KR20220011639A - Multipore Determination of the Distribution Fraction of a Polynucleotide Sequence in a Sample - Google Patents

Abstract

Disclosed herein is a nanopore sensor having one or more nanopores, for example, by correcting for errors inherent in identifying and correlating an electrical signal to the amount of a target analyte or reference analyte in a sample, thereby allowing a target analyte in a sample ( Methods and compositions for determining an improved estimate of the fraction of distribution (eg, a particular polynucleotide sequence) are disclosed.

Description

Multipore Determination of the Distribution Fraction of a Polynucleotide Sequence in a Sample

field of invention

A method is provided for determining the distribution fraction of a particular polynucleotide sequence in a sample using a system equipped with one or more solid state nanopores and a mathematical method for precise and accurate quantitation.

background of the invention

Characterization of a liquid sample by determining the relative abundance of components present in the liquid sample can provide useful information for many scientific fields and applications. For example, the relative abundance of point mutations in circulating cell-free DNA can be used to diagnose or monitor cancer progression in a patient. As another example, determining the fraction of a transgene sequence of a genetically modified organism (GMO) relative to a non-GMO reference sequence in genomic DNA, such as obtained from a seed swarm, is important for regulatory and economic reasons.

Although some methods exist for highly sensitive detection of an amount of a target analyte in a sample, these methods are generally expensive, time consuming, or have other limitations. For example, quantitative real-time PCR (qPCR) assays remain the standard method for determining the relative amount of a target nucleic acid sequence to a constant reference sequence in a test sample. However, the quantitative performance of qPCR is limited by the variability of amplification efficiency per sample and per amplicon. Factors affecting amplification efficiency include inhibitors and residual contaminants of the sample substrate and extraction reagents themselves. These factors vary by sample and preparation, and to the extent that they can affect the efficiency of amplification of one sequence versus another. Slight, variable differences in the amplification efficiency of target versus reference amplicons limit qPCR to resolving >1.5-fold quantitative differences. In addition, amplification reactions require special reagent sets and must be properly stored, can be time consuming and sensitive to reaction conditions.

The use of nanopore devices has emerged as a sensitive tool for single molecule identification, in which individual molecules are identified upon translocation through the nanopore under an applied voltage. Nanopore devices can be modified depending on the intended use, and are inexpensive and efficient enough for everyday use cases, whether in human health, agriculture, or anywhere else. However, the use of data from nanopores can introduce errors that can affect the determination of quantitative estimates of analytes in a sample, so the use of these data is unreliable.

Accordingly, there is a need for an improved method for determining the fraction of a target analyte relative to a reference analyte in a sample that is versatile, inexpensive, and easy to use.

Summary of the invention

According to some embodiments, there is provided a method of determining an improved estimate of the true relative abundance of a target analyte in a mixed unknown sample using a nanopore device, wherein a voltage is applied across the nanopore within the nanopore device to generate a detectable electronic signal and for each of a control sample comprising a known relative abundance of a target analyte to a reference analyte, and a mixed unknown sample comprising the target analyte and the reference analyte: separately via the nanopore via the nanopore inducing a potential of a charged analyte, wherein the relative abundance of the target analyte in the sample is determined; generating for each sample a plurality of event signals generated by translocation of the target analyte or the reference analyte through the nanopores; A detected relative abundance of first and second event signals for each sample by identifying an amount of a first event signal associated with the target analyte and an amount of a second event signal associated with the reference analyte from the plurality of event signals determining; and adjusting the detected relative abundances of the first and second event signals in the mixed unknown sample using the detected relative abundances of the first and second event signals in the control sample to obtain an error in the detected relative abundances. A method is provided comprising correcting for , thereby determining an improved estimate of the true relative abundance of the target analyte in the mixed unknown sample. In some embodiments, the sample is a liquid sample.

In some embodiments, the control sample is a target control sample comprising the target analyte but not the reference analyte. In some embodiments, the control sample is a reference control sample comprising the reference analyte but not the target analyte.

In some embodiments, a method of determining an improved estimate of the true relative abundance of a target analyte in a mixed unknown sample using the nanopore device includes applying a voltage to the nanopore device to include the target analyte, but see supra The method further comprises inducing a potential of the charged analyte through the nanopore sensor with respect to a target control sample that does not contain the analyte.

In some embodiments, the adjustment of the detected relative abundances of the first and second event signals in the unknown sample comprises the detected relative abundances of the first and second event signals in the target control sample and in the reference control sample. and correcting for the error of the detected relative abundance using In some embodiments, the error comprises a false positive or false negative detection error of the target analyte.

In some embodiments, a method of determining an improved estimate of the true relative abundance of a target analyte in a mixed unknown sample using the nanopore device comprises applying a voltage to the nanopore device to include the target analyte and the reference analyte inducing a potential of a charged analyte through a nanopore sensor with respect to a mixed control sample, wherein the relative abundance of the target analyte and the reference analyte is known.

In some embodiments, the adjustment of the detected relative abundance of the first and second event signals in the unknown sample comprises the first and second event signals in the target control sample, the reference control sample, and the mixed control sample. and correcting an error of the detected relative abundance using the detected relative abundance of .

In some embodiments, the error comprises a false positive target analyte detection error, a false negative target analyte detection error, a difference in capture rate constants of the target analyte and the reference analyte, or any combination thereof.

In some embodiments, the control sample is a mixed control sample comprising the target analyte and the reference analyte, and the relative abundance of the target analyte and the reference analyte is known. In some embodiments, the error comprises a difference in capture rate constants of the target analyte and the reference analyte.

In some embodiments, the mixed control sample is the target assay for the reference analyte that differs by no more than 1.2 fold, no more than 1.5 fold, no more than 2 fold, no more than 5 fold, or no more than 10 fold compared to the mixed unknown sample. Includes the relative abundance of water.

는

에 의해 결정되며, 여기서 상기 파라미터 ρ는 위양성 검출 오차, 위음성 검출 오차, 또는 둘 다를 보상할 수 있는 비율의 추정치이고, 상기 파라미터 α는 상기 표적 분석물 및 상기 참조 분석물의 포획률 상수 차이를 보상하기 위하여 사용될 수 있다. 일부 구현예에서, 상기 파라미터 α는 참조 분석물 포획률을 표적 분석물 포획률로 나눈 비율의 추정치이다.In some embodiments, the estimate of the true relative abundance is an estimate of the true ratio of the target analyte to the reference analyte in the mixed unknown sample. In some embodiments, an estimate of the true ratio (

Is

, wherein the parameter ρ is an estimate of a ratio capable of compensating for false-positive detection error, false-negative detection error, or both, and wherein the parameter α is to compensate for a difference in capture rate constants of the target analyte and the reference analyte can be used for In some embodiments, the parameter α is an estimate of the ratio of the reference analyte capture rate divided by the target analyte capture rate.

는

에 의해 결정되며, 여기서 상기 파라미터 ρ는 위양성 검출 오차, 위음성 검출 오차, 또는 둘 다를 보상할 수 있는 비율에 대한 추정치이고, 상기 파라미터 α는 상기 표적 분석물 및 상기 참조 분석물의 포획률 상수 차이를 보상하는 데 사용될 수 있다. 일부 구현예에서, 상기 파라미터 α는 상기 참조 분석물 포획률을 표적 분석물 포획률로 나눈 비율의 추정치이다.In some embodiments, the estimate of the true relative abundance is an estimate of the true fraction of the target analyte in the population of the reference analyte and the target analyte in the mixed unknown sample. In some embodiments, the estimate of the true fraction (

Is

, wherein the parameter ρ is an estimate for a rate capable of compensating for false-positive detection error, false-negative detection error, or both, and wherein the parameter α compensates for a difference in capture rate constants of the target analyte and the reference analyte. can be used to In some embodiments, the parameter α is an estimate of the ratio of the reference analyte capture rate divided by the target analyte capture rate.

이고,

이다. 일부 구현예에서, 상기 파라미터

는 상기 대조군 샘플을 사용한 경우 상기 표적 대조군 샘플 내에서 관찰되는 상기 제 1 이벤트 신호의 분율이고, 표적 대조군 샘플을 사용하지 않은 경우

1이다. 일부 구현예에서, 상기 파라미터

는 상기 참조 대조군 샘플을 사용한 경우 상기 참조 대조군 샘플 내에서 관찰되는 상기 제 1 이벤트 신호의 분율이고, 참조 대조군 샘플을 사용하지 않은 경우

0이다. 일부 구현예에서, 상기 파라미터

는 상기 혼합 대조군 샘플 내에서 관찰되는 상기 제 1 이벤트 신호의 분율이고 여기서 상기 대조군 샘플을 사용한 경우

는 상기 혼합 대조 샘플 내 참조 분석물(Y)에 대한 표적 분석물(X)의 공지된 비율이거나, 혼합 대조군 샘플을 사용하지 않은 경우

이다. 일부 구현예에서, 상기 파라미터

는 상기 혼합 미공지 샘플 내에서 관찰되는 상기 제 1 이벤트 신호의 분율이다.In some embodiments, the parameter

ego,

to be. In some embodiments, the parameter

is the fraction of the first event signal observed in the target control sample when the control sample was used, and when the target control sample was not used

1 is In some embodiments, the parameter

is the fraction of the first event signal observed in the reference control sample when the reference control sample was used, and when the reference control sample was not used

0. In some embodiments, the parameter

is the fraction of the first event signal observed in the mixed control sample where using the control sample

is the known ratio of target analyte (X) to reference analyte (Y) in the mixed control sample, or when a mixed control sample is not used

to be. In some embodiments, the parameter

is the fraction of the first event signal observed in the mixed unknown sample.

In some embodiments, the unknown or control sample is prepared by nucleic acid amplification. In some embodiments, the unknown or control sample is not prepared by nucleic acid amplification. In some embodiments, the sample is purified to consist substantially of reference and target molecules. In some embodiments, the sample is not purified.

In some embodiments, the amount or concentration of the reference analyte in the mixed unknown sample is known. In some embodiments, the method of determining an improved estimate of the true relative abundance of a target analyte in a mixed unknown sample using the nanopore device comprises: the true relative of the target analyte to the reference analyte in the mixed unknown sample. determining an estimate of the absolute amount or concentration of the target analyte in the mixed unknown sample using the estimate of abundance and the known amount or concentration of the reference analyte in the mixed unknown sample. In a related embodiment, the absolute amount or concentration of the target analyte may be determined using information derived from a plurality of nanopores of one or more nanopore devices.

In some embodiments, an amount of a first event signal associated with the target analyte and an amount of a second event signal associated with the reference analyte are identified according to a defined limit. In some embodiments, the method for determining an improved estimate of the true relative abundance of a target analyte in a mixed unknown sample using the nanopore device comprises optimizing the limit to perform a Q-test, a support vector machine, or an expectation maximization algorithm. The method further comprises increasing the detection accuracy of the reference analyte and/or the target analyte using the method. In some embodiments, the support vector machine is trained using electronic signals generated from a control sample comprising known amounts of a target analyte and a reference analyte.

In some embodiments, the defined limits are: event duration, maximum δG, median δG, average δG, standard deviation of event signal, average or median of noise power of events less than 50 Hz, unique pattern within the event signal, event is a function of a characteristic of one or more event signals selected from the group consisting of area, or any combination thereof.

In some embodiments, adjusting the detected relative abundance of the first and second event signals in the mixed unknown sample to correct for the error in the detected relative abundance is a Q-test, a support vector machine, or an expectation maximization This is done using an algorithm.

In some embodiments, the target analyte and the reference analyte each comprise a polynucleotide. In some embodiments, the target analyte polynucleotide and the reference analyte polynucleotide have different lengths. In some embodiments, the lengths differ by at least 10 nucleotides, at least 20 nucleotides, at least 50 nucleotides, at least 100 nucleotides, at least 150 nucleotides, or at least 200 nucleotides.

In some embodiments, the method of determining an improved estimate of the true relative abundance of a target analyte in a mixed unknown sample using the nanopore device comprises: combining the control or unknown sample with a first probe bound to a first payload; contacting, wherein the first probe is configured to specifically bind to the first analyte. In some embodiments, the method of determining an improved estimate of the true relative abundance of a target analyte in a mixed unknown sample using the nanopore device comprises: combining the control or unknown sample with a second probe coupled to a second payload; contacting, wherein the second probe is configured to specifically bind to the second analyte.

In some embodiments, the target analyte is associated with a genetically modified organism. In some embodiments, the target analyte comprises a marker associated with the presence or absence of cancer in the patient.

Also provided herein is a method for determining the relative amount of a target analyte in a sample, a first control sample including a reference analyte but not including a target analyte, and a second control sample including a target analyte but not including a reference analyte a control sample, a third control sample comprising a known relative abundance of the target analyte and the reference analyte, and an experimental sample comprising an unknown relative abundance of the target analyte and the reference analyte: each in a nanopore system running in; detecting for each sample an amount of a first event signal associated with a reference analyte and an amount of a second event signal associated with a target analyte; and the relative abundance of the amounts of the first and second event signals from the experimental sample of the first and second event signals from each of the first control sample, the second control sample, and the third control sample. A method is provided comprising determining an estimate of the true relative abundance of the reference analyte and the target analyte in the experimental sample as compared to the relative abundance of an amount.

In some embodiments, the event signal comprises an electrical signal induced by the potential of the reference analyte through the nanopore.

In some embodiments, the target analyte and the reference analyte each comprise a polynucleotide. In some embodiments, the target analyte and the reference analyte are distinguished by a length.

In some embodiments, the target analyte and the reference analyte are each bound to a sequence-specific probe comprising a payload that facilitates discrimination between the reference analyte and the target analyte in the nanopore device.

In some embodiments, the relative abundance is the quantity of the target analyte compared to the total population of the target analyte and the reference analyte in the sample.

Also provided herein is a method for determining the relative abundance of a target analyte in an unknown sample, the method comprising: providing an unknown sample comprising a plurality of reference analytes and a plurality of target analytes; loading the unknown sample into a first chamber of a nanopore device comprising nanopores disposed between the first chamber and the second chamber; applying a voltage to the nanopores so that the reference analyte and the target analyte pass through the nanopores from the first chamber to the second chamber; detecting a number of first electrical signals related to the potential of the reference analyte through the nanopores; detecting a number of second electrical signals related to the potential of the target analyte through the nanopores; and analyzing the target in the unknown sample for the relative abundance of the number of detected first electrical signals and the number of detected second electrical signals using a reference value that accounts for at least one error associated with the relative abundance of electrical signals. A method is provided comprising converting to an estimate of the true relative abundance of water.

In some embodiments, the reference value is determined from a fraction of the distribution of the first electrical signal determined from a mixed control sample comprising a known amount of a target analyte and a reference analyte. In some embodiments, the reference value is determined from a fraction of the distribution of the first electrical signal determined from a mixed control sample comprising a known amount of a target analyte and a reference analyte. In some embodiments, the reference value is determined from a fraction of the distribution of the first electrical signal determined from a mixed control sample comprising a known amount of a target analyte and a reference analyte.

In some embodiments, the mixed control sample, the target control sample, or the reference control sample are under substantially the same conditions as the conditions in the nanopore device during the detection of the first and second electrical signals from the unknown sample. It is carried out in the nanopore device under

In some embodiments, the nanopore device comprises a membrane separating an internal space of the device into a first chamber and a second chamber, wherein the membrane comprises the nanopore, wherein the first chamber and the second chamber are In fluid communication through the nanopores, the device includes in each chamber an electrode for applying a voltage to the nanopores. In some embodiments, the electrode is configured to detect a current through the nanopore. In some embodiments, the electrode is connected to a power source.

In some embodiments, the methods provided herein provide false positive or false negative detection error, or the accuracy of an estimate of the fraction of distribution of a target analyte in a mixed unknown sample by accounting for a difference in capture rate constants of the target analyte and the reference analyte. to improve In some embodiments, a series of controls are run to improve the accuracy of the estimate of the fraction of distribution, a control sample of reference analyte only to account for false-positive target analyte detection errors, and to account for false-negative target analyte detection errors. a control sample of only the target analyte for , and one or more mixed control samples to account for differences in capture rate constants between the target analyte and the reference analyte.

In some embodiments, the capture rate between the target analyte and the reference analyte in the mixed unknown sample is relatively constant, such that the mixed control need not be used to improve the estimate of relative abundance. In some embodiments, the relative capture rate between the target analyte and the reference analyte in a mixed sample is determined such that a correction term is applied to data from a mixed unknown sample to compensate for this difference to provide an estimate of the fraction of distribution without running a mixed control sample. known to be improved. In some embodiments, data from a mixed control sample run under substantially identical nanopore conditions using the same target analyte and reference analyte species as in the mixed unknown sample is actually running the mixed control sample as part of a method. It is used to improve the estimate of the fraction of distribution without

In some embodiments, a threshold is determined such that false positive values from the mixed unknown sample are negligible and that a reference-only control need not be used to improve the estimate of relative abundance. In some embodiments, false positive values from a mixed sample are known so that a correction term can be applied to data from a mixed unknown sample to compensate for false positive errors to improve the estimate of the fraction of distribution without running a reference-only control sample. In some embodiments, data from a reference analyte-only control sample run under substantially identical nanopore conditions using the same reference analyte species as in the mixed unknown sample is, as part of a method, actually reference-only control. It is used to improve the estimate of the fraction of distribution without running.

In some embodiments, a threshold is determined such that false negative values from the mixed unknown sample are negligible and that a target-only control need not be used to improve the estimate of relative abundance. In some embodiments, false negative values from a mixed sample are known so that a correction term can be applied to data from the mixed unknown sample to compensate for false negative errors to improve the estimate of the fraction of distribution without running a target-only control sample. In some embodiments, data from a target analyte-only control sample run under substantially identical nanopore conditions using the same target analyte species as in the mixed unknown sample is, as part of a method, actually target-only control. It is used to improve the estimate of the fraction of distribution without running.

In some embodiments, disclosed herein is a method of determining an estimate of the relative abundance of a target analyte to a reference analyte in a mixed sample, comprising applying a voltage to a nanopore device to include a known relative abundance of a target analyte to a reference analyte inducing a potential of a charged analyte through a nanopore sensor for each of a mixed control sample, and a mixed unknown sample comprising the target analyte and the reference analyte, wherein the reference analyte is an unknown relative abundance ratio of the target analyte to the target analyte; detecting for each sample an amount of a first event signal associated with the reference analyte and an amount of a second event signal associated with a target analyte; and the detected relative abundance of the first and second event signals in the mixed control sample and the true relative abundance of the target analyte to the reference analyte in the mixed control sample. A method is provided comprising determining an estimate of the true relative abundance of the target analyte to the reference analyte in the mixed unknown sample by adjusting the detected relative abundances of first and second event signals.

In some embodiments, disclosed herein is a method of determining an estimate of the relative abundance of a target analyte to a reference analyte in a mixed sample, wherein a voltage is applied to a nanopore device to include the target analyte but not the reference analyte a target control sample, a reference control sample comprising a reference analyte but no target analyte, and the target analyte and the reference analyte, wherein the relative abundance of the target analyte to the reference analyte is unknown. A mixed unknown sample comprising: inducing an electric potential through the nanopore sensor individually for each; detecting for each sample an amount of a first event signal associated with the reference analyte and an amount of a second event signal associated with a target analyte; and adjusting the detected relative abundances of the first and second event signals in the mixed unknown sample using the detected relative abundances of the first and second event signals in the target control sample and the reference control sample. A method is provided comprising determining an estimate of the true relative abundance of said target analyte to said reference analyte in a mixed unknown sample. In some embodiments, the target control sample provides a correction for false-negative detection of a target analyte from the mixed unknown sample. In some embodiments, the reference control sample provides a correction for false-positive detection of a target analyte in the mixed unknown sample.

In some embodiments, disclosed herein is a method of determining an estimate of the relative abundance of a target analyte to a reference analyte in a mixed sample, comprising applying a voltage to a nanopore device to include a known relative abundance of a target analyte to a reference analyte a mixed control sample comprising a target analyte but no reference analyte, a reference control sample comprising a reference analyte but no target analyte, and the target for the reference analyte A mixed unknown sample comprising the target analyte and the reference analyte, wherein the relative abundance of the analyte is unknown: inducing a potential through the nanopore sensor individually for each; detecting for each sample an amount of a first event signal associated with the reference analyte and an amount of a second event signal associated with a target analyte; and the detected relative abundances of the first and second event signals in the target control sample and the reference control sample, and the detected relative abundances of the first and second event signals in the mixed control sample and in the mixed control sample. to the reference analyte in the mixed unknown sample by adjusting the detected relative abundances of the first and second event signals from the mixed unknown sample using the true relative abundance of the target analyte to the reference analyte. A method is provided comprising determining an estimate of the true relative abundance of the target analyte for

In some embodiments, the method of determining an estimate of the relative abundance of a target analyte to a reference analyte in the mixed sample comprises applying a voltage to a nanopore device to include the target analyte, but not the reference analyte. Inducing a potential of the charged analyte through the nanopore sensor relative to the control sample.

In some embodiments, the method of determining an estimate of the relative abundance of a target analyte to a reference analyte in the mixed sample comprises applying a voltage to a nanopore device to include the reference analyte, but not the target analyte. Inducing a potential of the charged analyte through the nanopore sensor relative to the control sample. In some embodiments, determining said estimate of the true relative abundance of said target analyte to said reference analyte in said mixed unknown sample comprises said target control sample, said reference control sample, and said mixed control in said mixed control sample. and adjusting the detected relative abundance of the first and second event signals in the mixed unknown sample using the true relative abundance of the target analyte to the reference analyte in the sample.

In some embodiments, the mixed control sample is the target assay for the reference analyte that differs by no more than 1.2 fold, no more than 1.5 fold, no more than 2 fold, no more than 5 fold, or no more than 10 fold compared to the mixed unknown sample. Includes the relative abundance of water.

)는

에 의해 결정되며, 여기서 상기 파라미터 ρ는 위양성 검출 오차, 위음성 검출 오차, 또는 둘 다에 대한 보상을할 수 있는 비율에 대한 추정치이고, 상기 파라미터 α는 상기 표적 분석물 및 상기 참조 분석물의 포획률 상수 차이에 대한 보상에 사용될 수 있다. 일부 구현예에서, 상기 파라미터 α는 참조 분석물 포획률을 표적 분석물 포획률로 나눈 비의 추정치이다.In some embodiments, the relative abundance comprises a ratio of target analyte:reference analyte. In some embodiments, an estimate of the true ratio of a target analyte to the reference analyte in the mixed unknown sample (

)Is

, where the parameter ρ is an estimate of the rate that can compensate for false-positive detection error, false-negative detection error, or both, and the parameter α is the capture rate constant of the target analyte and the reference analyte. It can be used to compensate for differences. In some embodiments, the parameter α is an estimate of the ratio of the reference analyte capture rate divided by the target analyte capture rate.

)는

에 의해 결정되며, 여기서 상기 파라미터 ρ는 위양성 검출 오차, 위음성 검출 오차, 또는 둘 다에 대한 보상을 할 수 있는 비율의 추정치이고, 상기 파라미터 α는 상기 표적 분석물 및 상기 참조 분석물의 포획률 상수 차이에 대한 보상에 사용될 수 있다. 일부 구현예에서, 상기 파라미터 α는 참조 분석물 포획률을 표적 분석물 포획률로 나눈 비의 추정치이다.In some embodiments, the relative abundance is the amount of the target analyte in the population of the target analyte and the reference analyte. In some embodiments, an estimate of the true amount of the target analyte in the population of the reference analyte and the target analyte in the mixed unknown sample (

)Is

, wherein the parameter ρ is an estimate of a ratio capable of compensating for false-positive detection error, false-negative detection error, or both, and wherein the parameter α is the difference in capture rate constants of the target analyte and the reference analyte. can be used for compensation. In some embodiments, the parameter α is an estimate of the ratio of the reference analyte capture rate divided by the target analyte capture rate.

In some embodiments, disclosed herein is a control sample comprising known relative abundances of a target analyte and a reference analyte; and running the control sample and an unknown sample comprising the reference analyte and the target analyte in a nanopore device to determine the relative abundance of the reference analyte and the target analyte in the unknown sample. A kit is provided with instructions for

In some embodiments, a first control sample comprising a target analyte herein, wherein the first control sample does not comprise a reference analyte; a second control sample comprising the reference analyte, wherein the second control sample does not include the target analyte; a third control sample comprising known relative abundances of said target analyte and said reference analyte; and running the first control sample, the second control sample, the third control sample, and the unknown sample comprising the reference analyte and the target analyte separately in a nanopore device, so that the in the unknown sample is Kits are provided comprising a reference analyte and instructions for use in determining the relative abundance of the target analyte.

In some embodiments, provided herein is a computer-implemented method of determining an estimate of the true fraction of a target analyte in a sample, comprising: obtaining data from a nanopore sensor from at least one reference analyte control or target analyte control; wherein the data comprises a plurality of event signals from a target analyte or a reference analyte translocating through the nanopore; identifying one or more characteristics of the event signal to differentiate a signal associated with a target analyte and a signal associated with a reference analyte; training a support vector machine to identify an optimized limit for differentiating the first event from the second event and generating an estimate of the true relative abundance of the reference analyte and the target analyte in a sample, wherein the training comprises the use of a control selected from the group consisting of a reference control sample, a target control sample, and a mixed control sample, wherein the training comprises validating using a known mixed sample; and determining a distribution fraction of a target analyte in the sample from events recorded in the nanopore device from the mixed sample using the trained support vector machine.

In some embodiments, disclosed herein is a computer-implemented method of determining an estimate of a true fraction of a target analyte in a sample, comprising: obtaining a data set from a nanopore device, the data comprising at least one control sample and at least one including an event signal from an unknown sample of identifying a set of features and using them to create a limit for differentiating a first event signal associated with the target analyte from a second event signal associated with the reference analyte; and estimating the true value of the fraction of distribution in the unknown sample using a trained support vector machine.

In some embodiments, disclosed herein is a computer-implemented method of determining an estimate of a fraction of a distribution of a target analyte in a sample, comprising: obtaining a data set from one or more nanopore devices, each nanopore device comprising one or more nanopores. including; processing a set of inputs, derived from the data set, into a multipore analytical model, the set of inputs comprising calls from each of the set of nanopores; and generating, based on the response output of the multipore analysis model, an estimate of the distribution fraction of the target analyte in the sample and the reliability of the distribution fraction.

가 q 값에서 그래픽으로 어떻게 나타나는지 보여준다. q = 5 pA*ms 한계(수직 점선)는 위양성 0.05(즉, Q _ref = 0.05) 및 위음성 0.1(즉, Q _targ = 0.9)에 해당한다.
도 5A는 표적 유전자의 상대적 존재비의 추정치(GMO(%) (

) 대 표적 유전자의 참 상대적 존재비(GMO(%))의 결정 결과를 보여준다. 비교를 위하여 영오차선(기울기 = 1)의 위 및 아래에 10% 오차 한계가 도시된다.
도 5B는 2 개의 분리된 대조군 및 6 개의 공지된 혼합물을 사용하여 샘플 내 유전자 변형 생물체의 참 상대적 존재비의 추정치의 결정의 결과를 보여준다. 예측된 표적 존재비 백분율 값은 참 표적 존재비 백분율에 대해 도시된다. 비교를 위하여 영오차선(기울기 = 1)의 위 및 아래에 10% 오차 한계가 도시된다.
도 6은 이벤트 면적에 따라 참조 분석물로부터 표적 분석물을 구별하기 위한 한계의 범위에 대한 표적 분석물 존재비의 추정치(GMO(%))의 결과를 보여준다.
도 7은 표적 및 참조 분석물로부터의 이벤트 신호를 구별하기 위한 최적 파라미터가 있는 훈련된 서포트 벡터 머신으로부터의 시험 데이터 세트에 대한 정확도를 예측을 보여준다.
도 8은 동일한 포어 상에서 분리된 대조군으로서 연속적으로 실행된 2 가지 분자 유형(프로브/페이로드에 결합된 94bp 표적 dsDNA 및 프로브/페이로드에 결합된 74bp 참조 dsDNA)에 대한 이벤트를 보여준다.
도 9A는 중첩된 100% 표적 분석물 대조군 샘플(닫힌 원) 및 100% 참조 분석물 대조군 샘플(열린 사각형)에 대하여 평균 δG 대 지속시간의 대표적인 이벤트 플롯을 나타낸다. 표적 분석물은 3-분지 PEG에 연결된 G12D-결합 프로브가 있는 89bp DNA(이하, G12D-3bPEG)이다. 참조 분석물은 8-암 PEG에 연결된 야생형(c.35G)-결합 프로브가 있는 89bp DNA(이하, WT-8armPEG)이다. 나노포어를 통과하는 표적 분석물로부터 이벤트 신호를 식별하기 위한 한계(

msec,

nS 및

nS)가 표적 태그 상자(점선)를 생성한다.
도 9B는 도 9A로부터, 플롯에 중첩된 표적 분석물과 참조 분석물을 포함하는 미공지 샘플 A(삼각형) 및 샘플 B(별)로부터의 데이터가 있는 플롯을 보여준다.
도 10은 중첩된 100% 표적 분석물 대조 샘플(닫힌 원)과 100% 참조 분석물 대조 샘플(열린 사각형)에 대하여 평균 δG 대 지속시간의 대표 이벤트 플롯을 보여준다. 참조 분석로부터 표적 분석물을 구별하기 위한 서포트 벡터 머신-식별된 결정 경계(즉, 한계)도 표시된다.
도 11은 최대 δG 대 지속시간의 모든 이벤트 산점도에 플롯팅된 50% 표적/50% 참조 혼합 샘플로부터의 이벤트를 보여준다. 표적 도메인 상자는 프로브-결합 돌연변이 표적과 관련된 이벤트를 포함한다.
도 12는 표적(돌연변이) 및 참조(야생형) 집단의 식별을 위해 도 11에 도시된 50% 표적 / 50% 참조 혼합 샘플로부터의 데이터에 3-가우스 혼합 모델을 사용하여 가우스 혼합에 대한 기대 최대화 알고리즘(EMGM)을 적용한 결과를 보여준다.
도 13은 위양성 분율을 결정하기 위하여 참조 분석물만의 대조군 샘플로부터의 데이터에 3-가우스 혼합 모델을 사용하여 EMGM을 적용한 결과를 보여준다.
도 14는 미공지 샘플 내 돌연변이(표적) 분자의 상대적 존재비를 확인하기 위하여 혼합 미공지 샘플로부터의 데이터에 3-가우스 혼합 모델을 사용하여 EMGM을 적용한 결과를 보여준다.
도 15A는 샘플 내 표적 분석물의 분포분율을 결정하기 위하여 다수의 포어로부터 컨센서스 호출을 이용하는 방법의 흐름도를 도시한다.
도 15B는 분포분율을 결정하기 위하여 4 개의 포어로부터 컨센서스 호출을 이용하는 구현예를 도시한다.
도 15C는 임의의 수의 나노포어로부터, 도 15B의 시스템의 구현예에 의해 구현되는 방법의 흐름도를 도시한다.
도 15D는 단일 나노포어 및 다수의 나노포어로부터 유래된 정보를 이용하여 표적 분자의 농도를 결정하기 위한 방법의 출력을 도시한다.
도 16A는 하나 이상의 구현예에 따라, 나노포어로부터의 데이터를 사전 필터링하기 위한 방법의 흐름도를 도시한다.
도 16B는 도 16A에 도시된 방법의 출력을 도시하며, 나노포어로부터의 이벤트 데이터는 전기 신호 대 체류시간의 진폭에 의해 플롯팅된다.
도 16C는 도 16A 및 16B에 도시된 방법의 출력을 도시하며, PCA 과정의 성분들(PC1 및 PC2)을 사용하여 각 샘플 집단에 대한 가우스 분포 계수 대 PC1를 생성한다.
도 16D는 도 16C에 도시된 PCA 과정의 출력을 이용하여 분리 점수를 생성하는 데 사용되는 데이터를 도시한다.
도 16E는 샘플의 보정 비율을 확인하기 위하여, 최대 전류 진폭을 초 단위의 체류시간의 로그 함수에 대해 플롯팅한 것을 도시한다. Brief description of the drawing
The foregoing and other objects, features and advantages will become apparent from the following description of specific embodiments of the present invention, as illustrated in the accompanying drawings in which like reference numerals indicate like parts throughout different drawings. The drawings are not necessarily to scale, emphasis instead being placed on illustrating the principles of various embodiments of the invention.
1A shows a typical electronic signal of a single-molecule event caused by a dsDNA passing through a nanopore, showing the characteristic duration of the translocation and the decrease in current during the translocation.
1B shows an all-event scatterplot of maximum δG versus duration for a 5.6 kb dsDNA recorded in 22 nm diameter nanopores.
2A shows typical events that occur when 727 bp DNA passes through a 25 nm diameter solid-state nanopore at 100 mV in 1 M LiCl. The discharge area is shaded.
Figure 2B shows an increase in event duration with increased dsDNA length, while event depth was preserved.
2C shows a plot of the distribution of _{log 10} of all event areas recorded for dsDNA at each length shown.
3A depicts an example of a boundary created between events from a first type analyte (square) and a second type analyte (circle).
3B shows an example of the result of transformation of an input feature into a higher dimensional space to increase the accuracy of the linear limit between events from a first type analyte (square) and a second type analyte (circle). did it
4A shows the probability histograms for all events from the reference analyte sample, the target analyte sample, and the mixed sample as a function of event area.
4B depicts a graph of the percentage of events below the area limit from reference analyte only (Q _ref ), target analyte only (Q _targ ), and a mixed sample of target analyte and reference analyte (Q _{mix ).}
Figure 4C shows the aliquot parameters.

shows how is represented graphically at the q value. The q = 5 pA*ms limit (vertical dashed line) corresponds to a false positive of 0.05 (ie, Q _ref = 0.05) and a false negative of 0.1 (ie, Q _targ = 0.9).
5A is an estimate of the relative abundance of target genes (GMO (%) (

) versus the true relative abundance (GMO (%)) of the target gene. For comparison, the 10% margin of error is shown above and below the zero lane (slope = 1).
5B shows the results of determination of an estimate of the true relative abundance of genetically modified organisms in a sample using two separate controls and six known mixtures. Predicted target abundance percentage values are plotted versus true target abundance percentages. For comparison, the 10% margin of error is shown above and below the zero lane (slope = 1).
6 shows the results of estimates of target analyte abundance (GMO (%)) for a range of limits for differentiating target analytes from reference analytes according to event area.
7 shows prediction of accuracy on a test data set from a trained support vector machine with optimal parameters for discriminating event signals from target and reference analytes.
8 shows events for two molecular types (94 bp target dsDNA bound to probe/payload and 74 bp reference dsDNA bound to probe/payload) run serially as controls isolated on the same pore.
9A is Representative event plots of mean δG versus duration are shown for overlaid 100% target analyte control samples (closed circles) and 100% reference analyte control samples (open squares). The target analyte is 89 bp DNA (hereinafter G12D-3bPEG) with a G12D-binding probe linked to a 3-branched PEG. The reference analyte is 89 bp DNA (hereafter WT-8armPEG) with a wild-type (c.35G)-binding probe linked to an 8-arm PEG. Limitations for identifying event signals from target analytes that pass through the nanopores (

msec,

nS and

nS) creates a target tag box (dotted line).
FIG. 9B shows a plot from FIG. 9A with data from unknown sample A (triangles) and sample B (stars) comprising target and reference analytes superimposed on the plots.
FIG. 10 shows representative event plots of mean δG versus duration for overlapped 100% target analyte control samples (closed circles) and 100% reference analyte control samples (open squares). Support vector machine-identified decision boundaries (ie, limits) for differentiating target analytes from reference analyses are also indicated.
11 shows events from a 50% target/50% reference mixed sample plotted on a scatter plot of all events of maximum δG versus duration. The target domain box contains events related to the probe-binding mutation target.
Figure 12 is an expectation maximization algorithm for Gaussian mixing using a 3-Gaussian mixture model on data from the 50% target / 50% reference mixed sample shown in Figure 11 for identification of target (mutant) and reference (wild-type) populations. (EMGM) is applied.
13 shows the results of applying EMGM using a 3-Gaussian mixture model to data from a control sample of reference analyte only to determine the false positive fraction.
14 shows the results of applying EMGM to data from a mixed unknown sample using a 3-Gaussian mixture model to determine the relative abundance of mutant (target) molecules in the unknown sample.
15A depicts a flow diagram of a method of using a consensus call from multiple pores to determine the distribution fraction of a target analyte in a sample.
15B shows an implementation that uses a consensus call from four pores to determine the distribution fraction.
15C shows a flow diagram of a method implemented by an implementation of the system of FIG. 15B, from any number of nanopores.
15D depicts the output of a method for determining the concentration of a target molecule using information derived from single nanopores and multiple nanopores.
16A depicts a flow diagram of a method for pre-filtering data from nanopores, in accordance with one or more implementations.
Figure 16B shows the output of the method shown in Figure 16A, wherein event data from the nanopores is plotted by the amplitude of the electrical signal versus residence time.
16C shows the output of the method shown in FIGS. 16A and 16B , using the components of the PCA process (PC1 and PC2) to generate Gaussian distribution coefficients versus PC1 for each sample population.
FIG. 16D shows the data used to generate a separation score using the output of the PCA process shown in FIG. 16C.
Figure 16E shows the plot of the maximum current amplitude as a log function of the dwell time in seconds to confirm the calibration rate of the sample.

details

The details of various embodiments of the invention are set forth in the description that follows. Other features, objects, and advantages of the present invention will be apparent from the following description, and drawings, and from the claims.

Justice

Throughout this application, the text describes various embodiments of the present nutrients, compositions, and methods. The various implementations described are for the purpose of providing various examples and should not be construed as alternative descriptions. Conversely, it should be noted that the descriptions of various embodiments provided herein may be in overlapping scope. The embodiments discussed herein are illustrative only and are not intended to limit the scope of the invention.

Throughout this specification, various documents, patents, and published patent specifications are referenced by identification citation. The disclosures of these documents, patents, and published patent specifications are incorporated herein by reference in order to describe in more detail the state-of-the-art in the technical field to which the present invention pertains.

As used herein and in the claims, the singular forms “a,” “an,” and “the” include the plural unless the context clearly dictates otherwise. For example, the term “an electrode” includes a plurality of electrodes, including mixtures thereof.

As used herein, the term “comprising” is intended to mean that apparatus and methods include the recited components or steps, but do not exclude others. "Consisting essentially of," as used when defining apparatus and methods, means excluding other elements or steps of essential importance to the combination. "Consisting of" means excluding other elements or steps. Embodiments defined by each of these transition terms are included within the scope of the present invention.

All numerical designations, including ranges, such as distance, size, temperature, time, voltage, and concentration, are approximations intended to include general experimental deviations in the measurement of parameters, such deviations being included within the scope of the described embodiments. it is intended to be It is to be understood that all numerical designations are preceded by the term “about,” although not always explicitly stated. It is also to be understood that, although not explicitly recited, the components described herein are exemplary only and equivalents thereof are known in the art.

As used herein, the term “analyte” refers to any molecule, compound, complex, or other entity capable of detecting its presence using a nanopore sensor to facilitate determination of the relative abundance of an analyte in a pore. refers to When referring to a target or reference analyte, the terms target or reference molecule may be used interchangeably.

As used herein, the term “target analyte” refers to a molecule or complex of interest in a sample. In some embodiments, the target analyte comprises a portion of a polynucleotide having a sequence of a nucleic acid of interest. The target analyte can be specifically targeted for binding by a probe to facilitate detection of the target analyte in the nanopore sensor, as described herein.

As used herein, the term "reference analyte" refers to a molecule or complex of interest in a sample, whose abundance is used as a relative measure of quantification for a target analyte. In some embodiments, the reference analyte comprises a portion of a polynucleotide having a sequence of a nucleic acid of interest. The reference analyte can be specifically targeted for binding by a probe to facilitate detection of the target analyte in the nanopore sensor, as described herein.

As used herein, the term “specific binding” or “specifically binding” refers to the targeted binding of a probe to a target analyte or reference analyte.

As used herein, the term “probe” refers to a molecule that specifically binds to a target analyte or fragment thereof. In some embodiments, the probe comprises a payload molecule configured to affect an electronic signal generated upon translocation of the probe-payload molecule or complex comprising a target or reference analyte bound to the complex. In some embodiments, the probe comprises a payload molecule binding moiety modified to bind to a payload molecule.

As used herein, the term “payload molecule” refers to a molecule having a physical dimension that facilitates the generation of a distinctive electrical signal when captured within a nanopore within a relevant range of dimensions. The payload molecule may be bound to a target analyte or reference analyte to facilitate detection of the target analyte or reference analyte in the nanopore device. In some embodiments, the payload molecule may be charged to act as a driving molecule. In some embodiments, the payload molecule comprises a probe binding moiety capable of specifically binding to a probe molecule, and the probe specifically binds to a target analyte or a reference analyte.

As used herein, the term “nanopore” (or simply “pore”) refers to a single nano-scale opening in a membrane that separates two volumes. The pores may be, for example, protein channels embedded within a lipid bilayer membrane, or drilled through or through a thin solid-state substrate, such as a layer made of silicon nitride or silicon dioxide or graphene or a combination thereof or other materials. It can be manipulated using etching or voltage-pulsing methods. geometrically, the pores have dimensions that are greater than or equal to 0.1 nm in diameter and less than or equal to 1 micron in diameter; The length of the pores is determined by the thickness of the film, which can be up to a nanometer thick, or up to 1 micron thick or more thick. For membranes thicker than several hundred nanometers, the nanopores may be referred to as “nanochannels”.

As used herein, the term “nanopore device” or “nanopore device” refers to a device that combines (in parallel or series) one or more nanopores with circuitry for sensing single molecule events. Each nanopore within a nanopore device, including a chamber and electrodes used to facilitate sensing with the nanopore, is referred to herein as a nanopore sensor. Specifically, the nanopore device applies a specific voltage to the pore or pores while measuring the ion current through the pore(s) using a sensitive voltage-clamp amplifier. When a single charged molecule, such as double-stranded DNA (dsDNA), is captured and driven through a pore by electrophoresis, the measured current changes, resulting in a capture event (i.e., the potential of the molecule through the nanopore, or within the nanopore). entrapment of molecules), the amount of change (current amplitude) and the duration of the event are used to characterize the entrapped molecules within the nanopore. After recording many events during the experiment, the distribution of events is analyzed to characterize the corresponding molecules according to the amount of change (ie the current signal). In this way, nanopores provide a simple, label-free, fully electrical single-molecule method for the detection of biomolecules.

As used herein, the term “electrical signal” includes a series of data of current, impedance/resistance, or voltage collected over time depending on the configuration of an electronic circuit. In general, current is measured in a “voltage clamp” configuration; Voltage is measured in a "current clamp" configuration, and resistance can be derived from either of the above using Ohm's Law V = IR. Impedance can also be generated by measuring from current or voltage data collected from the nanopore device. Although the types of electrical signals referred to herein include current signals and current impedance signals, a variety of other electrical signals may be used to detect particles within the nanopores.

As used herein, the term “event” refers to a change in the potential through a nanopore of a detectable molecule or molecular complex and a related measurement through an electrical signal, e.g., a current through the nanopore over time. . This may be defined as current, change in current from a reference open channel, duration, and/or other characteristics of detection of molecules within the nanopore. Multiple events with similar characteristics represent a population of molecules or complexes with the same or similar characteristics (eg, size, charge).

As used herein, "area" of an event is the duration of the event (i.e., the duration at which the current varies from the open channel current signal) multiplied by the average change in current from the open channel over the duration of the event (i.e., pA*ms).

As used herein, the term “relative abundance” refers to the amount of an item relative to the total number of related items in a group. For example, with respect to a target analyte in a sample, the relative abundance of the target analyte refers to the amount of the target analyte present in the sample relative to a reference analyte. This can be expressed as a fraction of distribution, eg, the percentage of target analyte in the sample compared to the total population of target analyte and reference analyte. The relative abundance can also be expressed as, for example, the ratio of target analyte:reference analyte. With respect to electronic signals, the relative abundance of a group of electronic signals may refer to an amount of a first electronic signal associated with a target analyte compared to an amount of a second electronic signal associated with a reference analyte. To distinguish between the true relative abundance of a target analyte in a sample (i.e., previously measured or prepared to have a known relative abundance) and a relative abundance determined according to the methods provided herein, the true relative abundance is referred to as "true relative abundance". , the relative abundance determined according to the methods provided herein is also referred to as an “estimate of the true relative distribution”.

As used herein, the term “control sample” refers to a sample comprising a co-located relative abundance of a target analyte to a reference analyte. Control samples, such as reference control samples, target control samples, and mixed control samples, are used herein to increase the accuracy of estimates of the fraction of distribution in the unknown sample. In some embodiments, a control sample comprises a target analyte, a reference analyte, or both.

As used herein, the term “unknown sample” or “unknown mixed sample” or “mixed unknown sample” refers to a sample comprising the relative abundance of an unknown reference analyte. When the relative abundance of a reference analyte is to be determined by the methods provided herein, it is considered unknown, even if some value of that estimate is already known. For some unknown samples, the amount or concentration of the reference analyte in the sample is known.

As used herein, the term "known sample" refers to a sample comprising a known relative abundance of a target analyte to a reference analyte, which is trained on features of that model, such as a fractional distribution estimation model or limits; Used to verify or provide an estimate of accuracy.

Introduction/Overview

In some embodiments, the invention provided herein provides methods for determining an estimate (eg, quantity or ratio) of the true relative abundance of a target analyte to a reference analyte present in a sample. This method utilizes a nanopore single molecule counting device (ie, a nanopore device) to detect and discriminate between target and reference analytes in a sample.

The use of raw electronic event signals associated with a target analyte and a reference analyte to determine an estimate of the relative abundance of a target analyte in a sample includes false positive detection error, false negative detection error, and between target analyte and reference analyte in a mixed sample. It can be inaccurate for a number of reasons, including errors related to differences in capture rate constants. Provided herein, in accordance with some embodiments, are methods for improving the accuracy of estimating the true distribution fraction of a reference and target analyte in a sample. In some embodiments, such methods involve the use of a control sample specifically designed for correction for one or more errors associated with detection of an electronic signal within the mixed sample. When the mixed sample includes a known amount or concentration of a reference gemstone, an improved estimate of the relative abundance can be used to provide an improved estimate of the true amount or concentration of the target analyte in the sample.

In some embodiments, the methods provided herein allow for false positive or false negative detection errors, or differences in capture rate constants between the target analyte and the reference analyte, to account for the accuracy of an estimate of the distribution of a target analyte in a mixed unknown sample. to improve In some embodiments, reference analyte only-control sample to account for false-positive target analyte detection error, target analyte only-control sample to account for false-negative target analyte detection error, and between the target analyte and the reference analyte A series of control samples, including one or more mixed control samples to account for differences in capture rate constants, are used to improve the accuracy of estimates of the fraction of distribution.

In some embodiments, the capture rate between the target analyte and the reference analyte in the mixed unknown sample is relatively constant, thus eliminating the need to use a mixed control to improve the estimate of relative abundance. In some embodiments, the relative capture rate between the target analyte and the reference analyte in a mixed sample is an estimate of the fraction of distribution without running a mixed control sample by applying a correction term to data from a mixed unknown sample to compensate for this difference. is known to improve In some embodiments, data from a mixed control sample run under substantially identical nanopore conditions using the same target analyte and reference analyte species as in the mixed unknown sample is actually running the mixed control sample as part of a method. It can be used to improve the estimate of the fraction of distribution without

In some embodiments, a threshold is determined such that false positive values from the mixed unknown sample are negligible and that a reference analyte-only-control does not need to be used to improve the estimate of relative abundance. In some embodiments, false positive values from a mixed sample are such that a correction term can be applied to data from a mixed unknown sample to compensate for false positive errors to improve the estimate of the fraction of distribution without running a reference analyte-only-control sample. is known In some embodiments, data from a reference analyte-only-control sample run under substantially the same nanopore conditions using the same reference analyte as in the mixed unknown sample actually determines the reference analyte-only-control sample. It can be used to improve the estimate of the fraction of a distribution without running as part of it.

In some embodiments, a threshold is determined such that false negative values from the mixed unknown sample are negligible and that a reference analyte-only-control does not need to be used to improve the estimate of relative abundance. In some embodiments, false negative values from a mixed sample are such that a correction term can be applied to data from a mixed unknown sample to compensate for false negative errors to improve the estimate of the fraction of distribution without running a reference analyte-only-control sample. is known In some embodiments, data from a reference analyte-only-control sample run under substantially the same nanopore conditions using the same reference analyte as in the mixed unknown sample actually determines the reference analyte-only-control sample. It can be used to improve the estimate of the fraction of a distribution without running as part of it.

sample use

As compared to a reference nucleic acid molecule, determining the amount of a target sequence in a nucleic acid fragment has many applications.

In one application, the methods herein are used to determine the amount of a transgene of a genetically modified organism (GMO) relative to a non=GMO reference sequence in genomic DNA, eg obtained from the collection of several seeds. This decision has important regulatory and economic reasons. Buyers and sellers of seeds with desired traits need precise and accurate knowledge about the proportion of seeds with desired traits for fair pricing and trading.

Thus, in some embodiments, the methods provided herein provide for determining % GMO content from seeds, grains, powders, and feeds estimated to include between 1-100% GMO content. Seed developers, growers, and supervisory agencies want precise measurements and the ability to resolve within 10% differences (1.1 times) in GMO content. % GMO is defined as 100 x (GMO event copies) / (taxon-specific genomic reference copies).

In another application, the methods described herein are used to monitor the relative abundance of polynucleotide sequences comprising point mutations to non-mutant (wild-type) sequences in cell-free circulating DNA from blood or urine samples. The relative abundance of point mutations at specific genomic loci was associated with cancer type and treatment outcome. Determination of the relative abundance of mutations to non-mutant sequences can be used for diagnosis, treatment, and monitoring of disease progression. Although tumor imaging results can take weeks to reveal shrinking/growing masses, the method provided herein can rapidly identify the relative abundance of mutant markers, thus enabling efficient and frequent use of easily accessible sample types. Testing (eg, daily) may be performed. Clinically, these techniques can more efficiently reveal treatment responses by providing more perspectives into disease epidemiology, while also enabling early detection of relapses.

In some embodiments, the methods provided herein provide for copy number variation determination (CNV) in a hereditary cancer screening assay. The copy number variation (CNV) test is for hereditary cancer predisposition. The objective is to detect deletions or duplications of gene regulatory elements that differ from the reference by <1.5 fold. For example, a 10% difference (1.1 fold) in the number of copies of the BRCA1 gene may warrant clinical action.

Nanopore detection

Nanopores are formed on a solid-state silicon-based substrate, and single-molecule experiments are performed by applying a voltage across the pores in a buffered electrolyte solution.

1A shows a typical single-molecule event caused by dsDNA passing through a nanopore. Events are quantified in terms of duration width and maximum conductance depth, and maximum δG. The maximum δG is a value obtained by dividing the current attenuation amount δI by the applied voltage V. 1B shows an all-event scatter plot of maximum δG versus duration for 1072 events of 5.6 kb dsDNA recorded for 5 min with 22 nm diameter nanopores (V = 100 mV, 1 nM DNA, 1M LiCl, 10 mM Tris, 1 mM EDTA, pH = 8.8).

In addition to the maximum δG and duration, characteristics of the event profile that can be quantified are: mean δG, median δG, standard deviation of the event signal, and other higher-order characteristics. Another useful feature is the absolute value of the integral area of an event, which can be calculated by multiplying the mean δG by the duration (Storm, AJ, JH Chen, HW Zandbergen, and C Dekker. “Translocatoin of Double-Strand DNA Through a Silicon Oxide Nanopore." Physical Review E 71, no. 5 (May 2005): 051903, doi:10.1103/PhysRevE.71.051903). The integral area, or simply "area", is also known as charge depletion (Folgea, Daniel, Marc Gershow, Bradley Ledden, David S McNabb, Jene A Golovchenko, and Jiali Li. "Detecting Single Stranded DNA with a Solid State Nanopore. " Nano Letters 5, no. 10 (October 2005): 1905-9. doi: 10.1021/nl051199m).

For dsDNA of sufficient length (>700 bp) to pass through the nanopore in the folded state, the event may exhibit one or more amplitudes. 1B is an example of a fully folded event exhibiting a larger maximum δG and shorter duration, and an unfolded event exhibiting a longer duration and shallower maximum δG. A partially collapsed event shows both amplitude levels within an event, starting at a deeper level and ending at a shallower level, with the width of the overall duration between the unfolded and fully collapsed events. The δG and duration distributions show a mixture of modes for foldable dsDNA, whereas the event area is a single mode distribution for dsDNA, whether or not the DNA has sufficient length to fold as it passes through the nanopore. has

The differentiation of target analytes and reference analytes using nanopores is based on the detection of sufficiently different event signals upon each translocation through the nanopores to enable reliable and sensitive detection. Differences in the average event signal may be based on signal duration, change in current, characteristics within the signal, or other distinguishable characteristics, and combinations thereof. The characteristics used are used as a basis for determining limits that serve as a method of identifying event signals associated with reference analytes and target analytes used in the fraction of distribution determination described herein.

In some embodiments, the target and reference fragments are dsDNA molecules of sufficiently different lengths to produce different nanopore event durations.

In some embodiments, the target analyte and reference analyte are both dsDNA, and the characteristic that generates distinct event types may be a difference in length of the target analyte and reference analyte. In such embodiments, differences in target and reference event areas, created by differences in lengths of target and reference analytes, can be used to differentiate target and reference event signals (ie, event profiles).

The event area distribution for dsDNA has a single mode. This is when the target and reference analytes are dsDNA of sufficiently different lengths, the area becomes a useful event characteristic for classifying the event into a target phenotype or reference type. To generate sufficiently different area distributions, the length should be at least 100 bp difference in length for nanopores with a diameter greater than 20 nm. For smaller nanopores 1-20 nm in diameter, such as those formed by controlled breakdown (Yanagi, Itaru, Rena Akahori, Toshiyuki Hatano, and Ken-ichi Takeda. "Fabricating Nanopores with Diameters of Sub-1 Nm to 3 Nm Using Multilevel Pulse-Voltage Injection." Scientific Reports 4 (2014): 5000 doi:10.1038/srep05000), dsDNA for target and reference must have a length difference of at least 20 bp.

There is no clear upper limit on how different dsDNA lengths can be for target and reference molecules.

Figure 2A shows a typical event when 727 bp of DNA passes through a solid-state nanopore with a diameter of 25 nm at 100 mV in 1M LiCl. The event area is shown as a shaded area. Figure 2B shows how the event area increases with dsDNA length. Primarily, event duration increases while event depth is maintained, and event area (average depth × duration) is proportional to duration, thus capturing this length-dependent increase. 2C shows the distribution of the base 10 logarithm of the area (pA*ms) of all events recorded for each DNA indicated, run sequentially on the same nanopore. The distribution of the logarithmic base 10 of the event area is nearly normal (Gaussian). As the length of DNA increases, the mean of the distribution increases.

Two dsDNAs of at least 300 bp in length and at most 100,000 bp in length, a target-sequence comprising dsDNA and a reference-sequence comprising dsDNA were generated. In some embodiments, the target and reference dsDNA analytes are at least 10 bp, 20 bp, 30 bp, 40 bp, 50 bp, 60 bp, 70 bp, 80 bp, 90 bp 100 bp, 150 bp, 200 bp or It has a difference of 300 bp in length. In general, increased differences in length between target and reference dsDNA analytes, when distinguished by size, facilitate higher sensitivity and specificity of the determination of event signals associated with target and reference analytes, and increase the relative abundance in the sample. Estimation is improved.

In some embodiments, characterizing polynucleotide fragments cleaved from genomic DNA (gDNA) is part of the workflow for fractional distribution determination. Such fragment characteristics may include, for example, its sequence, length, and secondary structure. In some embodiments, such fragment features enhance capture and detection of specific sequences by the nanopore device.

In some embodiments, the target and reference fragments are bound to different payload molecules such that the target/payload molecule and the reference payload molecule generate sufficiently different nanopore event signals. In some implementations, the other event signal is a combination of event duration, event maximum depth, event average depth, and/or other event characteristics.

In some embodiments, the target and reference analytes are sequence-specific, wherein a unique nanopore event signal is generated as each molecule or complex type (target-payload, reference-payload) passes through the pore. distinguished by the load. Methods using probes bound to payloads that bind to each molecular type to facilitate discrimination are described in International Patent Application No. WO/2015/171169 "Target Detection with a Nanopore", International Patent Application No. WO/2014/182634 " A Method of Biological Target Detection Using a Nanopore and a Fusion Protein Binding Agent", International Patent Application No. WO/2016/049657 "Target Sequence Detection by Nanopore Sensing of Synthetic Probes" International Patent Application No. WO/2016/126746 "Nanopore" Detection of Target Polynucleotides from Sample Background" and International Patent Application No. WO/2017/173392 "Nanopore Discrimination of Target Polynucleotides from Sample Background by Fragmentation and Payload Binding", each of which is incorporated herein by reference in its entirety included as

In some embodiments, the target and/or reference analyte is a dsDNA, which invades each dsDNA type (target and reference) resulting in two types of macromolecules that are detected as nanopores with native payload-binding PNAs. In some embodiments, the target and/or reference analyte is a single-stranded nucleic acid (ssNA), including DNA or RNA. A payload-bound complementary nucleic acid (eg, LNA) hybridizes to one region on the ssNA and one or more flanking primers to another region on the ssNA to generate a double-stranded molecule with the bound payload, wherein The load is unique to the target and reference to create a unique target and reference event profile.

Fractional Distribution Framework

In some embodiments, the fractional distribution framework comprises: 1) designing and applying a biochemical method for converting a sample material into a nanopore detection format, for both target analyte and reference analyte types; 2) applying a specific nanopore experimental protocol; and 3) applying an analytical step that produces a quantitative estimate of the relative abundance of the target relative to the reference analyte. This section focuses on Part 1 of the framework.

Sample preparation for nanopore detection

A molecule comprising a target sequence (referred to as a "target analyte" or "target molecule") and a molecule comprising a reference sequence (referred to as a "reference analyte" or "reference molecule") may be physically similar: For example, the target and reference molecules may be similar in molecular weight, or polynucleotide length, and may differ by only one nucleotide. The goal of biochemical methods is to render target and reference molecules without bias to generate distinct “target” or “reference” event profiles upon translocation through nanopores. In this way, the target:reference mixture measured on the nanopore represents the concentration ratio of the target:reference in the sample.

In some applications it may be advantageous to create distinct event profiles by adding polynucleotide sequences to the target, reference, or both molecules. For example, most of the DNA fragments obtained from cell-free circulating DNA in blood or urine have a uniformly short length of 150-200 bp. Adding polynucleotide sequences by conventional methods including PCR, ligation, and direct oligonucleotide hybridization allows flexibility to maximize the differentiation of nanopore events. In other cases, hybridization of chemically modified oligonucleotide probes carrying covalently linked polymeric payloads can be used to alter the charge and molecular weight of a target or reference analyte without affecting polynucleotide length. In all cases, the target is a separate event profile per group of target and reference molecules.

There are applications where there is sufficient starting material to use the enrichment strategy without the required PCR prior to nanopore detection, including examples of GMOs (amounts of soybean seeds containing GMO target sequences). There are other instances where PCR is required as part of the enrichment, including liquid biopsies, as blood or urine samples may contain < 10 target sequences per mL of fluid. The proposed method is independent of sample preparation requirements, including sample collection, purification and concentration of targets and references. Nanopore measurement and/or analysis if the target and reference are sufficiently enriched (>10 pM) relative to background (< 1 pM) and the target and reference analytes are present and produce electrical event signals that are distinguishable from the background and from the other. Subsequent quantification of the distribution fraction can be implemented.

In some embodiments, the target or reference analyte comprises a polynucleotide sequence (including double and single stranded DNA, RNA, and synthetic polynucleotides) between 20 nt and 100,000 nt in length. In some embodiments, a polynucleotide comprising a target sequence is derived from gDNA of a living organism, including a plant, human, animal, insect, bacterium, or virus. In some embodiments, the target polynucleotide sequence is derived from an exogenous, non-genomic sequence comprising double or single-stranded RNA or DNA from a source comprising a plasmid, a BAC, a linear sequence-verified gene block, an expression cassette. .

In some embodiments, specific enrichment is provided for detection of a fraction of a distribution (eg, copy number variation) by a nanopore device. In some embodiments, site-directed fragmentation methods are used to prepare samples for detection by nanopores. In some embodiments, the detection methods provided herein include upstream fragmentation of a polynucleotide fragment of a nucleic acid sample, eg, fragmentation of gDNA to a length of 20-100,000 nt or base pairs in size. In some embodiments, the nucleic acid is prepared using restriction enzymes or using site-directed nucleases including Cas9/sgRNA, TALENS, zinc finger protein/nuclease or other fragmentation methods known in the art. It is sequence-specifically fragmented.

In some embodiments, target or reference analyte enrichment is performed using positive and negative size selection that maintains, excludes, and elutes the target fragment size. For example, after retention and exclusion of high molecular weight polynucleotide species (e.g., >8,000 bp DNA) with low SPRI bead:DNA ratio (0.6) in the presence of PEG, fragment size with SPRI bead:DNA (1.5:1) (eg 2000-8000 bp) bind, wash and elute. In some embodiments, a target or reference nucleic acid may be subjected to nucleic acid amplification to facilitate detection within the nanopore.

Nanopore detection

The fraction of distribution framework comprises the steps of: 1) designing and applying biochemical methods to convert sample materials into nanopore sensing formats, for both target analytes and reference types; 2) applying a specific nanopore experimental protocol; and 3) applying a mathematical method to generate a quantitative estimate of the dose of the target (target:reference) relative to the reference analyte. This section focuses on the experimental protocol, part 2.

Repetition of a sample run within a nanopore is described herein to provide an improved estimate of the true relative abundance of a target analyte in a mixed unknown sample. In some embodiments, the target analyte and the reference analyte are prepared to enable reliable discrimination between species using a nanopore sensor. In some embodiments, characteristics of a fragment comprising a target sequence (i.e., a "target fragment") and a fragment comprising a reference sequence (i.e., a "reference fragment") are such that the two fragments are distinguished by one or more signal properties. nanopores are selected to generate an event signal that can be

In some embodiments, one or more control mixtures (ie, control samples) are used to calibrate an estimate of the amount of target relative to a reference in the unknown mixture. In some embodiments, the calibration compensates for differences in nanopore capture efficiency between target and reference molecule types.

In some embodiments, an unknown mixture of a target and a reference analyte is measured on the nanopore, and the fraction distribution of the target relative to the reference is quantified mathematically. In some embodiments, one or more unknown mixtures of target and reference molecule types, derived from the same sample, are sequentially measured on the same nanopore. In some embodiments, one or more unknown mixtures of target and reference molecule types, derived from the same sample, are measured in parallel on different nanopores.

In some embodiments, one or more controls, comprising 100% target alone, 100% reference alone, and a mixture of known target and reference molecules, are measured on the nanopores before and/or after said unknown mixture.

In some embodiments, the experimental protocol comprises sequentially running one or more controls on the nanopores before or after, or before and after running the unknown mixture on the nanopores. The control may consist of 100% target analyte, or 100% reference analyte, referred to as “isolated controls”. The control may also be any known mixture of target and reference analytes, referred to as “mixed control” or “control mixture”. The control mixture is a target:reference analyte ratio of 1:1, any other target:control ratio from 0.01:1 to 100:1, or any ratio less than 0.01:1 of target:reference analyte (eg, 0.001: 1) or any ratio greater than 100:1 (eg 1000:1). One or more controls may be run more than once. Controls (isolated and mixtures) and unknown mixtures can be run sequentially in any order on the same nanopore. Between the control and unknown samples, the fluidic channels (ie chambers) in which the nanopores trap molecules are flushed.

In some embodiments, no controls are run, only unknown mixtures are run and compared to reference tables established by running controls in a separate prior experiment, ie no controls are run at the point of use.

In some embodiments, the one or more fluidly isolated channel and nanopore sensors measure a control in parallel with the one or more fluidly isolated channel and nanopore sensors that measure unknowns. One or more nanopores may access each fluidic channel. In a parallelized implementation, flushing may not be necessary as each pore sees only one set of reagents, either a control (isolated or mixture) or unknown (in one or more unknown sets).

In some embodiments, the ratio of reference analyte to target analyte in the control mixture concentration is close to the expected ratio of reference analyte to target analyte in an unknown sample, although this may not be known in advance.

Any number of unknown mixtures can be run sequentially on the same nanopore by flushing the old blanks before each new blank is added for measurement. This requires that the unknown mixture consist of the same target and reference analyte types, but their proportions may be the same or different in different unknowns.

Each recording time should be long enough to detect a minimum of 100 events for each reagent type, and the higher the number of recorded events, the better the performance, where the improvement is significant when more than 500 events are recorded; It is very noticeable when more than one event is recorded. The recording time for each reagent set can be the same or different. The adaptive approach can dynamically stop recording when a target molecule number of molecules is detected. We have established a method for determining the number of molecules required to obtain a desired level of confidence (e.g., 95%, 98%, 99%, 99.9%, etc.) applicable to any set of reagents (control or unknown). (SI Section 10.2, Morin, Trevor J, Tyler Shropshire, Xu Liu, Kyle Briggs, Cindy Huynh, Vincent Tabard-Cossa, Hongyun Wang, and William B Dunbar. “Nanopore-Based Target Sequence Detection” Edited by Meni Wanunu. PloS One 11, no. 5 (May 5, 2016): e0154426-21. doi:10.1371/journal.pone.0154426).

In some embodiments, experimental protocols using single nanopores are 1) running 100% target for recording time T, 2) flushing the nanopore chamber, 3) running 100% reference for recording time T, 4) flushing the nanopore chamber, 5) running a 50:50 target:reference mixture for recording time T, 6) flushing the nanopore chamber, and 7) running unknown mixture for recording time T. The recording time T may be 15 seconds, 30 seconds, 45 seconds, 1 minute, 5 minutes, 10 minutes, or any period between 1-15 seconds or 10-60 minutes.

Another common experimental protocol is to run (1)-(7), then 8) flush the nanopore chamber, 9) repeat 100% target for recording time T, 10) flush the nanopore chamber, 11 ) repeat 100% reference for recording time T, 12) flush the nanopore chamber, and 13) repeat 50:50 target:reference mixture for recording time T.

Another common experimental protocol is to run (1)-(7), then 8) flush the nanopore chamber, 9) repeat the 50:50 target:reference mixture for recording time T, and 10) flush the nanopore chamber. flush, 11) repeat 100% reference for write time T, 12) flush the nanopore chamber, 13) repeat 100% target for write time T.

Another common experimental protocol is 1) running a suspected target:reference control mixture ratio close to the target:reference ratio in the unknown mixture, during recording time T, 2) flushing the nanopore chamber, 3) recording time T while running the unknown mixture.

Another common experimental protocol is to 1) run a 1:1 target:reference control mix ratio for recording time T, 2) flush the nanopore chamber, and 3) run unknown mixture for recording time T.

In some embodiments, an experimental protocol using a single nanopore is 1) running 100% target for recording time T, 2) flushing the nanopore chamber, 3) running 100% reference for recording time T, 4) flush the nanopore chamber, and 5) run the unknown mixture for recording time T.

In some embodiments, one experimental protocol using single nanopores is to 1) run 100% target for recording time T, 3) flush the nanopore chamber, and 4) run unknown mixture for recording time T.

In some embodiments, one experimental protocol using single nanopores is to 1) run a 100% reference for a recording time T, 2) flush the nanopore chamber, and 3) run an unknown mixture for a recording time T.

In some embodiments, one experimental protocol using single nanopores runs only unknown mixture for recording time T, 100% reference control sample, 100% target control sample, known target:reference control mixture, or any thereof Data from a lookup table containing error correction information derived from the combination or previous data is used, each run under conditions substantially similar to the experimental protocol for the unknown mixture, and data generated from recording time T At least one correction term is provided to improve the estimate of the fraction of distribution of the target analyte in the unknown mixture.

Upon completion of the experimental protocol, events recorded from controls (if executed) and events recorded from unknown(s) are mathematically analyzed to predict the amount of target relative to a reference within one or more unknowns.

Estimating fractions of distribution and determining limits

The fraction of distribution framework comprises the steps of: 1) designing and applying biochemical methods to convert sample materials into nanopore sensing formats, for both target analytes and reference types; 2) applying a specific nanopore experimental protocol; and 3) generating a quantitative estimate for the amount of the target analyte (target:reference) referenced. This section focuses on Part 3 of the framework.

In some cases, the estimated concentration ratio R=[t]/[r] of the target sequence “t” to the reference sequence “r” is quantified. The transgene percentage or GMO% converts the ratio R into a percentage. In some cases, the estimated amount of the target sequence relative to the total (target and reference sequence) F=[t]/([t] + [r]) is quantified. There is a simple conversion between said ratio R and the quantity F: F = R/(R + 1) or, equivalently, R = F/(1 - F).

Fractional distribution methods predict the relative amount of a target to a reference, or a target to the whole (sum of target and reference). In some embodiments, a calibration molecule may be added to determine the absolute concentration of the target or reference molecule. In some embodiments, the relative capture rate of the target molecule to the calibration molecule at a constant concentration may be related to the concentration of the target molecule in the sample, wherein information derived from the plurality of nanopores is used to determine the concentration of the target molecule. can be

In some embodiments, single nanopore event characteristics are compared between a target and a reference analyte to calculate a fraction of distribution. In some embodiments, one or more nanopore event characteristics are compared between target and reference analyte types to calculate a fraction of distribution.

There are three methods described herein to improve limit determination, which uses event signals from nanopores to differentiate between target and reference analytes and to determine the fraction of distribution, to correct for error: 1 ) Q- test, 2) Support Vector Machine (SVM), 3) Expectation Maximization Algorithm (EMGM) method for Gaussian mixture.

및

로 표시된다. 표적 및 참조 분자 구성은 구별되는 나노포어 이벤트 신호를 제공하도록 고안 및 생성된다.The following general concepts apply to methods. First, the true ratio of target analyte "t" to reference analyte "r" is expressed as R=[t]/[r]. The true fraction of target analyte relative to total (target and reference) analytes is expressed as F=[t]/([t]+[r]). A simple conversion between said ratio R and fraction F is F = R/(R+1) or, equivalently, R = F/(1 - F). _{The true fraction of the unknown} mixture is R _{mix and} the true fraction of the mixture is denoted as F mix. The mathematical method _{produces estimates of F mix} and R _mix , which are

and

is displayed as Target and reference molecular constructs are designed and generated to provide distinct nanopore event signals.

Q test

The mathematical method first devises criteria for classifying all recorded events into one or two categories: target positive (equivalently, reference negative) or target negative (equivalently, reference positive). An event criterion utilizes one or more event characteristics. In some implementations, a single characteristic is used to create a criterion to classify an event. Given the criteria, each event is tagged as either a target event or a reference event. These are referred to as "target-tags" or "reference-tags".

The fraction of target-tagged events is expressed as Q , equal to the number of target-tagged events divided by the total number of events. The fraction of reference-tagged events is 1- Q . The tagged fraction Q is a function of the nanopore concentration fraction F , denoted Q (F).

의 경우, Q(z) = Q _X:Y 이다. 일부 구현예에서, z = 0.5인 1:1 비율 대조군 혼합물이 바람직하고 태그된 분율은 Q _1:1 또는 Q _50:50 으로 표기된다. The fraction Q (F _mix ) of target-tagged events in the mixture is expressed as Q _mix; The fraction Q (1) of target-tagged events in 100% target control is expressed as Q _{targ ;} The fraction Q (0) of target-tagged events in the 100% reference control is expressed as Q _{ref ;} The fraction of target-tagged events in the target:reference control mixture is expressed as Q _X:Y where X:Y is the ratio of the target-to-reference mixture in the control mixture. fraction

For , Q(z) = Q _X:Y . In some embodiments, a 1:1 ratio control mixture with z =0.5 is preferred and the tagged fraction is denoted Q _1:1 or Q _50:50.

In general, Q _targ is close to 1, and 1- Q _targ represents the false negative fraction. In general, Q _ref is close to 0, and Q _ref represents the false positive fraction. The control group satisfies Q _targ > Q _X:Y > Q _{ref .} The mixed group satisfies Q _targ > Q _mix > Q _reff .

In some embodiments, target-tagged fractions from controls ( Q _targ , Q _ref , Q _X:Y ) are run individually, and lookup tables are used to reference values for any new assay that measures Q _mix. used In some embodiments, ( Q _targ , Q _ref , Q _X:Y ) is set at the point of use as part of the analysis. In some implementations, ( Q _targ , Q _ref ) is run separately and a lookup table is used to reference its values, whereas ( Q _X:Y ) values are used as part of an analysis to measure Q _{mix at the point of use.} is set

In some embodiments, target-tagged fractions from controls ( Q _targ , Q _ref , Q _X:Y ) are run one or more times at the point of use, the values of which are averaged for subsequent use in the equation below.

에 대한 식은 다음과 같이 주어진다:Estimate of true quantity F _mix

The expression for is given as:

(식 1)

(Equation 1)

, 및

이다.here

, and

to be.

에 대한 식은 다음과 같이 주어진다:Estimate of true ratio R _mix

The expression for is given as:

(식 2)

(Equation 2)

(%)와 같다.In the example for predicting the amount of transgene (GMO), GMO (%) is

Same as (%).

는 위양성 검출 오차, 위음성 검출 오차, 또는 둘 다에 대하여 보상할 수 있는 비율의 추정치이다. 일부 구현예에서,

의 값은 위양성 오차에 대하여 보상하는 데 사용될 수 있다. 위양성 오차에 대한 보상이 사용되지 않으면,

은 0으로 설정될 수 있다. 일부 구현예에서,

의 값은 위음성 오차에 대하여 보상하는 데 사용될 수 있다. 위음성 오차에 대한 보상이 사용되지 않으면,

은 0으로 설정될 수 있다.the above parameters

is an estimate of the ratio that can be compensated for false-positive detection error, false-negative detection error, or both. In some embodiments,

The value of can be used to compensate for false-positive errors. If compensation for false-positive errors is not used,

may be set to 0. In some embodiments,

The value of can be used to compensate for false negative errors. If compensation for false negative error is not used,

may be set to 0.

는 비율 보상 승수이다. 분석적으로, 상기 파라미터

는 두 포획률 상수의 비율이다. 포획률 상수는 나노포어 이벤트 비율을 주어진 분자 유형에 대한 농도로 나눈 값이다. 구체적으로, 상기 파라미터

는 참조 분자 포획률 상수를 표적 분석물 포획률 상수로 나눈 값이다. 따라서, 상기 승수

는 표적 및 참조 분자 유형 간의 나노포어 포획 및 검출의 차이에 대하여 보상한다.the above parameters

is the ratio reward multiplier. Analytical, the parameter

is the ratio of the two capture rate constants. The capture rate constant is the rate of nanopore events divided by the concentration for a given molecular type. Specifically, the parameter

is the reference molecule capture rate constant divided by the target analyte capture rate constant. Therefore, the multiplier

compensates for differences in nanopore capture and detection between target and reference molecule types.

이다.When the control mixture is in a 1:1 ratio,

to be.

는 식 (1) 및 (2)에서 1로 설정되어,

및

각각에 대한 추정치를 제공한다.When compensation for the difference in capture rate constants of the target analyte and the reference analyte is not used,

is set to 1 in equations (1) and (2),

and

An estimate is provided for each.

및

각각에 대한 추정치를 제공한다.

및

에 대한, 불확실성 추정치, 또는 오차 막대 역시 계산될 수 있다. 분리된 및 혼합 대조군에 대한 및 미공지 혼합물에 대한 각 Q는 그와 관련된 표준 오차, std(Q)

를 가지며, 여기서 N은 전체 이벤트의 수이다. 수치적으로, 각각의 Q 분포로부터의 무작위적 샘플이 여러 번 추출되어, 식 (1) 및 (2)를 적용하여

및

에 대한 값의 분포를 생성할 수 있다. 그런 다음

및

에 대한 분포를 이용하여 불확실성 한계를 계산할 수 있으며, 결과적으로

및

이 된다.Applying equations (1) and (2), we get

and

An estimate is provided for each.

and

Uncertainty estimates, or error bars, can also be computed for . Each Q for the isolated and mixed controls and for the unknown mixture is the standard error associated with it, std( Q)

, where N is the total number of events. Numerically, a random sample from each Q distribution was taken multiple times, applying equations (1) and (2) to

and

You can create a distribution of values for . after that

and

The uncertainty limit can be calculated using the distribution for

and

becomes this

In some embodiments, the proportion or fraction of events meeting or deviating from the event characteristic criterion is used to estimate the amount of target relative to a reference in the unknown mixture. In some embodiments, the criterion is a limit.

이며, 여기서 N은 전체 이벤트의 수이다. 동일한 기준이 모든 대조군, 분리된 것 및 혼합물, 및 모든 미공지에 적용되어, 위의 식들(식 (1)-(2))에 사용되는 모든 Q 값을 계산한다. Our previous work describes a method for calculating Q and its error bars using a single tag criterion (Morin, Trevor J, Tyler Shropshire, Xu Liu, Kyle Briggs, Cindy Huynh, Vincent Tabard-Cossa, Hongyun Wang, and William B Dunbar. "Nanopore-Based Target Sequence Detection." Edited by Meni Wanunu. PloS One 11, no. 5 (May 5, 2016): e0154426-21. doi:10.1371/journal.pone.0154426). As detailed in the work above, applying the criteria, each event j has a variable Z _j assigned to it. If event j is tagged, then Z _j = 1; Otherwise, Z _j = 0. For each reagent set (control and unknown),

, where N is the total number of events. The same criteria were applied to all controls, isolates and mixtures, and to all unknowns to calculate all Q values used in the above equations (Eqs. (1)-(2)).

The criterion includes one or more inequalities and may be a linear or non-linear function of one or more event characteristics. Each inequality has an associated limit or range of limits. Thus, the criterion is fully specified by a set of inequalities and a corresponding set of limits.

In some embodiments, the criteria are established for classes of target and reference molecule types, and new analyzes using the types of molecules for that class may use the predefined criteria.

In some embodiments, the criteria are identified from control data collected for any new analysis. That is, the criteria are set at run time as part of the operational fractional distribution protocol.

In some embodiments, the set of inequalities for the criterion is established a priori from a previous set of experiments with comparable target and reference molecule types, while the limit set for one or more criterion inequalities is set at run time using control data. .

In some embodiments, a single event characteristic is used to establish the criteria.

The limit, denoted " q, ", is a scalar value that distinguishes target-tagged events from non-target-tagged (ie, reference-tagged) events based on the inequality. When more than one inequality is used within a criterion, q may represent a vector of bound values used in the set of inequalities.

Consider the example of using two dsDNAs of different lengths for target and reference. In general, a single inequality using the event area is a viable criterion. If the target is a dsDNA that is longer than the reference dsDNA, the event is tagged if the area exceeds the limit. If the target is a dsDNA that is shorter in length than the reference dsDNA, an event is tagged if the area is less than the limit.

Various methods can be used to automate the selection of a q- threshold value or threshold values, where one q value is identified for each inequality within the criterion.

In some embodiments, the q- limit is found to be a value that produces a desired false positive for Q _{ref .} For example, the q- limit may be set to the 95th percentile of Q _ref to produce 5% false positives. In this case, 95% of the reference molecular events have an area less than q. Alternatively, the SFT q- limit is found to be a value that produces a desired false negative for Q _targ , ie, the q- limit may be set to the 5th percentile of Q _targ to produce a false negative of 5%.

에 대한 해로서 발견된다. 상기 한계는 Q _targ 와 Q _ref 사이의 최대 거리에 해당하는 값일 수 있다.In some embodiments, the SFT q- limit is

is found as a solution to The limit may be a value corresponding to the maximum distance between Q _targ and Q _ref.

In some embodiments, the q- limit range is calculated as a value that produces a desired false positive range for Q _{ref .} For example, the q- limit range may be the 95th to 99th percentile range of Q _{ref .}

및

값의 범위를 생성하고, 이들 범위의 평균이 계산되고 예측된

및

으로 보고된다. In some embodiments, where the q- limit range is used, equations (1) and (2) are

and

Generate a range of values, and the average of these ranges is calculated and predicted

and

is reported as

Consider an example using two different payloads bound to a target DNA and a reference DNA. In general, three inequalities using event mean conductance and event duration are viable criteria. Specifically, for a particular payload-target DNA molecule construct, a target event creates a unique subspace on a 2D event plot of mean δG versus duration, with duration greater than the limit, and average δG above one limit, or An event is tagged when it is below another limit. In this case, the tag criterion is expressed as three linear inequalities and three limits, using two event features (mean δG and duration).

SVM method

In some embodiments, machine learning is used to identify feature sets and feature criteria for tagging each event as either a target analyte event or a reference analyte event. In some embodiments, a support vector machine is used to classify an event as a target or reference analyte.

In some embodiments, development of a support vector machine workflow comprises the following steps: 1) loading nanopore data, 2) selecting nanopore event characteristics to distinguish events, 3) creating a control group training and testing the model used, 4) calibrating the data using controls, 5) predicting an unknown target:reference mixture. In some implementations, a conventionally developed and reduced support vector machine workflow for automated fractional distribution prediction is implemented.

In some implementations, machine learning tools are applied to automate the selection of criteria, including the selection of event characteristics, the shape of the inequality (linear and/or non-linear), and the threshold q used for the inequality. In some implementations, a support vector machine (SVM), a supervised machine learning method for solving a classification problem, is implemented to generate tag criteria. References on SVM are: Cortes, C. & Vapnik, V. Machine Learning (1995) 20: 273 and Boser, BE, Guyon, IM, and Vapnik, VN (1992). "A training algorithm for optimal margin classifiers," including the Proceedings of the fifth annual workshop on Computational learning theory , each of which is incorporated herein by reference in its entirety .

An example of applying the SVM method to the distributed fraction framework of the present invention is as follows:

For linearly separable data, let { x ₁ , ..., x _n } be the data set and y _i ∈ {1, -1} be the class label of x _i , the decision boundary is You should classify:

To maximize the margin of classifying all points, the classification problem becomes an optimization problem as follows:

(식 3)Minimize

(Equation 3)

Data points close to the decision boundary are called support vectors.

For practical problems, the data are usually not linearly separable due to some singular value or noise. To optimize classification, margins were adjusted to allow for some misclassified points. On the other hand, misclassified cases were punished at a high cost. These margins become soft margins. Soft margin classification can be used by adding a “slack” variable to the cost function ( FIG. 3A ).

로 표시하면, 그 다음 커널 함수 K는 다음과 같이 표현될 수 있다.A second method for dealing with data that is not linearly separable is the kernel method (Boser, BE, et al., cited earlier). This transforms the input feature space into a higher dimensional space. In this way, the data can be separated linearly ( FIG. 3B ). mapping function

, the next kernel function K can be expressed as

(식 4)

(Equation 4)

There is a set of types of kernel functions available. The most common types are:

linear kernel

polynomial kernel

Gaussian (RBF) kernel

In general, both kernel tricks and soft margins are used to generate better solutions to classification problems.

및

를 예측하는 단계. 제공된 예에서, 상기 5 단계의 적용은 더 상세하게 설명된다. 식 (3) 및 (4), 커널 유형, 소프트 마진 상수, 및 커널 함수가 의존하는 임의의 파라미터를 포함하는 하이퍼-파라미터 그리드 검색은 상기 방법을 적용하는 일부로 해결된다. 공통 결정 경계 및 공통 보정 비율을 포함하는 SVM에서 생성된 분석 기반 일반화 모델은 대조군 데이터 세트의 필요 없이 미공지 혼합물에 적용될 수 있다.Applying SVM to nanopore data for fraction distribution has the following steps: 1) loading control and unknown data sets, containing all events for each set; 2) selecting a feature; 3) model training and testing phase; 4) data correction step; and 5)

and

step to predict. In the example provided, the application of the above five steps is described in more detail. A hyper-parameter grid search including equations (3) and (4), kernel type, soft margin constant, and any parameters that the kernel function depends on is solved as part of applying the above method. Analysis-based generalization models generated in SVM with common decision boundaries and common correction ratios can be applied to unknown mixtures without the need for a control data set.

Analysis-based generalization models generated in SVM with common decision boundaries and common correction ratios can be applied to unknown mixtures without the need for a control data set. Other data mining methods including decision trees, neural networks, Native Bayer, logistic regression, K-nearest neighbors and boosting are also claimed as applicable methods for nanopore data.

EMGM Method (Expectation Maximization Algorithm for Gaussian Mixture Models)

In some embodiments, clustering methods are applied to generate criteria for tagging target events and reference events. Each event is tagged as either a target event or a reference event. In some embodiments, fraction distribution is the ratio of target events to the sum of target and reference events. Running a control that provides reward information allows adjustments to improve the estimate of the fraction of distribution.

In some embodiments, the clustering method is a maximum likelihood method applied to a parameterized model of the distribution of one or more event parameters. Iterative application of the maximum likelihood estimate to the control set, along with one set of distributions associated with the target analyte type and another set of distributions associated with the reference analyte type, produces suitable model parameters. Application of the parameterized model to the unknown mixture then produces an assignment of events to the target of the reference distribution(s), and assignment to the target distribution(s) to the target and the total number of events assigned to the reference distribution. The proportion of events that have occurred is used to generate a volume estimate.

The log-likelihood function is used as a metric to track progress in iterations of the algorithm, which iteratively updates the membership assignment of each event in the control data and improves the fit of the distribution to the data. In some implementations, the data is modeled using a mixture of parameterized Gaussian distributions. Methods using finite mixture models to characterize numerical data, including Gaussian mixture models, are well characterized in the fields of statistics and applied mathematics (Hand, David J., Heikki Mannila, and Padhraic Smyth. Principles of data mining. MIT press , 2001).

In some embodiments, given a Gaussian mixture (GM) model, the method maximizes the likelihood function over parameters including mean and covariance of components and mixing coefficients. Since there is no closed solution to the log-likelihood, mode parameters and weights for assigning data are iteratively calculated using an expectation maximization (EM) technique (C.M. Bishop, Pattern Recognition and Machine Learning, Springer, 2006).

The method of applying the EM algorithm applied to the GM model to the nanopore data in order to generate an estimate of the distribution fraction is called EMGM. Like the Q-test, the EMGM method utilizes prior knowledge of one or more nanopore event signals that can be used to differentiate a target event from a reference event.

As noted, the target population may be represented by a single distribution, or by two or more distributions. Likewise, a reference population may be represented by a single distribution, or by two or more distributions. Target and reference distribution(s) are established by applying the above algorithm to one or more separate controls and one or more control mixtures.

Then, after the target distribution(s) have been established, events in the unknown mixture are tagged as target events if they relate to the modeled target distribution(s).

For example, a total of three Gaussian distributions may fit the entire data set in a 1:1 control mixture, with one mode related to target type and two modes to reference.

The algorithm requires only one control mixture for application of EMGM. Consequently, the resulting model can be applied to unknown mixtures. In some embodiments, an additional isolated reference control is used to counteract the effect of false positives. Specifically, the application of the EMGM model to a 100% reference control set the false positive fraction, which was subtracted from the predicted fraction generated by applying the EMGM model to the unknown mixture. This subtraction may be referred to as false positive compensation (or “FP” compensation).

Implementation example implementing multipore consensus call

In some applications, implementations of the system(s) described above and below may generate estimates of the fraction and/or concentration of target sample components from a plurality of pores to produce a more accurate estimate of the fraction and/or concentration. can As such, outputs from individual pores (eg, one or more nanopore devices) can be combined and processed to produce improved estimates of fraction distribution and/or concentration. In particular, estimates of fractions of distribution and/or concentrations can be derived from single or multiple micropores processing different aliquots of a common sample (i.e., from a control, from an unknown mixture) with results from one or more nanopores contributing to the generation of the estimate. Fluid nanopores can be generated from devices (eg, consumables).

Accordingly, the one or more methods may include one or more of the following, as described in greater detail below: applying a voltage to a set of nanopores within the one or more nanopore devices to generate a detectable electronic signal and generating a detectable electronic signal and a target analyte and reference inducing a potential of a charged analyte through the set of nanopores with respect to a portion of the sample comprising units of the analyte; generating a set of event signals from the potential of units of the target analyte and a reference analyte through the set of nanopores; generating, from the set of event signals, a set of parameters corresponding to the set of nanopores and related to the distribution fraction of the target analyte; evaluating each of the parameter sets according to corresponding limit conditions to produce a validated parameter set, each of the corresponding limiting conditions being based on a function of a measure of variability determined over values of the parameter set, wherein generating the parameter set includes maintaining parameter values that satisfy the corresponding limit condition; combining the values of the (verified) parameter sets by a parameter combination operation; and returning an estimate of the distribution fraction of the target analyte based on the output of the parameter combination operation.

15A depicts a flow diagram of a method of using a consensus call from any number of nanopores to determine the distribution fraction of a target analyte in a sample. 15A , a computational system associated with a nanopore device (eg, such as an implementation of a nanopore device described below) receives a set of pore parameters corresponding to the set of pores of the nanopore device (1510). ). The pore parameter set may be derived from an electrical signal (eg, current measurement) output from electrodes connected to the pore set, and the electrical signal, as described above, is [t] to determine an estimate of the relative fraction of distribution. ] and [r] are useful for determining; However, in other embodiments, the pore parameter set may additionally or alternatively be derived from other signals.

The computational system then evaluates ( 1520 ) each set of pore parameters according to one or more limit conditions. The limit condition may include a limit condition based on outputs derived from a plurality of pores. For example, a statistical measure of variability across the two or more sets of pores could be used to design limiting conditions for evaluating whether the output of each pore should be processed along with the output of the other pore to determine an estimate of the fraction of distribution. can Statistical measures of variability include: range of parameter values across multiple pores (eg, quartile ranges), variance of parameter values across multiple pores, standard deviation of parameter values across multiple pores, and any other suitable statistical measure. or one or more of non-statistical measures. In other embodiments, the limit condition may be configured in such a way that each pore is evaluated independently of the output of the other pore.

As shown in Fig. 15A, the calculation system then combines (1530) a subset of the Fore parameters that satisfy each limit condition into a parameter combining operation, based on the evaluation of the Fore parameter sets. The parameter combination operation may output an average parameter value (eg, an average value, a median value, a mode) determined from a set of pore parameters that satisfy each limit condition. In some embodiments, the average parameter value may be a weighted average, wherein the weight given to each parameter used to calculate the weighted average is determined by the limit-based comparison of step 1520 (eg, satisfaction of the corresponding limit condition). degree) can be determined. In one implementation, the distance between a parameter value and its respective limit condition can be used to determine the weight. For example, a parameter value with a lower degree of satisfying the limit condition may be given less weight, and a parameter value with a higher degree of satisfying the limit condition may be given more weight.

As shown in Figure 15A, the computational system then returns a fraction of distribution estimate based on the output of the parameter combination operation of step 1530 (1540), wherein the fraction of distribution estimate is subjected to target analysis, as described above. Describes the percentage of target analyte in the sample compared to the total population of water and reference analytes. Methods similar to those shown in FIG. 15A may also be used to determine the concentration of a target sample component, as described in more detail in the example associated with FIG. 15D below.

15B depicts an embodiment using a consensus call from four pores of a nanopore device to determine the fraction distribution of a target analyte in a sample, wherein FIG. 15B is a more specific embodiment of the method shown in FIG. 15A. As shown in Figure 15B, the outputs derived from the four pores (e.g., electrical signals or other measurements) are used to generate parameter values P1, P2, P3 and P4, in order from smallest to largest: By the calculation system, it is arranged. The logic of the calculation system then defines a quartile range (IQR) as P3-P2 and implements a set of limit conditions to determine whether any of the values P1, P2, P3, and P4 should pass for further analysis. do.

In the first evaluation step 1501, the calculation system compares the value of IQR divided by the average of the parameter values P2 and P3 with the first limit, and if IQR/mean(P2, P3) is greater than the first limit, the experiment is decide to fail (thus P1, P2, P3 and P4 should be discarded). However, if IQR/mean(P2, P3) is less than or equal to the first limit, P2 and P3 are passed for further analysis, and the calculation system evaluates P1 and P4 according to the respective limit conditions.

In a second evaluation step 1502, the computational system compares P1 to a second limit defined as a function of P2, IQR, and X, where X is a constant. 15B , the second limit is defined as P2-X*IQR, and in a specific example, the value of X is set to 1.5. If P1 is less than the second limit, then P1 should be discarded; However, if P1 is greater than or equal to the second limit, then P1 should be passed for further analysis.

In a third evaluation step 1503, the computational system compares P4 to a third limit defined as a function of P3, IQR, and Y, where Y is a constant. As shown in FIG. 15B , the third limit is defined as P3+Y*IQR, and in a specific example, the value of Y is set to 1.5. However, in other embodiments, X and Y need not be equal to each other, and the second and third limits may be defined in any other suitable way. If P4 is greater than the third limit, then P4 should be discarded; However, if P4 is less than or equal to the second limit, then P4 should be passed for further analysis.

As shown in FIG. 15B , the values of P1 , P2 , P3 , and P4 that pass their respective limits are combined 1504 (eg, by determining an average value, determining a weighted average value). In embodiments where a weighted average is determined, the weights w1, w2, w3, and w4 corresponding to each of P1, P2, P3, and P4 are used to generate the weighted average.

15C shows a flow diagram of a method implemented by an implementation of the system of FIG. 15B. More specifically, FIG. 15C shows the structure of logic for evaluating a call to parameter values from multiple nanopores as input and returning an output (eg, average parameter value) with or without confidence values. The above structure may process an input array of the form [ P1, P2, ..., PN ], returning an output of the form (state, result ), where state is reliability (eg, state 0 represents no-response output) , state 1 indicates a return output with no confidence value, and state 2 indicates a return output with confidence value), and the result is a combined parameter value determined from a number of nanopores (eg, average parameter value, weighted average parameter). value). In embodiments where the result is a mean parameter, the mean can be an arithmetic mean, a geometric mean, a weighted mean and/or any suitable mean or combination function.

More specifically, for a zero-length input array 1511, the logic provides a return output (0,0). For an input array 1512 of length 1, the logic provides a return output (1, x), where x is the value of the parameter in the input array. For an input array 1513 of length 2, the logic compares the difference between the two parameter values to a limit condition and provides a return output (2, mean(X)) if the limit condition is satisfied, where mean (X) is the average value of two parameters in the input array. However, if the limit condition is not satisfied, the logic provides a return output (1, mean(X)).

For an input array of length greater than two, the logic determines the IQR of the array and processes a subset of the array that falls within a desired range based on the IQR (1514), where the desired range of Figure 15C is the It is determined based on the values of the 25th and 75th quartile parameters in the array. If the length of the subset is less than half the length of the original array, the logic provides a return output (1, mean(X_good)), where X_good is the subset of the array that falls within the desired range. If the length of the subset is less than half the length of the original array, the logic enters region 1516, where the difference between the maximum and minimum values of the subset of the array is greater than a limit TH, and the array If the difference between the maximum and minimum values of the IQR values of is greater than TH, the logic provides a return output (1, mean(X_in_IQR)), where X_in_IQR represents the value of the input array that falls within the IQR. If the difference between the maximum and minimum values of a subset of the array is greater than TH, but the difference between the maximum and minimum values of the IQR values of the array is not greater than TH, the logic determines that the length of the IQR values of the array is greater than 1 Determines if it is greater, and if so, gives the return output (2, mean(X_in_IQR)). However, if the length of the IQR values of the array is not greater than 1, the logic provides a return output (1, mean(X_in_IQR)).

Finally, if the difference between the maximum and minimum values of the subset of the array is not greater than a limit TH, the logic determines whether the length of the subsent of the array is greater than one, and if so, returns the output ( 2, mean(X_good)). If the length of the subset is not greater than 1, the logic provides a return output (1, mean(X)).

In one example, the logic outputs (state, result) of (0, 0) for the input array [], where there is no return output since there is no input nanopore data. In another example, the logic outputs a (state, result) of (1, 0) for an input array [0], where the return output is result 0 and state 1 (there were results from only one nanopore) Considering that, there is no reliability). In another example, the logic outputs (state, result) of (2, 0) for an input array [0, 0], where the return output is result 0, state 2 (both nanopores have the same value 0) Considering that it has been returned, it is reliable). In another example, the logic outputs (state, result) of (1, 30) for input array [30], where the return output is result 30, state is 1 (result from only one nanopore is Considering that there was, there is no reliability). In another example, the logic outputs (state, result) of (2, 30.5) for the input array [30, 31], where the return output is result 30.5, state 2 (data values from both nanopores are Considering the closeness to the limit condition, it is reliable). In another example, the logic outputs (state, result) of (1, 65.0) for the input array [30, 100], where the return output is result 65.0, state 1 (data values from both nanopores are Considering that it is widely separated with respect to the limit condition, it is unreliable). In another example, the logic outputs (state, result) of (2, 30.5) for the input array [30, 31, 100], where the return output is result 30.5, state 2 (data from two nanopores) value is close to the limit condition, and considering that the data values from the third nanopore are treated as outliers, they are reliable). In another example, the logic outputs (status, result) of (2, 33.0) for the input array [30, 31, 32, 33, 34, 35, 36], where the return output is result 33.0, state 2 (reliable considering that the data values from all nanopores are close to the limit condition). In another example, the logic outputs (status, result) of (2, 32.0) for the input array [30, 31, 32, 33, 34, 100, 98], where the return output is result 32.0, state 2 (reliable considering that data values from multiple nanopores are close to the limit condition, and data values from both nanopores are treated as singular values).

As such, embodiments of the described methods may be applied to information derived from any suitable number of nanopores, using any suitable limiting conditions based on other measures of variability.

Also, as noted above, aspects of the method described with respect to FIGS. 15A-15C can be adapted to determine the concentration of a target molecule using information derived from multiple pores. 15D depicts the output of a method for determining the concentration of a target molecule using information derived from multiple nanopores. More specifically, with regard to the output shown in FIG. 15D , one embodiment of the system utilizes information derived from multiple nanopores to compare target sample components (eg, molecules, other analytes) versus specific ones at a constant concentration. The relative capture rate is determined with the concentration of the target sample component to determine the relative capture rate of the calibration molecule, and information derived from the plurality of nanopores is used to determine the concentration of the target sample component.

In the specific example with respect to the output shown in FIG. 15D , the relative capture rates of 74 bP DNA (target molecule) to 217 bP calibration molecule at a concentration of 0.5 nM were compared to other concentrations of 74 bP DNA (e.g., 0.5 nM to 15 nM). concentration) was used to determine the concentration of 74 bP DNA in the samples prepared. Using 1, 2, or 7 control data points with boundary conditions of (0, 0) to fit a linear curve, a particular case derived the concentration of the target molecule from a ratio of the number of events. The output shown in FIG. 15D shows the percentage error of estimates of different concentrations of target molecules using 1, 2, or 7 control data points and using predictions from single versus multiple nanopores. As shown in Figure 15D (top), there is a linear correlation between the ratio of the concentration of the 74 bP DNA target molecule and the number of capture events with the 217 bP calibration molecule. As shown in Figure 15D (middle and bottom), the use of data from multiple nanopores is generally 0.5 nM, 1 nM, 3 nM compared to estimates of the true concentration of the target molecule using data from a single nanopore. , produce lower percentage errors of the estimate of the true concentration of the target molecule at concentrations of 5 nM, 7 nM, 10 nM, and 15 nM, especially as the number of control data points increases.

Pre-filtering data from nanopores based on pore status and data quality

With respect to using data from multiple nanopores to determine an estimate (eg, an estimate of a fraction of a distribution, an estimate of a concentration), some implementations of the system and its associated computational logic depend on the quality of the nanopores themselves (eg, , low- or high-frequency noise, coarse noise statistics including root-mean-square noise, pore diameter, growth rate during experiments, etc. It may be configured to omit the use of pore-derived information and this may be algorithmically automated. As such, the method(s) may include omitting consideration data from the nanopores of the set of nanopores based on an assessment of the quality of the data from the nanopores.

In particular, the system measures the low-frequency noise content (eg, acceptable if the average/median noise power is in the range 0.1-10 Hz or greater, <-50 dB/Hz) for a selected time period (eg, every 5 seconds) within the time domain. ), and the use of data derived from nanopores associated with low-frequency noise content exceeding the threshold level may be omitted. The system optionally or alternatively evaluates the high-frequency noise content (eg, with an average/median noise power in the range of 0.5-30 kHz or greater) for a selected period of time (eg, every 5 seconds) within a time domain and exceeds a threshold level. It is possible to omit the use of data derived from nanopores related to high-frequency noise content. The system optionally or alternatively evaluates a coarse noise content (e.g., RMS of a time domain signal, acceptable if < 20 pA at 30 kHz) over a selected time period (e.g., every 5 seconds) in the time domain and can omit the use of data derived from nanopores related to coarse noise content exceeding the threshold level.

The system also evaluates the rate of change of pore diameter (or other pore morphological characteristics) and/or pore characteristics for a selected period of time (eg, every 5 seconds) within a time domain, where < 0.25 nm/min is acceptable; The use of data derived from nanopores related to out-of-range morphological features and/or rates of change of pore features out of bounds may be omitted. In particular, the method for determining the nanopore diameter over time involves the implementation of a model of open channel conductance, measured as G = I/V, which is a current I and a voltage V. Specifically, G is calculated as the average current divided by the voltage across all pairs of events. The first model for G ignores any access resistance contribution to the total resistance (total resistance is the reciprocal of the total conductance) and depends on the nanopore diameter d and the film thickness L as follows: G ₁ (d) = σ(πd ² )/(4L), where σ is the conductivity of the bulk electrolyte. The first model matches the conductance versus nanopore diameter data when d/L < 3/4. The second model incorporates the effect of access resistance and empirically accommodates film thinning in nanopores when the film thickness L is replaced by an effective thickness l = L/α, α ≥ 1, with all reported d/ Match data against L values (ie, values greater or less than 1). This model is: G ₂ (d, l) = σ(πd ² )/(4l)[1/(1 + (πd)/(4l)].

For a given range of G recorded over the time course of an experiment, we can determine which model is more suitable for comparing the range of modeled values; However, the choice for α has a significant impact on the _{estimated d 2 value, although empirically.} Typically, in the course of an experiment, the value of G = I/V increases slowly over time. If there is an observed increase, there are two potential causes. First, the nanopores expand over time, resulting in an increase in the amount of current; This occurs at a higher rate in the case of less "stable" pores (ie, pores in a film that are fragile for one reason or another, or where the film is very thin and/or can grow due to application of high voltages that can etch the film material). Second, the increase may be due to evaporation of water and a corresponding relative increase in the concentration of ions in the “open” chamber on the nanopores to which the reagent has been added. By replacing the buffer in the exposed chamber and measuring the conductance, it is possible to determine which of these causes are at work - once the values return to their original values, the size and shape of the pores are likely to remain the same; If the value is higher, the pores may be enlarged. In summary, using known L values, if d/L < ¾, the first model G ₁ (d) can be used to estimate the diameter. Otherwise, the second model can be used to estimate the diameter.

The system can also evaluate sample quality content to omit information derived from nanopores (e.g., prior to implementation of the method shown in Figures 15A and 15B) and to omit the use of information derived from nanopores related to poor sample quality content. can The system can evaluate sample quality content with throughput (lower limit, e.g., at least 1 event per minute, and upper limit, e.g., at least 10,000 events per minute) through capture rate or event rate per unit time.

The system can evaluate sample quality content as an isolate of a population for a sample having two or more species present, including one or more unknowns potentially present and one or more controls present. As an example, the sample comprises one control or reference component and one unknown/target sample component present in at least a minimal amount (eg, 2%). When processing such a sample, the system determines a degree of separation value from a model estimate (eg, an SVM-based model) that distinguishes populations within the sample. The separation value output from the model(s) may include a distance value (eg, the shortest distance to a reference separation boundary/hyperplane of the center of a population of events) or other suitable separation.

In one example to evaluate sample quality content, a 50:50 mixture of S2 samples (0%, 100%, as described above) was evaluated with a universal model according to the following steps, as shown in Figure 16A. As shown in FIG. 16A , the method 1600 of the general-purpose model implemented by the system first provides consideration information from the nanopores that generate poor quality data (eg, nanopore size related, abnormal nanopore morphology related, sample It is determined whether a sample population (eg, a population of 50:50 mixtures) is within a target area to filter out or eliminate aberrant interactions with the processing assay, detection of contamination of the sample and/or sample processing material, etc.). (1610). FIG. 16B depicts the output of step 1620, wherein the different reagents used to process the sample (ie, 0% S-adenosylmethionine, 100% S-adenosylmethionine) from nanopores The event data is plotted versus the amplitude versus dwell time of the electrical signal. When generating the plot of FIG. 16B, the system collects event data from all populations, defines a target area, and external events are classified as noise. The system then determines a measure of the noise percentage for each nanopore based on the number of true versus noise events defined by the target area, and if the noise percentage for that particular nanopore is greater than a limit, the specific were removed from the considered data from Fore. As such, the system defined a target area boundary that separates a subset of noise events from a subset of true events within the nanopore data, and determines the noise percentage based on the subset of noise events and the subset of real events. did However, other implementations of step 1610 may be implemented in other ways, with respect to estimating the noise percentage of the nanopores.

As shown in FIG. 16A , the system is then configured to account for information from the nanopores that generate poor quality data (eg, related to abnormal interactions with sample handling analysis, related to detected contamination of the sample and/or sample handling material, etc.) To provide another step for filtering or removing , the separation of components of the sample population is evaluated (1620). When implementing step 1620, the system includes: a residence time (eg, residence time of a sample component with respect to the nanopore), a central amplitude of the electrical signal output from the nanopore, a maximum amplitude of the electrical signal output from the nanopore, the nanopore A principal component analysis (PCA) operation may be performed on one or more of area, and other suitable nanopore-related factors. In particular, the PCA task implements a transformation that transforms data from a first coordinate system to a second coordinate system (i.e., an orthogonal linear transformation) so that the largest variance by some projection of the data is in the first coordinate system of the second coordinate system (i.e., first component). The second largest variance lies at the second coordinate, and the third largest variance lies at the third coordinate. In this way, the PCA task maps the data to new coordinates associated with different levels of variance of the data. When using the PCA component instead of a single parameter value (eg, residence time, median amplitude of an electrical signal output from a nanopore, maximum amplitude of an electrical signal output from a nanopore, nanopore area, etc.) Data separation can be effectively evaluated irrespective of the overlap of parameter values.

As such, the system uses the first component of the PCA task to check the separation of the sample population (eg, with respect to the Gaussian distribution representing each sample population). When checking the separation of a sample population, the system can define the separation score as: SS = (u1-u2)/(s1+s2), where u1 and u2 are the mean of each Gaussian distribution (u2>u1) , and s1 and s2 are the standard deviations of each Gaussian distribution. The separation score SS is then evaluated against limits to determine if the degree of separation is appropriate. Figure 16C shows the output of step 1620 (using the same sample and sample population as Figure 16B), wherein the system uses the components of the PCA task (PC1 and PC2) to provide a Gaussian distribution of coefficients for each sample population. Creates a versus PC1 contrast. 16D , the system determines for the sample population a separation score SS = (u1-u2)/(s1+s2) of 2.4, and if the marginal separation is greater than a pre-defined limit, the system continues to evaluate the data generated from the nanopores. However, other embodiments of step 1620 may be implemented in other ways, with respect to assessing population segregation for a sample.

16A , the system then implements a general-purpose model (eg, a general-purpose model including an instrumented model and a clustering method, which can operate independently and/or cooperatively) for the sample. It is determined whether a correction rate for the sample population is within a target range to produce a prediction of the correction rate ( 1630 ). As shown in Fig. 16E, the system outputs the maximum amplitude of the current from the nanopores using a general-purpose model to check whether the 50:50 calibration ratio of the sample is within the target range of the dwell time in seconds. Generate in nanoamps (max_amp (nA)) for logarithms to base 10. In FIG. 16E , class_labels S2 0% and 100% represent different sample populations, and the universal model outputs %(S2_100%) as 68.91%, which is used to check the 50:50 calibration ratio. However, other implementations of step 1630 may be implemented in other ways, with respect to checking the sample correction ratio. nanopore device

As provided, a nanopore device includes at least one pore forming an opening in a structure that separates the interior space of the device into two volumes, and identifying an object passing through the pore (eg, a parameter representing the object). at least one sensor configured to (by detecting a change). Nanopore devices used in the methods described herein are also disclosed in PCT Publication No. WO/2013/012881, the entire contents of which are incorporated by reference.

The pore(s) of the nanopore device are nano-sized or micro-sized. In one aspect, each pore is sized to allow passage of small or large molecules or microorganisms. In one aspect, each pore is at least about 1 nm in diameter. Alternatively, each pore has a diameter of at least about 2 nm, 3 nm, 4 nm, 5 nm, 6 nm, 7 nm, 8 nm, 9 nm, 10 nm, 11 nm, 12 nm, 13 nm, 14 nm, 15 nm, 16 nm, 17 nm, 18 nm, 19 nm, 20 nm, 25 nm, 30 nm, 35 nm, 40 nm, 45 nm, 50 nm, 60 nm, 70 nm, 80 nm, 90 nm, or 100 nm is nm.

In one aspect, the pores have a diameter of about 100 nm or less. Alternatively, the pores have a diameter of about 95 nm, 90 nm, 85 nm, 80 nm, 75 nm, 70 nm, 65 nm, 60 nm, 55 nm, 50 nm, 45 nm, 40 nm, 35 nm, 30 nm. nm, 25 nm, 20 nm, 15 nm, or 10 nm or less.

In one aspect, the diameter of the pores is between about 1 nm and about 100 nm, or between about 2 nm and about 80 nm, or between about 3 nm and about 70 nm, or between about 4 nm and about 60 nm, or between about 5 nm and about 5 nm in diameter. from about 50 nm, or from about 10 nm to about 40 nm, or from about 15 nm to about 30 nm.

In some aspects, the nanopore device further comprises means for moving the polymer scaffold across the pore and/or means for identifying an object passing through the pore. More details are provided in the section related to the double-pore device below.

Compared to single-pore nanopore devices, double-pore devices can be more easily configured to better control the speed and direction of movement of the polymer scaffold across the pores.

In one embodiment, the nanopore device comprises a plurality of chambers, each chamber communicating with an adjacent chamber through at least one pore. Two of these pores, a first pore and a second pore, are positioned such that at least a portion of the target polynucleotide can exit the first pore and move into the second pore. The device also includes a sensor in each pore capable of identifying the target polynucleotide during the movement. In one aspect, the identifying comprises identifying individual components of the target polynucleotide. In another aspect, the identifying comprises identifying a payload molecule bound to the target polynucleotide. When a single sensor is used, the single sensor may include two electrodes disposed across the pore to measure an ionic current across the pore. In another embodiment, the single sensor comprises a configuration other than an electrode.

In one aspect, the device includes three chambers connected through two pores. Devices having four or more chambers can be devised to include one or more chambers on the side of a three chamber device, or between any two of the three chambers. Likewise, three or more pores may be included in the device to connect the chambers.

In one aspect, there may be two or more pores between two adjacent chambers, such that multiple polymer scaffolds can move simultaneously from one chamber to the next. This multi-pore design can improve the throughput of target polynucleotide analysis in the device. In the case of multiplexing, one chamber may have one type of target polynucleotide and the other chamber may have another type of target polynucleotide.

In some aspects, the device further comprises means for moving the target polynucleotide from one chamber to another. In one aspect, the shift causes simultaneous loading of a target polynucleotide (eg, an amplification product or amplicon comprising a target sequence) across both the first pore and the second pore. In another aspect, the means further enable movement of the target polynucleotide through the two pores, in the same direction.

For example, in a three-chamber double-pore device ("two-pore" device), each chamber may include an electrode for connection to a power source so that a separate voltage is applied across each pore between the chambers. can

According to one embodiment of the present invention, there is provided an apparatus comprising an upper chamber, a central chamber, and a lower chamber, wherein the upper chamber communicates with the central chamber through a first pore, and wherein the central chamber communicates with the central chamber through a second pore. A device is provided in communication with the lower chamber. Such a device may have any size and other characteristics previously disclosed in US Patent Publication No. 2013-0233709 entitled Dual-Pore Device, the entire contents of which are incorporated herein by reference.

In one aspect, each pore is at least about 1 nm in diameter. Further, each pore has a diameter of at least about 2 nm, 3 nm, 4 nm, 5 nm, 6 nm, 7 nm, 8 nm, 9 nm, 10 nm, 11 nm, 12 nm, 13 nm, 14 nm, 15 nm, 16 nm, 17 nm, 18 nm, 19 nm, 20 nm, 25 nm, 30 nm, 35 nm, 40 nm, 45 nm, 50 nm, 60 nm, 70 nm, 80 nm, 90 nm, or 100 nm.

In one aspect, each pore is about 100 nm or less in diameter. In addition, the pores have a diameter of about 95 nm, 90 nm, 85 nm, 80 nm, 75 nm, 70 nm, 65 nm, 60 nm, 55 nm, 50 nm, 45 nm, 40 nm, 35 nm, 30 nm, 25 nm, 20 nm, 15 nm, or 10 nm or less.

In one aspect, the pores have a diameter of from about 1 nm to about 100 nm, or from about 2 nm to about 80 nm, or from about 3 nm to about 70 nm, or from about 4 nm to about 60 nm, or from about 5 nm to about 50 nm, or from about 10 nm to about 40 nm, or from about 15 nm to about 30 nm.

In some aspects, the pores have a substantially round shape. As used herein, “substantially round” refers to a shape that is at least about 80 or 90% cylindrical. In some embodiments, the pores are square, rectangular, triangular, oval, or hexagonal in shape.

In one aspect, the pores have a depth of from about 1 nm to about 10,000 nm, or from about 2 nm to about 9,000 nm, or from about 3 nm to about 8,000 nm, and the like.

In some aspects, the nanopores extend through the membrane. For example, the pore may be a protein channel embedded in a lipid bilayer membrane, or drilling through a solid-state substrate such as a layer made of silicon dioxide, silicon nitride, graphene, or a combination thereof or other materials; It may be formed by etching or other methods of forming pores. The nanopore is sized to allow passage of the pore of the scaffold:fusion:payload, or the product of this molecule after enzymatic activity. In other embodiments, transient blocking of the pores may be desirable for differentiation of molecular types.

In some aspects, the length or depth of the nanopore is of sufficient size to form a channel connecting two separate volumes. In some of these aspects, the depth of each pore is greater than 100 nm, 200 nm, 300 nm, 400 nm, 500 nm, 600 nm, 700 nm, 800 nm, or 900 nm. In some aspects, the depth of each pore is no more than 2000 nm or 1000 nm.

In one aspect, the pores are spaced apart by a distance of about 10 nm to about 1000 nm. In some aspects, the distance between the pores is greater than 1000 nm, 2000 nm, 3000 nm, 4000 nm, 5000 nm, 6000 nm, 7000 nm, 8000 nm, or 9000 nm. In some aspects, the pores are spaced apart by no more than 30000 nm, 20000 nm, or 10000 nm. In one aspect, the distance is at least about 10 nm, or at least about 20 nm, 30 nm, 40 nm, 50 nm, 60 nm, 70 nm, 80 nm, 90 nm, 100 nm, 150 nm, 200 nm, 250 nm, or 300 nm. In other aspects, the distance is less than or equal to about 1000 nm, 900 nm, 800 nm, 700 nm, 600 nm, 500 nm, 400 nm, 300 nm, 250 nm, 200 nm, 150 nm, or 100 nm.

In another aspect, the distance between the pores is from about 20 nm to about 800 nm, from about 30 nm to about 700 nm, from about 40 nm to about 500 nm, or from about 50 nm to about 300 nm.

The two pores enable fluid communication between the chambers and may be arranged in any position as long as they have a defined size and distance between them above. In one aspect, the pores are arranged such that there is no direct blockage therebetween. Also, in one aspect, the pores are substantially coaxial.

In one aspect, the device has electrodes in the chamber that are connected to one or more power sources. In some aspects, the power source comprises a voltage-clamp or patch-clamp, which can supply a voltage across each pore and independently measure the current through each pore. In this aspect, the power supply and the electrode configuration may establish the central chamber as a common ground for both power sources. _{In one aspect, the power source or power sources provide a first voltage V 1} between the upper chamber (chamber A) and the central chamber (chamber B) and a second voltage V between the central chamber and the lower chamber (chamber C). ₂ is configured to apply.

In some aspects, the first voltage V ₁ and the second voltage V ₂ are independently adjustable. In one aspect, the central chamber is adjusted to be a grounding point for the two voltages. In one aspect, the central chamber includes a medium for providing conductance between each pore and an electrode within the central chamber. In one aspect, the central chamber includes a medium for providing resistance between each pore and an electrode within the central chamber. Keeping these resistances small enough relative to the nanopore resistance is useful for decoupling both voltages and currents across the pore, which aids in independent regulation of the voltage.

Adjustment of the voltage can be used to control the movement of charged particles within the chamber. For example, when both voltages are set to the same polarity, properly charged particles can move sequentially from the upper chamber to the central chamber and to the lower chamber, or vice versa. In some aspects, when the two voltages are set to opposite polarities, charged particles can travel from the upper or lower chamber to the central chamber and stay there.

Tuning the voltage in the device can be particularly useful for controlling the movement of large molecules, such as charged polymer scaffolds, that are of sufficient length to pass through two pores at the same time. In this respect, the direction and speed of movement of the molecule can be controlled by the relative magnitude and polarity of the voltage, as described below.

The device may comprise a material suitable for holding a liquid sample, particularly a biological sample, and/or a material suitable for nanofabrication. In one aspect, such materials include dielectric materials, such as silicon, silicon nitride, silicon dioxide, graphene, carbon nanotubes, TiO ₂ , HfO ₂ , Al ₂ O ₃ , or other metal layers, or any combination of these materials, but , but not limited thereto. In some aspects, for example, a single-layer graphene film with a thickness of about 0.3 nm can be used as the pore-retaining film.

Devices can be made by various means and methods that are microfluidic and are equipped with dual-pore microfluidic chip implementations. For microfluidic chips consisting of two parallel membranes, both membranes can be drilled simultaneously with a single beam to form two concentric pores, although using different beams on each side of the membrane is also possible with any suitable alignment technique. . In general, the housing ensures hermetic separation of chambers A-C.

In one aspect, the device comprises a microfluidic chip (denoted as a "double-pore chip") consisting of two parallel membranes connected by a spacer. Each membrane contains pores drilled with a single beam through the center of the membrane. In addition, the device preferably has a Teflon® housing or a polycarbonate housing for the chip. The housing ensures hermetic separation of chambers A-C and provides minimal access resistance to the electrodes so that each voltage is applied primarily across each pore.

More specifically, the pore-containing film may be made of a transmission electron microscope (TEM) grid with a silicon, silicon nitride, or silicon dioxide window of 5-100 nm in thickness. The spacer uses SU-8, photoresist, PECVD oxide, ALD oxide, an insulator, such as ALD alumina, or a vaporized metallic material, such as Ag, Au, or Pt, and the other aqueous portion of chamber B between the films. It occupies a small volume within it, so it can be used to separate membranes. The holder is seated in a water bath consisting of the largest volume fraction of chamber B. Chambers A and C are accessible with larger diameter channels (to lower access resistance) leading to the membrane seal.

A focused electron or ion beam can be used to break through the membrane and align them naturally. The pores can be sculpted (shrunk) into smaller sizes by irradiating the correct beam to focus on each layer. Any single nanopore drilling method can also be used to drill a pair of pores in two membranes, given the drill depth and membrane thickness possible by a given method. Pre-drilling the micro-pores to a predetermined depth and then nanopores through the remainder of the membrane also makes it possible to further refine the film thickness.

Thanks to the voltage present at the pores of the device, charged molecules can move through the pores between chambers. The moving speed and direction can be controlled by the magnitude and polarity of the voltage. In addition, since each of the two voltages can be adjusted independently, the movement direction and speed of the charged molecules can be finely controlled within each chamber.

One example contemplates a target polynucleotide having a length greater than the bound distance comprising the depth of the two pores plus the length between the two pores. For example, a 1000 bp dsDNA would be about 340 nm in length and substantially longer than 40 nm spanning two 10 nm deep pores 20 nm apart. In a first step, the polynucleotide is loaded into either the upper chamber or the lower chamber. Thanks to its negative charge, under physiological conditions of about 7.4 pH, the polynucleotide can migrate across the pore to which a voltage is applied. Therefore, in a second step, two voltages of equal polarity and equal or similar in magnitude are applied to the pores to sequentially move the polynucleotide across the two pores.

By the time the polynucleotide reaches the second pore, one or both of the voltages may be altered. The distance between the two pores was chosen to be shorter than the length of the polynucleotide, so that when the polynucleotide reaches the second pore, it is also within the first pore. Thus, an immediate change in voltage at the first pore will create a force that pushes the polynucleotide away from the second pore.

Both pores have the same voltage force effect | V ₁ | = | V ₂ | Assuming + δV , a value of δV > 0 (or < 0) is | V ₁ | (or V ₂ ) can be adjusted for directional movement. In practice, although the voltage-induced force at each pore is not the same as V ₁ = V ₂ , calibration experiments can identify an appropriate bias voltage that results in the same repulsive force for a given dual-pore chip; A change around that bias voltage can be used for direction control.

At this time, if the magnitude of the voltage-induced force at the first pore is smaller than the voltage-induced force at the second pore, then the polynucleotide passes through both pores and continues toward the second pore, but more will move at a slower pace. In this regard, it is readily understood that the movement speed and direction of the polynucleotide can be controlled by the polarity and magnitude of the two voltages. As will be discussed in more detail below, this precise movement control has broad application. To quantify a target polynucleotide, the usefulness of implementing a dual-pore device is that during controlled delivery and detection, the target polynucleotide or payload-bound target polynucleotide can be repeatedly measured, adding confidence to the detection result. will be.

Accordingly, in one aspect, a method of controlling the migration of a charged polymer scaffold through a nanopore device is provided. The method comprises (a) loading a sample comprising a target polynucleotide (eg, a target polynucleotide amplicon) into one of an upper chamber, a central chamber, or a lower chamber of the device of any of the embodiments, the device comprising: coupled to one or more power sources for providing a first voltage between the upper chamber and the central chamber and a second voltage between the central chamber and the lower chamber; (b) setting an initial first voltage and an initial second voltage such that the target polynucleotide moves between the two chambers so that the polymer scaffold is positioned across both the first and second pores; and (c) adjusting the first voltage and the second voltage such that both voltages generate a force that repels the charged target polynucleotide from the central chamber (voltage-competition mode), wherein the two voltages have a magnitude are different, so that under controlled conditions, the target polynucleotide scaffold crosses the two pores and moves in different directions in a controlled manner.

In one aspect, the sample comprising the target polynucleotide is loaded into the upper chamber and the initial first voltage is set to pull the target polynucleotide from the upper chamber towards the central chamber and the initial second voltage is the target configured to pull polynucleotides from the central chamber towards the lower chamber. Likewise, the sample may initially be loaded into the lower chamber, and the target polynucleotide may be pulled into the central and upper chambers.

In another aspect, the sample comprising the target polynucleotide is loaded into the central chamber, and the initial first voltage is set to pull the charged polymer scaffold from the central chamber to the upper chamber; The initial second voltage is set to pull the target polynucleotide from the central chamber toward the lower chamber.

In one aspect, the real-time or online adjustment of the first voltage and the second voltage in step (c) is performed by active control or feedback control using dedicated hardware and software, at clock rates of up to several hundred megahertz. . The automatic control of the first or second or both voltages is based on feedback of the first or second or both ion current measurements.

sensor

As discussed above, in various aspects, the nanopore device further comprises one or more sensors to effect detection of the target polynucleotide.

The sensor used in the device may be any sensor suitable for identifying a target polynucleotide amplicon bound to or not bound to a payload molecule. For example, the sensor can be configured to identify the target polynucleotide by measuring a current, voltage, pH value, optical property, or residence time associated with the polymer. In another aspect, the sensor may be configured to identify one or more individual components or one or more components of a target polynucleotide bound to or attached to the target polynucleotide. Said sensor may be formed of any component configured to detect a change in a measurable parameter, wherein said change is said target polynucleotide, a component of said target polynucleotide, or preferably a component bound to or attached to said target polynucleotide. indicates In one aspect, the sensor comprises a pair of electrodes disposed on two sides of the pore for measuring the ionic current across the pore as a molecule or other material, in particular a target polynucleotide, moves through the pore. In certain aspects, the ionic current across the pore changes measurably when a portion of the target polynucleotide passing through the pore binds to a payload molecule. This change in current may change in a predictable, measurable manner, corresponding, for example, to the presence, absence, and/or size of the target polynucleotide present.

In a preferred embodiment, the sensor comprises an electrode used to apply a voltage across the nanopore and measure the current. The potential of a molecule through the nanopore provides an electrical impedance (Z) that affects the current through the nanopore according to Ohm's Law, V = IZ, where V is the applied voltage and I is the nanopore The current through, and Z is the impedance. Conversely, the conductance G = 1/Z is monitored to signal and quantify nanopore events. When a molecule is translocated through a nanopore in an electric field (eg, under an applied voltage) the result is a current signal that can be correlated with the molecule passing through the nanopore upon further analysis of the current signal.

When a dwell time measurement from a current signal is used, the magnitude of the component may be associated with a particular component based on the length of time it takes to pass through the sensing device.

In one embodiment, a sensor is provided in a nanopore device that measures optical properties of a polymer, a component (or unit) of the polymer, or a component bound to or attached to the polymer. One example of such a measurement involves the identification of an absorption band unique to a particular unit by infrared (or ultraviolet) spectroscopy.

In some embodiments, the sensor is an electrical sensor. In some embodiments, the sensor detects a fluorescent signal. A radiation source at the exit of the pore may be used to detect the signal.

Equivalence and scope

Those skilled in the art will recognize, or be able to ascertain using no more than routine experimentation, many equivalents to various equivalents of the specific embodiments in accordance with the invention described herein. The scope of the present invention is not intended to be limited to the above detailed description, but is as set forth in the appended claims.

In the claims, articles such as "a," "an," and "the" may mean one or more than one, unless the context clearly dictates otherwise. Claims and descriptions that include “or” between one or more elements of a group indicate that one, two or more, or all elements of that group are present, included, or otherwise in a given product or process, unless the context clearly dictates otherwise. It is considered to be satisfied if it is relevant in this way. The invention includes embodiments in which exactly one element of the group is present, included, or otherwise related to a given product or process. The invention includes embodiments in which two or more, or all, elements of a given group are present in, included in, or otherwise related to in a given product or process.

It should also be noted that the term "comprising" is intended to be open-ended and does not permit, but does not require, the inclusion of additional elements or steps. Where the term “comprising” is used herein, the term “consisting of” is accordingly also included and disclosed.

Where ranges are provided, endpoints are included. Also, unless otherwise clear from the context and understanding of one of ordinary skill in the art, values expressed in ranges are not intended to be used in context with any particular value or subrange within the stated range in other embodiments of the invention. Unless otherwise specified, up to tenths of the unit of the lower limit of the range may be assumed.

All sources cited, such as references, publications, databases, database entries, and techniques cited herein, are incorporated herein by reference, even if not explicitly recited in the citation. In case of any conflict between the content of the cited sources and the present application, the description of the present application shall control.

Section and table headings are not intended to be limiting.

Example

The following are examples of specific embodiments for carrying out the present invention. The following examples are provided for illustrative purposes only and are not intended to limit the scope of the present invention in any way. Efforts have been made to ensure accuracy with respect to numbers used (eg, amounts, temperature, etc.), however, some experimental errors and deviations should, of course, be tolerated.

Unless otherwise specified, in the practice of the present invention, methods conventional in the fields of protein chemistry, biochemistry, recombinant DNA technology and pharmacology within the skill of the art will be used. These techniques are fully described in the literature. For example, T.E. Creighton, Proteins: Structures and Molecular Properties (W.H. Freeman and Company, 1993); A.L. Lehninger, Biochemistry (Worth Publishers, Inc., current addition); Sambrook, et al., Molecular Cloning: A Laboratory Manual (2nd Edition, 1989); Methods In Enzymology (S. Colowick and N. Kaplan eds., Academic Press, Inc.); Remington's Pharmaceutical Sciences, 18th Edition (Easton, Pennsylvania: Mack Publishing Company, 1990); Carey and Sundberg Advanced Organic Chemistry 3rd Ed. (Plenum Press) Vols A and B (1992).

Example 1 - Q-test based FA using dsDNA of different lengths for target and reference

In this example, a transformation (GMO) target sequence is present in a 788 bp target dsDNA (ie, target analyte) and a reference sequence (lectin housekeeping gene) is present in a 466 bp reference dsDNA (ie, reference analyte) We present the results obtained by applying the fraction of distribution (FA) framework to the data. Quantification of the amount of transgene target in a sample is accomplished by first applying the Q test as a single attribute criterion based on event area and using equations (1) and (2), second applying the SVM method and applying equations (3) and ( 4) is done by using

The 466 bp reference DNA and 788 bp target transforming DNA fragments were generated by PCR using sequence specific oligonucleotide primers from a mixture of conventional transgene-containing genomic DNA samples. The PCR product was purified and concentrated using a standard silica membrane column. Precise fraction mixtures of the two amplicons were prepared from the resulting large number of individual amplicons, and aliquots of the fraction mixture and single amplicons were used as standard reference materials for all analyses.

First, a reference control sample containing 466 bp reference DNA was measured in the nanopore device. Next, a target control sample containing 788 bp transforming DNA was prepared and measured in the nanopore device. The difference in length between the target analyte (788 bp) and the reference analyte (466 bp) creates a unique event signal upon translocation through the nanopore that can be discerned based on the area of the event signal.

가 q 값에서 그래픽으로 어떻게 나타나는지 보여준다. q = 5 pA*ms 한계(수직 점선)는 0.05의 위양성(즉, Q _ref = 0.05) 및 0.1의 위음성(즉, Q _targ = 0.9)에 해당한다.4A shows all event area histograms for two separate control runs, one for 466 bp reference DNA and one for 788 bp target transforming DNA. Also shown are area histograms from a 3:10 target:reference control mixed sample. 4B shows the trends of the control mixture ( Q _targ , Q _ref ) and the known mixture ( Q _mix ) as a function of area-based limit q , where Q _mix = Q _3:10 . Figure 4C shows the aliquot parameters.

shows how is represented graphically at the q value. The q = 5 pA*ms limit (vertical dashed line) corresponds to a false positive of 0.05 (ie, Q _ref = 0.05) and a false negative of 0.1 (ie, Q _targ = 0.9).

을 생성하여, 공지 혼합물에 대한 분포 분율의 추정값 생성하기 위해 참조 전용 및 표적 전용 대조군을 이용하는 방법의 정확도 및 정밀도를 테스트한다. 먼저 공지 혼합물에 식 (2)를 적용하였다. Q _X:Y 를 생성하는 데 대조군 혼합 샘플을 사용하지 않았기 때문에, 모델을 검증하지 위해 표적 분석물과 참조 분석물의 포획률 상수 차이에 대한 보상을 사용하지 않고(즉, α = 1로 설정) 추정치가 생성되었다. 도 5A는 예측된 GMO(%)(

) 대비 참 GMO(%)의 플롯을 보여주며, 비교를 위하여 영오차선(기울기 = 1)의 위, 아래에 10% 오차 한계를 도시하였다. 이러한 결과는 단일 나노포어에서 연속적으로, 100% 표적 및 100% 참조(분리된) 대조 샘군을 실행한 다음, 5개의 공지 혼합물을 실생하여 설정되었다. 표 1은 도 5A에 도시된 예측값과 오차 막대, 그리고 각 혼합물에 대하여 검출된 전체 이벤트의 수를 보고한다.Using the control mixture, GMO predicted value (%) by applying Equation (2)

to test the accuracy and precision of the method using reference-only and target-only controls to generate estimates of the fraction of distribution for a known mixture. First, formula (2) was applied to a known mixture. Because no control mixed samples were used to generate Q _X:Y , we did not use compensation for differences in capture rate constants of target and reference analytes to validate the model (i.e., set α = 1) and estimate the estimate has been created 5A shows the predicted GMO (%) (

) versus true GMO (%), and 10% error limits are shown above and below the zero lane (slope = 1) for comparison. These results were established by running a population of 100% target and 100% reference (isolated) control samples sequentially on a single nanopore followed by running five known mixtures. Table 1 reports the predicted values and error bars shown in Figure 5A, and the total number of events detected for each mixture.

Table 1. GMO prediction results for the data in FIG. 5A

Separate nanopore experiments and the resulting results according to a similar protocol (2 separate controls and 6 known mixtures) are presented in Figure 5B and Table 2.

Table 2. GMO prediction results for Figure 5B data

를 도시하고, 평균값

을 공지된 15% GMO와 비교하였다. 이는 본 명세서에서 제공된 분석 프레임워크가 한계 범위에 걸쳐, 상기 한계가 최적화되지 않은 경우에도, 위양성 및 위음성 오차를 보상하여, 샘플 내 표적 분석물의 상대적 존재비의 개선된 추정값을 제공한다.The results in Figures 5A and 5B and Tables 1 and 2 suggest that it is possible to differentiate the two DNAs with a GMO% prediction accuracy within 5% using a single nanopore. These results were obtained without compensating for the difference in capture rate constants of the target analyte and the reference analyte (set α = 1 in Equation (2)). Compensation for differences in capture rate constants is expected to further improve the results. An example of using a q -limit range instead of a single value is shown in FIG. 6 . Specifically, the 75th to 99th percentile of Q _ref was selected as the q- limit range. Results for the above q range

is shown, and the average value

was compared to known 15% GMO. This provides an improved estimate of the relative abundance of a target analyte in a sample, allowing the analytical framework provided herein to compensate for false-positive and false-negative errors over a range of limits, even when those limits are not optimized.

The procedure for quantification of the abundance of the target sequence validated in this example does not require any amplification, purification, concentration or buffer exchange steps. This process is compatible with inexpensive, disposable sample preparation cartridges, allowing sample-in, answer-out procedures in compact (portable or table-top) devices.

을 예측된 GMO%로 보고하였다. 식 (2)에서, 대조군 혼합물을 사용하여 표적 분석물과 참조 분석물의 포획률 상수 차이를 보상하였다. 실험은 1:1, 0.75:1 또는 0.35:1의 표적:참조 대조군 혼합물을 사용하였다.In another series of experiments, various % GMO samples were tested as unknowns. The protocol for each nanopore is as follows: a) 100% 466 bp reference for 5 min followed by flushing; b) 100% 788 bp target for 5 min followed by flushing; c) 1 to 4 blanks each run for 5 minutes, flushing in the middle; d) Run the control mixture. The area criterion was used and the q- limit range spanning the 75th to 99th percentile of Q _ref was used, and the mean

was reported as the predicted GMO%. In equation (2), a control mixture was used to compensate for the difference in capture rate constants of the target analyte and the reference analyte. Experiments used 1:1, 0.75:1 or 0.35:1 target:reference control mixtures.

Table 3 reports the predictive results of one nanopore analysis on four “unknown” mixed samples (S1-S4), using a 0.35:1 control mixture (35% GMO) as compensation. Since unknowns were blinded in each nanopore analysis, percentage errors were not reported in the table. In addition, the table above reports the total number of events recorded during each 5 minutes.

Table 3. GMO prediction results for blinded samples S1-S4

A total of 12 nanopore experiments were performed according to the aforementioned protocol, and each mixed sample was always tested 2-5 times on different nanopores, by different experimenters or on different days. The nanopore size range was 25-35 nm in diameter. A total of 11 mixed samples (S1-S11) were analyzed. Table 4 reports the combined estimates of the predicted % GMO values, from smallest to largest. The reported average GMO% values were calculated by averaging the results of single-nanopore predictions. The uncertainty of each mean estimate was calculated by repeated random sampling (Monte Carlo method) of the distribution of the individual estimates. Reported are the numerically generated 95th percentile confidence intervals. The number of tests for each sample and the % true GMO for each sample were also recorded.

Table 4. Combined GMO% predicted values for samples S1-S11 (mean ± 2 sigma)

The results in Table 4 show that the method of the present invention can predict the distribution fraction (eg, GMO%) of the target analyte with high accuracy. Within the GMO% range of 10-90%, the accuracy is within 2% of combined single-nanopore estimates. Combining the two nanopore estimates resulted in <5% error at 5-10% and 100% GMO, which could be expected to increase the prediction error by approaching the saturation limit. In general, the use of compensation for the difference in capture rate constants between the target analyte and the reference analyte improves accuracy compared to the case where no compensation for the difference in capture rate constants is used (Table 1-2). For the full GMO% prediction range, more nanopore estimates greatly improve accuracy and precision. Arranged nanopores, each measured in a common pool, can also further reduce uncertainty by eliminating variations between person, date, and reagent sets that were present as part of this study.

Example 2 - SVM based FAA using dsDNA of different lengths for target and reference

The same nanopore data as recorded and analyzed in Example 1 were reanalyzed here using the conventional SVM method (Equations (3) to (4)).

A first isolated control set was used for the initial selection selection. The initial selection aims to remove highly correlated features that can cause multicollinearity problems in certain classification methods. The seven identified characteristics are: (i) log ₁₀ (dwell time) or "dwell time", log base 10 of event duration; (ii) maxAmp: δG maximum; (iii) sdAmpSub: standard deviation of the event signal, excluding rise and fall times; (iv) medAmp: δG median; (v) LFNmean: mean value of noise power of events below 50 Hz; (vi) LFNmedian: median noise power of events below 50 Hz; (vii) Area: the same event area used in Example 1.

Additional feature extraction was performed to reduce the data dimension. The purpose of this step is to balance computation time with classification accuracy. Two algorithms were implemented: 1) a univariate feature selection method. ANOVA F values were calculated between each feature and marker of the event. Limits were set manually to select the feature part with the highest F-score. 2) Recursive feature removal (RFE). An estimator (eg, SVM) is trained on an initial set of features and the importance of each feature is obtained. The features with the lowest importance will be excluded from the current feature set. This process is repeated recursively until the desired number of feature sets is reached.

For the data of Example 1, a univariate feature selection method was used. The percentage limit of features was manually set to 60%. The four optimal features selected by the algorithm are: (i) residence time, (ii) sdAmpSub, (iii), medAmp, (iv) area.

The next step in the method is model training and testing. All events collected within separate controls were randomly classified using a 7:3 split into the training data set and the test data set. In order to find optimal parameters for performing classification, an SVM was trained based on the training data set using a hyper-parameter search algorithm. The hyper-parameters tested in the grid algorithm are: kernel type (linear, rbf), normalization parameter (C) and kernel coefficient (gamma). The area under the curve (roc_auc) of the ROC curve was used to evaluate the performance of each hyper-parameter combination. The model with the highest roc_auc score was used for downstream data processing. For the best parameter combination, the mean precision and recall for each class from the test data were calculated. The model with optimal parameters was then trained on the training data set and tested on the test data set. Accuracy predictions for the test data set were generated and shown in FIG. 7 . The accuracy of the whole set exceeded 97.5%.

The next step in the method was data calibration. Calibration can be performed by applying the three-step model to the control mixture data to the control mixture data generating the correction ratio. The correction factor is then multiplied by each predicted quantity for the unknown mixture. This is equivalent to multiplying the parameter α in equations (1) and (2). In the SVM method, values of the parameter α are generated by applying the model to a control mixture, whereas (1) and (2) involve the direct calculation of α from the control data set Q values.

Table 5 shows a comparison of the GMO% prediction of the Q-test and the SVM-based method.

Table 5. Comparison of Single Nanopore GMO% Prediction, Q-Test and SVM

The samples were classified as follows: a) the SVM prediction is more accurate (1, 5, 6, 8, 9, 16, 19, 20, 21), b) the Q-test prediction is more accurate (3, 4) , 7, 10, 11, 12, 14, 15, 17), and c) both methods are equivalent in accuracy (2, 3, 18, 22). For these 22 samples, the overall performance of the two methods was roughly equivalent, and each outperformed the other for 9/22. The values of the SVM method have no clear criteria against which the requirements for the Q-test can be applied. Something that can be automated to apply to datasets that may not exist. On the other hand, the Q-test is simpler to calculate and may be more suitable for application of fractions of a distribution where well-characterized criteria can be utilized in the Q-test format.

Example 3 - Q-test based FA using short DNA with native payload (74 bp reference, 94 bp target transgene)

In the context of the predictive application of % GMO, this example shows that two comparable lengths can be used for target and reference dsDNA, where the distinction of nanopore event signals is two distinct sequence-specific payloads. is achieved using

Methods: 94 bp transgene-specific and 74 bp taxa-specific fragments were synthesized from conventional and transgene-containing genomic DNA samples using a validated qPCR primer set (publicly available from the European Union Reference Laboratory for GM Food and Feed). amplified from the mixture. Prior to nanopore detection, these amplicons were combined with a sequence-specific oligonucleotide probe covalently linked to a PEG polymer probe (International Patent Application No. WO/2016/187159, "Methods and Compositions for Target Detection in a Nanopore Using a Labeled Polymer"). Scaffold", which is incorporated herein by reference in its entirety) (method described in Data Storage patent #5520281-v2-29517, 5/16/2016). Specifically, the transgene-targeting probe was ligated to the 4-arm 40 kDa PEG, and the reference-targeting probe was ligated to the 8-arm 40 kDa PEG.

As a representative example of an overall event scatterplot, FIG. 8 shows event plots for two molecular types run serially as separate controls on the same pore. First, a sample containing 96 bp DNA/probe-payload complex was prepared and measured in a nanopore device. The complex is a model for a fragment comprising the target sequence and binding to the probe-payload. The probe-payload was PNA-PEG with a 4-arm PEG structure. Next, a fragment comprising a reference sequence was designed to generate a unique event signal upon translocation through the nanopore, from which calculation of the fraction of distribution could be achieved. The reference molecule is a 74 bp DNA bound to PNA-PEG, wherein PEG has an 8-arm structure. Target/probe-payload molecules create distinct event subpopulations that are distinct from reference/probe-payload molecules, and it is important to distinguish them from background events when both are present.

을 예측된 GMO%로 보고하였다. 식 (2)에서, 1:1 대조군 혼합물을 표적 분석물과 참조 분석물 사이의 포획률 상수 차이에 대한 보상으로 사용하였다.The protocol for each nanopore is as follows: a) 100% 74 bp/payload-2 reference for 5 min, followed by flushing; b) 100% p4 bp/payload-1 target for 5 min followed by flushing; c) 1-4 unknown runs for each 5 min, flushing between runs; d) Running the control mixture. An area criterion was used and a q- limit range spanning the 75th to 99th percentile of Q _ref was implemented, and the mean

was reported as the predicted GMO%. In equation (2), a 1:1 control mixture was used as compensation for the difference in capture rate constants between the target and reference analytes.

Nanopore set experiments were performed according to the protocol mentioned above, and each mixed sample was tested 2-4 times, always on different nanopores, by different experimenters or on different days. The nanopore size ranged from 25-35 nm in diameter. A total of six mixture samples (Sp1-Sp6) were analyzed. Table 6 reports the combined estimates of the predicted % GMO predicted values, from smallest to largest. The reported average % GMO values are calculated by averaging single-nanopore predictions. The uncertainty of each mean estimate was calculated and recorded as the 95th percentile confidence interval. The number of tests for each sample and the % true GMO for each sample are also recorded.

Table 6. Prediction of % bound GMO using separate payloads for target/reference discrimination

The predictive performance for both payloads does not appear to be as good as when using dsDNA length discrimination (Examples 1 and 2). In either case, the accuracy is greater than 6% in all cases, and can be further improved by combining the resulting estimates with the use of more nanopores measuring the molecular pool in parallel.

Example 4 - Q-test and SVM method for KRAS G12D SNP compared to wild type using short DNA (89 bp) and two unique payloads

Primers were designed to amplify short (58 bp, 70 bp, or 89 bp) fragments of the human KRAS gene from highly fragmented, cell-free, circulating DNA. (cfDNA primer sequences were designed to anneal to either side of the KRAS G12D SNP sequence (CosmicID 521). Amplicons were generated from cell-free circulating DNA fractions obtained from plasma and oligonucleotide probes targeting both wild-type and mutant KRAS alleles were used. Hybridized and covalently linked to a PEG polymer payload: a probe targeting the KRAS wt allele (c.35G) was ligated to a 40 kDa 8-arm or 80 kDa two-branched PEG polymer and a G12D (c.35G->A) allele The gene targeting probe was linked to a 40 kDa 3-branched PEG polymer.

9A shows representative event plots of mean δG versus duration versus overlapped 100% target analyte control samples (closed blue circles) and 100% reference analyte control samples (open squares in black). The target analyte is 89 bp DNA with a G12D-binding probe (hereinafter, G12D-3bPEG) linked to a 3-branched PEG. The reference molecule is an 89 bp DNA with a wild-type (c.35G)-binding probe (hereinafter WT-8armPEG) linked to an 8-arm PEG. Both controls were run continuously at 215 mV using 35 nm diameter nanopores (1.0 M LiCl, 10 mM tris 1 mM EDTA). Visually, the plot suggests criteria based on three inequalities for tagging target events:

msec,

nS 및

nS는 또한 도 9A에 도시된 표적 태그 상자(점선)를 생성한다. 명시된 한계값과 함께 상기 세 부등식의 기준을 사용하여, 분리된 대조군은

과

를 생성한다. 표적-페이로드 및 참조-페이로드 분자의 등몰 농도는

가 되었고, 대조군 혼합물로 사용되었다. 두 개의 후속 미공지 샘플, A 및 B는,

,

를 등록하였다. 두 샘플은 도 9B에 도시된 대로, 이벤트 플롯에서 두 개의 분리된 대조군에서 중첩되었다. 시각적으로, 샘플 A가 샘플 B보다 높은 G12D 함량을 보이지만, 둘 다 100% WT 대조군의 위양성 비율 0.6%에 비해 양성이다. 식 (1)을 적용하고 보상에 대해 대조군 혼합물을 이용한 경우, 야생형에 대한 G12D 돌연변이의 예측 분율은 샘플 A 및 샘플 B에 대해 각각 F _A ^* = 11.1±0.9%와 F _B ^* = 6.0±0.7%이다.limit value

msec,

nS and

nS also generates the target tag box (dashed line) shown in Figure 9A. Using the criteria of the three inequalities above with the specified limits, the isolated controls were

class

to create The equimolar concentrations of the target-payload and reference-payload molecules are

and was used as a control mixture. Two subsequent unknown samples, A and B, were

,

was registered. The two samples were superimposed in two separate controls in the event plot, as shown in Figure 9B. Visually, sample A shows a higher G12D content than sample B, but both are positive compared to the 0.6% false positive rate of the 100% WT control. Applying Equation (1) and using the control mixture for reward, the predicted fractions of G12D mutations for wild-type were F _A ^* = 11.1 ± 0.9% and F _B ^* = 6.0 ± 0.7% for sample A and sample B, respectively. to be.

Table 7 shows the results for samples A and B in column 1 and column 2. Results for all patient samples tested are also shown. A total of 5 different patient samples were analyzed. Samples C and C2 are subsamples of the same patient sample; So are samples D, D2 and E, E2. Different subsamples from the same patient sample were within 2% of each other in all three cases considered. This is despite the fact that different experimenters are running each nanopore experiment on different nanopores, and in two cases on different days. It proposes a reproducible process and quantitative distribution fraction method.

Table 7. Predicted fraction of G12D mutations in blood samples using Q-test

The true G12D amount for these samples is unknown. Samples were collected from patients several weeks after initiation of cancer treatment (chemotherapy), after which each patient's DNA was sequenced and found to be positive for the G12D mutation. Non-positive control samples from control patients were also analyzed and the predicted fraction of G12D was less than 2%, suggesting an overall course false positive rate of 2%. Further optimization of the workflow can further reduce the detection limit.

For comparison, the SVM method was applied. Using one representative experiment (Nanopore NP4 in Table 1), data were processed using the steps described for application of the SVM method. _{Event scatter plots of median δG versus log 10} (duration) for overlapped 100% reference control and 100% target control are shown in FIG. 10 . SVM-identified decision boundaries are also shown. The predicted G12D fraction in sample C2 is reported in Table 8 for both Q-test and SVM methods. Both methods are within 5% of each other.

Table 8. Predicted G12D fraction using Q-test and SVM method to determine the optimized limit (q)

Example 5: EMGM for FA of KRAS G12D SNP compared to wild-type using short DNA (89 bp) and two unique payloads

The application of the expectation maximization algorithm (EMGM) to a Gaussian mixture model on a representative data set is described. Targets and references are the mutant KRASG12D SNP and wild-type sequence in the payload-binding dsDNA fragment, as described in Example 4. As a representative working procedure, only the 1:1 control mixture was measured, only the 100% reference control was measured, and then the unknown mixture was measured.

Step 1: The logarithm (log (retention time)) and median amplitude (medAmp) of the retention times of the 50% target & 50% reference mixed samples were used as input data to the EMGM algorithm ( FIG. 11 ). Using prior knowledge of this analysis, the initially identified predicted region of the target, mutant KRASG12D SNP, was marked as a rectangular region in the plot. This prior knowledge was established by testing a 100% target control under similar conditions (same buffer) in a separate experiment. The box is not used for tags. Rather, after EMGM was applied to the control mixture, all events related to the Gaussian mixture in the box were tagged as target events.

Step 2: Based on the population, the model was trained using a 3 Gaussian mixture model. This model predicted the mutation (target) region within one cluster (diamond). The other two clusters (star and square) correspond to wild-type (Fig. 12). We observed that some events within the initial target domain box (Fig. 11) were related to the reference mode by the EMGM algorithm. This differs from the Q-test, where the box itself defines a population of events that are tagged as reference representatives.

Step 3: Apply the above model to 100% wild-type (reference) samples. The ratio of the number of events in the mutation (target) region to the number of total events establishes the false positive fraction (Figure 13), which can be used to improve the estimate of the fraction of distribution.

Step 4: Predict an unknown mixed sample using the model. The ratio of the number of events in the mutation region to the total number of events was used as a predictor of the percentage of mutant molecules in the unknown mixture (Fig. 14).

As a test of performance improvement by false-positive compensation, the false-positive fraction in step 3 was subtracted from the fraction calculated in step 4 as a correction. The results of applying EMGM to a number of mixtures in the nanopore set experiment are reported in Table 9. Mixtures were blinded until EMGM results were collected, and then the results were compared to true G12D fraction values.

Table 9. Predicted G12D fraction compared to EMGM with and without false positive (FP) reward

For NP-a, performance improved only in 20% of cases using false-positive compensation. For NP-b, the performance was improved in all cases. In the case of NP-c, there was no false-positive compensation, but the performance was already good, especially for the 50% and 20% estimates. In summary, before applying the EMGM model to the unknown mixture for estimating the fraction of distribution, the EMGM method Only the control mixture is needed to apply

Other implementations

It is to be understood that the words used are words of description rather than limitation, and that changes may be made within the scope of the appended claims without departing from the true scope and spirit of the invention in its broader aspects.

While this invention has been described in some detail and to some extent specific in connection with some described embodiments, it is not intended to be limited to that particular or embodiment or any particular embodiment, and in view of the prior art, the possible scope of such claims. In order to provide the broadest interpretation, and thus to effectively encompass the intended scope of the invention, it should be construed with reference to the appended claims.

All publications, patent applications, patents, and other references mentioned herein are incorporated herein by reference in their entirety. In case of conflict, the present specification, including definitions, will control. Also, section headings, materials, methods, and examples are for the purpose of illustration and are not intended to be limiting.

Claims (20)

A method of determining an estimate of the fraction of distribution of a target analyte using one or more nanopore devices, the method comprising:
Applying a voltage to a set of nanopores within one or more nanopore devices to generate a detectable electronic signature and to analyze a portion of a sample comprising units of a target analyte and a reference analyte charged through the set of nanopores. inducing a potential of water;
generating a set of event signatures from the translocation of units of the target analyte and a reference analyte through the set of nanopores;
generating, from the set of event signals, a set of parameters corresponding to the set of nanopores and related to the distribution fraction of the target analyte;
evaluating each of the parameter sets according to corresponding limit conditions to produce a validated parameter set, each of the corresponding limiting conditions being based on a function of a measure of variability determined over values of the parameter set, the validation generating the specified parameter set includes maintaining parameter values that satisfy the corresponding limit condition;
combining the values of the verified parameter sets by a parameter combination operation; and
returning an estimate of the distribution fraction of the target analyte based on an output of the parameter combination operation;
How to include.
The method of claim 1 , wherein the event signal in the set of event signals comprises a current induced and measured by the potential of one of a target analyte and a reference analyte through the nanopores in the set of nanopores.
The method of claim 1 , wherein the measure of variability is at least one of a quartile range, standard deviation, and variance determined over the values of the set of parameters, and wherein a function of the measure of variability includes a constant in the measure of variability. How to include multiplication.
2. The method of claim 1, wherein the corresponding limit condition against which at least one of the parameter sets is evaluated comprises a difference between a maximum value and a minimum value of a parameter in the parameter set.
2. The method of claim 1, wherein combining the values of the verified parameter sets with a parameter combination operation comprises determining a weighted average of values of the verified parameter sets, wherein the weights for the values of the verified parameter sets are: a method determined based on the satisfaction level of the corresponding limit condition for each of the set of verified parameters.
The method of claim 1 , wherein generating the parameter set comprises excluding from consideration data from nanopores in the set of nanopores based on an assessment of the quality of data from the nanopores.
7. The method of claim 6, wherein the quality assessment comprises a first assessment characterizing a morphological feature of each nanopore, a second assessment characterizing a rate of change of a morphological feature of each nanopore, and a signal noise associated with each nanopore. A method comprising at least one of a third evaluation that characterizes.
7. The method of claim 6, wherein the quality assessment comprises a first operation of filtering nanopore data based on noise observed with respect to each nanopore, and a second operation of filtering nanopore data based on separation of a sample population in the sample. A method comprising: an operation; and a third operation of determining a calibration ratio of a population of samples in the sample.
9. The method of claim 8, wherein the first task defines a target area boundary that separates a subset of noise events from a subset of real events within the nanopore data, and, for each set of nanopores, the A method for determining a noise ratio based on a subset and a subset of the actual events.
9. The method of claim 8, wherein the second task implements a principal component analysis (PCA) task on at least one of a dwell time and an amplitude of an electrical signal output, and wherein the first component of the PCA task is used to determine between a sample population of the samples. A method for determining a measure of separation.
The method of claim 1 , wherein the target analyte is associated with a genetically modified organism.
A system for determining an estimate of the fraction of distribution of a target analyte, comprising:
one or more nanopore devices comprising a set of nanopores for displacing material from the sample;
an electrical system comprising a set of electrodes connected to the set of nanopores; and
A system comprising a computing system comprising a non-transitory computer readable memory storing instructions for:
A voltage is applied to the set of nanopores to generate a detectable electronic signature and the potential of a charged analyte through the set of nanopores relative to a portion of the sample comprising units of the target analyte and the reference analyte. inducing;
generating a set of event signatures from the translocation of units of the target analyte and a reference analyte through the set of nanopores;
generating, from the set of event signals, a set of parameters corresponding to the set of nanopores and related to the distribution fraction of the target analyte;
evaluating each of the parameter sets according to corresponding limit conditions to produce a validated parameter set, each of the corresponding limiting conditions being based on a function of a measure of variability determined over values of the parameter set, the validation generating the specified parameter set includes maintaining parameter values that satisfy the corresponding limit condition;
combining the values of the verified parameter sets by a parameter combination operation; and
returning an estimate of the distribution fraction of the target analyte based on an output of the parameter combination operation.
13. The method of claim 12, wherein the measure of variability is at least one of a quartile range, a standard deviation, and a variance determined over values of the set of parameters, and wherein a function of the measure of variability gives a constant to the measure of variability. A system involving multiplication.
13. The system of claim 12, wherein the calculation system comprises a memory storing instructions for determining a difference between a maximum and a minimum value of a parameter in the parameter set, wherein the corresponding limit condition to which at least one of the parameter sets is evaluated is A system that is the difference between the maximum and minimum values of a parameter in the parameter set.
13. The method of claim 12, wherein the calculation system comprises a memory for storing instructions for determining a weighted average of values of the set of verified parameters, wherein the weights for the values of the set of verified parameters are determined for each of the sets of verified parameters. a system determined based on the satisfaction level of the corresponding limit condition for .
13. The system of claim 12, wherein said target analyte is associated with a genetically modified organism.
13. The system of claim 12, wherein the computational system includes a memory for storing instructions for discarding parameter values based on a quality assessment from each nanopore from which the parameter originated.
18. The method of claim 17, wherein the quality assessment characterizes a first assessment characterizing a morphological characteristic of each nanopore, a second assessment characterizing a rate of change of a morphological characteristic of each nanopore, and a signal noise associated with each nanopore. A system comprising at least one of a third rating to build.
18. The method of claim 17, wherein the quality assessment defines a target area boundary that separates a subset of noise events from a subset of real events within the nanopore data, and, for each set of nanopores, a subset of the noise events. and determining a noise ratio based on a set and a subset of the actual events.
18. The method of claim 17, wherein the quality assessment performs a principal component analysis (PCA) task on at least one of a residence time and an amplitude of an electrical signal output, and uses a first component of the PCA task to separate between the sample populations of the samples. A system that determines a measure of

Images