WO2017196112A1

WO2017196112A1 - Detection of abnormal signal in dataset

Info

Publication number: WO2017196112A1
Application number: PCT/KR2017/004902
Authority: WO
Inventors: Jong Yoon Chun; Mi Hyun Jang
Original assignee: Seegene, Inc.
Priority date: 2016-05-11
Filing date: 2017-05-11
Publication date: 2017-11-16
Also published as: KR20180135986A; KR102189356B1

Abstract

The present invention relates to a method for detecting an abnormal signal in a dataset, particularly a dataset for a target analyte. According to the present invention, it is possible to accurately determine a cycle number(s) indicative of an abnormal signal in a dataset using normality scores, but also to provide a corrected dataset in which an abnormal signal value at a cycle number is replaced with a normal signal value, if the candidate cycle number is determined to be indicative of an abnormal signal.

Description

DETECTION OF ABNORMAL SIGNAL IN DATASET

The present invention relates to detection of an abnormal signal in a dataset, particularly a dataset for a target analyte.

For detection of target nucleic acid sequences through nucleic acid amplification, real-time detection methods are widely used to monitor target amplification in a real-time manner. Real-time PCR methods use a signal-generating means for releasing a detectable fluorescent signal in proportion to the amount of target nucleic acid sequence in a PCR reaction, so as to detect a particular target nucleic acid sequence. The release of the detectable fluorescence signal may be achieved, for example, by using an intercalator that emits a fluorescence signal upon bound to a duplex DNA, or an oligonucleotide containing both a reporter molecule and a quencher molecule of inhibiting the release of the fluorescence thereof.

The real-time PCR method measures a fluorescence signal proportional to the amount of target nucleic acid at each cycle, thereby generating a dataset including a plurality of data points, each data point having a pair of coordinate values of a cycle number and a signal intensity (signal value) at the cycle number. The dataset may be represented by an amplification curve (also referred to as an amplification profile curve or growth curve) where fluorescent intensity values are plotted vs. cycle numbers for convenience of data analysis. The dataset representing an amplification curve can then be analyzed to determine the presence or absence of a target nucleic acid sequence in a sample. For example, if there is a cycle having a fluorescent signal more than a threshold applied to the dataset representing an amplification curve, it can be determined that a target nucleic acid sequence is present in a sample.

For detection of a target nucleic acid sequence, it is essential to obtain an accurate and reliable dataset. However, despite elaborate experiments, the resulting datasets may contain abnormal signals (e.g., noises or errors) due to changes in annealing temperatures, the formation of air bubbles in reaction tubes or the presence of contaminant materials in samples. Examples of such abnormal signals include a sharp rise (also referred to as jump, spike and step) or decline (also referred to as dip) in fluorescence signals. The occurrence of such abnormal signals may lead to misinterpretation in qualitative or quantitative analysis of datasets, impairing the accuracy and reliability of the analysis.

Although there have been many attempts to prevent the occurrence of abnormal signals, its exact cause has not yet been clarified. Even if the exact cause is found, it is more difficult to prevent them in advance. Therefore, it would be more practical to analyze the dataset to determine whether abnormal signals have occurred, and if need, correct or invalidate it prior to determining the presence or absence of the target nucleic acid sequence from the dataset.

In this regard, some data analysis methods have been reported for identifying abnormal signals.

U.S. Patent No. 8,560,247 discloses a technique for discriminating non-amplifying data, i.e., errors such as noise and jumps, which comprises receiving a set of data points, calculating a first function that approximates the set of data points, analyzing the first function to determine whether a slope of the first function exceeds a maximum amplification slope and if the maximum amplification slope is exceeded, identifying the exceeded slope segment as a non-amplifying segment of the curve. However, considering that a normal signal often exhibits amplification data exceeding the maximum amplification slope depending on the reaction conditions, the method is highly likely to determine a large number of normal signals as errors.

In addition, U.S. Patent Application Publication No. 2015/0186598 discloses a method for detecting jump errors based on determination of two consecutive cycles with different signs from a second derivative of a dataset. However, the method uses a threshold which is not so strict for determining a jump error, and the application of the threshold is complicated.

Throughout this application, various patents and publications are referenced and citations are provided in parentheses. The disclosure of these patents and publications in their entirety are hereby incorporated by references into this application in order to more fully describe this invention and the state of the art to which this invention pertains.

The present inventors have endeavored to improve conventional methods for detecting abnormal signals which may be found in a dataset for a target analyte. As a result, the present inventors have established a parameter, "normality-representing value", which represents the extent of normality of a signal value at a cycle number for the datasets, and have developed a novel method for detecting abnormal signals by selecting a candidate cycle number(s) by the normality-representing value, altering the signal value at the candidate cycle number(s), and comparing the normality-representing value after alteration to the normality-representing value before alteration at the candidate cycle number.

Accordingly, it is an object of this invention to provide a method for detecting an abnormal signal in a dataset.

It is another object of this invention to provide a computer readable storage medium containing instructions to configure a processor to perform a method for detecting an abnormal signal in a dataset.

It is still another object of this invention to provide a device for detecting an abnormal signal in a dataset.

It is further object of this invention to provide a computer program to be stored on a computer readable storage medium to configure a processor to perform a method for detecting an abnormal signal in a dataset.

Other objects and advantages of the present invention will become apparent from the detailed description to follow taken in conjugation with the appended claims and drawings.

The features and advantages of this invention will be summarized as follows:

(a) The method of the present invention makes it possible not only to accurately determine a cycle number or even a plurality of abnormal signals indicative of an abnormal signal in a dataset by using a normality-representing value. As such, the method of the present invention provides information about correction or invalidation of a dataset by displaying whether there is a cycle number indicative of an abnormal signal in a dataset and which cycle number indicates an abnormal signal.

(b) The method of the present invention can provide a corrected dataset in which an abnormal signal value at a cycle number is replaced with a normal signal value, if the candidate cycle number is determined to be indicative of an abnormal signal. The corrected dataset can be used to provide qualitative and/or quantitative information about the target analyte as a new dataset for a target analyte (in particular, a target nucleic acid sequence).

(c) The method of the present invention can accomplish both the determination of a cycle indicative of an abnormal signal and the correction of the abnormal signal in a simple and time-efficient manner.

(d) Furthermore, the use of the dataset corrected by the method of the present invention can significantly reduce false-positive or false-negative results for the target analyte.

Fig. 1 is a flow chart representing a process for determining a cycle number indicative of an abnormal signal in a dataset in accordance with a representative embodiment of the present invention.

Fig. 2 shows an exemplary dataset (pre-alteration dataset) used in the analysis of Example 1.

Fig. 3 is a graph depicting the normality scores (NS) calculated at each cycle number for the dataset of Fig. 2.

Fig. 4 shows an example of the signal alteration scheme of the present invention for altering the signal value at the candidate cycle number for the dataset of Fig. 2. According to the signal alteration scheme, the signal value at the 10^th cycle number (y_pre(10)) as a candidate cycle number is replaced with the post-alteration signal value (y_post(10)), and then signal values at all cycle numbers after the 10^th cycle number are also replaced with a respective suitable post-alteration signal value.

Fig. 5 shows a pre-alteration dataset (corresponding to the dataset of Fig. 2; dotted line); and a first post-alteration dataset (solid line) in which a signal value at a candidate cycle number (10^th cycle number) and signal values at all cycle numbers after the candidate cycle number (from 11^th cycle number to 45^th cycle number) in the pre-alteration dataset are replaced with a respective suitable post-alteration signal value in accordance with an embodiment of the present invention.

Fig. 6 is a graph depicting the normality scores calculated at each cycle number for the first post-alteration dataset (indicated by solid line in Fig. 5).

Fig. 7 shows the first post-alteration dataset (dotted line); and a second post-alteration dataset (solid line) in which a signal value at another candidate cycle number (11^th cycle number) and signal values at all cycle numbers after the candidate cycle number (from 12^th cycle number to 45^th cycle number) in the first post-alteration dataset are replaced with a respective suitable post-alteration signal value in accordance with an embodiment of the present invention.

Fig. 8 is a graph depicting the normality scores calculated at each cycle number for the second post-alteration dataset (indicated by solid line in Fig. 7).

Fig. 9 shows the second post-alteration dataset (dotted line); and a third post-alteration dataset (solid line) in which a signal value at another candidate cycle number (23^th cycle number) and signal values at all cycle numbers after the candidate cycle number (from 24^th cycle number to 45^th cycle number) in the second post-alteration dataset are replaced with a respective suitable post-alteration signal value in accordance with an embodiment of the present invention.

Fig. 10 is a graph showing the normality scores calculated at each cycle number for the third post-alteration dataset (indicated by solid line in Fig. 9).

Fig. 11 shows the pre-alteration dataset of Fig. 2 (dotted line); and a corrected dataset finally provided in accordance with an embodiment of the present invention (second post-alteration dataset; solid line).

I. Detection of Abnormal Signal in Dataset

In one aspect of this invention, there is provided a method for detecting an abnormal signal in a dataset, comprising:

(a) obtaining a dataset for a target analyte by a signal amplification reaction; wherein the dataset comprises a plurality of data points; wherein each of the data points has a cycle number and a signal value at the cycle number;

(b) providing a normality-representing value at each cycle number of the dataset by using the signal values; wherein the normality-representing value is a value representing the extent of normality of a signal value at a cycle number;

(c) selecting a candidate cycle number(s) for an abnormal signal by the normality-representing value; wherein the normality-representing value at the candidate cycle number(s) corresponds to a pre-alteration normality-representing value;

(d) altering the signal value at the candidate cycle number(s), and further providing a normality-representing value at the candidate cycle number(s) by using the altered signal value; wherein the normality-representing value further provided at the candidate cycle number(s) corresponds to a post-alteration normality-representing value;

(e) comparing the post-alteration normality-representing value to the pre-alteration normality-representing value at the candidate cycle number(s); and

(f) determining the candidate cycle number(s) to be indicative of an abnormal signal, if the post-alteration normality-representing value indicates a higher extent of normality than the pre-alteration normality-representing value at the candidate cycle number.

As used herein, the term "abnormal signal" refers to a signal which is not associated with a target analyte (e.g., target nucleic acid sequence), i.e., a signal which is abruptly increased or decreased by other factors than an analyte during a signal amplification reaction. The term "abnormal signal" is used interchangeably with "error signal", "erroneous signal", "aberrant signal", "outlier signal" and "noise signal". The abnormal signal herein includes a signal indicating a sharp rise (e.g., jump, spike or step) or decline (e.g., dip) of the signal values in the amplification curve obtained from the signal amplification reaction. The causes of the abnormal signal include, but are not limited to, changes in annealing temperatures, the formation of air bubbles in reaction tubes or the presence of contaminant materials in samples.

The present method will be described in more detail with reference to Fig. 1 as follows:

Step (a): Obtaining a dataset 110

In step (a), a dataset for a target analyte is obtained by a signal amplification reaction 110. The dataset comprises a plurality of data points, each of the data points having a cycle number and a signal value at the cycle number.

The term "target analyte" or "analyte" as used herein encompasses a variety of materials (e.g., biological and non-biological materials), particularly biological materials, more particularly nucleic acid molecules (e.g., DNA and RNA), carbohydrates, lipids, amino acids, biological compounds, hormones, antibodies, antigens, metabolites and cells. Most particularly, the target analyte is a target nucleic acid molecule. The target analyte is present in a sample.

The term "sample" as used herein refers to any material undergoing the method of the present invention. Particularly, the term "sample" refers to any material containing or presumed to contain a nucleic acid of interest or which is itself a nucleic acid containing or presumed to contain a target nucleic acid sequence of interest. More particularly, the term "sample" as used herein includes biological samples (e.g., cells, tissues, and fluid from a biological source) and non-biological samples (e.g., food, water and soil). The biological samples includes, but not limited to, virus, bacteria, tissue, cell, blood, serum, plasma, lymph, sputum, swab, aspirate, bronchoalveolar lavage fluid, milk, urine, feces, ocular fluid, saliva, semen, brain extracts, spinal cord fluid (SCF), appendix, spleen and tonsillar tissue extracts, amniotic fluid and ascitic fluid. In addition, the sample may include naturally occurring nucleic acid molecules isolated from biological sources and synthetic nucleic acid molecules.

The term used herein "target nucleic acid", "target nucleic acid sequence" or "target sequence" refers to a nucleic acid sequence of interest for analysis, detection or quantification. The target nucleic acid sequence comprises a sequence in a single strand as well as in a double strand. The target nucleic acid sequence comprises a sequence newly generated in reactions as well as a sequence initially present in a sample.

The target nucleic acid sequence may include any DNA (gDNA and cDNA), RNA molecules and their hybrids (chimera nucleic acid). The sequence may be in either a double-stranded or single-stranded form. Where the nucleic acid as starting material is double-stranded, it is preferred to render the two strands into a single-stranded or partially single-stranded form. Methods known to separate strands includes, but not limited to, heating, alkali, formamide, urea and glycoxal treatment, enzymatic methods (e.g., helicase action), and binding proteins. For instance, strand separation can be achieved by heating at temperature ranging from 80℃ to 105℃. General methods for accomplishing this treatment are provided by Joseph Sambrook, et al., Molecular Cloning, A Laboratory Manual, Cold Spring Harbor Laboratory Press, Cold Spring Harbor, N.Y.(2001).

The target nucleic acid sequence includes any naturally occurring prokaryotic, eukaryotic (for example, protozoans and parasites, fungi, yeast, higher plants, lower and higher animals, including mammals and humans), viral (for example, Herpes viruses, HIV, influenza virus, Epstein-Barr virus, hepatitis virus, polio virus, etc.), or viroid nucleic acid. The nucleic acid molecule can also be any nucleic acid molecule which has been or can be recombinantly produced or chemically synthesized. Thus, the nucleic acid sequence may or may not be found in nature.

The target nucleic acid sequence should not be construed as limiting the sequence known at a given time or the sequence available as of a given time, but instead should be read to encompass the sequence that may be available or known now or at any time in the future. In other words, the target nucleic acid sequence may or may not be known at the time of practicing the present method. In case of unknown target nucleic acid, its sequence may be determined by one of conventional sequencing methods prior to performing the present method.

When the target analyte is a target nucleic acid molecule, the sample may undergo a nucleic acid extraction procedure known in the art (see Sambrook, J. et al., Molecular Cloning, A Laboratory Manual, 3rd ed. Cold Spring Harbor Press (2001)). The nucleic acid extraction process may vary depending on the type of the sample. In addition, when the extracted nucleic acid is RNA, a reverse transcription process for synthesizing cDNA can be further performed (see Sambrook, J. et al., Molecular Cloning, A Laboratory Manual, 3rd ed., Cold Spring Harbor Press (2001)).

According to an embodiment of this invention, the target nucleic acid sequence comprises a nucleotide variation.

The term "nucleotide variation" used herein refers to any single or multiple nucleotide substitutions, deletions or insertions in a DNA sequence at a particular location among contiguous DNA segments that are otherwise similar in sequence. Such contiguous DNA segments include a gene or any other portion of a chromosome. These nucleotide variations may be mutant or polymorphic allele variations. For example, the nucleotide variation detected in the present invention includes SNP (single nucleotide polymorphism), mutation, deletion, insertion, substitution and translocation. Exemplified nucleotide variation includes numerous variations in a human genome (e.g., variations in the MTHFR (methylenetetrahydrofolate reductase) gene), variations involved in drug resistance of pathogens and tumorigenesis-causing variations. The term "nucleotide variation" used herein includes any variation at a particular location in a nucleic acid sequence. In other words, the term "nucleotide variation" includes a wild type and its any mutant type at a particular location in a nucleic acid sequence.

The dataset used herein is obtained by a signal-generating process.

The term "signal-generating process" as used herein refers to any process capable of generating signals in a dependent manner on the properties of a target analyte in a sample, i.e., activity, amount or presence (or absence), particularly presence (or absence). The signal-generating process herein includes biological and chemical reactions. Such biological reactions include genetic analysis such as PCR, real-time PCR and microarray, immunological analysis and bacterial growth assays. According to an embodiment, the signal-generating process comprises analyzing the generation, change or destruction of the chemical substance.

The signal-generating process is accompanied with signal change. The signal change may serve as an indicator indicating qualitatively or quantitatively the presence or absence of a target nucleic acid sequence.

The details of "signal-generating process" are disclosed in WO 2015/147412 filed by the present inventors, the teachings of which are incorporated herein by reference in its entirety.

According to an embodiment, the signal-generating process is a signal amplification process.

According to an embodiment of this invention, the signal-generating process is a process with amplification or with no amplification of a target nucleic acid sequence.

Particularly, the signal-generating process is a process with amplification of a target nucleic acid molecule. More particularly, the signal-generating process is a process with amplification of a target nucleic acid molecule and capable of increasing or decreasing signals (particularly, increasing signals) upon amplifying the target nucleic acid molecule.

The term used herein "signal generation" include appearance or disappearance of signals and increase or decrease in signals. Particularly, the term "signal generation" means increase in signals.

The signal-generating process may be performed in accordance with a multitude of methods known to one of skill in the art. The methods include TaqMan^TM probe method (U.S. Pat. No. 5,210,015), Molecular Beacon method (Tyagi et al., Nature Biotechnology, 14 (3):303(1996)), Scorpion method (Whitcombe et al., Nature Biotechnology 17:804-807(1999)), Sunrise or Amplifluor method (Nazarenko et al., Nucleic Acids Research, 25(12):2516-2521(1997), and U.S. Pat. No. 6,117,635), Lux method (U.S. Pat. No. 7,537,886), CPT (Duck P, et al., Biotechniques, 9:142-148(1990)), LNA method (U.S. Pat. No. 6,977,295), Plexor method (Sherrill CB, et al., Journal of the American Chemical Society, 126:4550-4556(2004)), Hybeacons^TM (D. J. French, et al., Molecular and Cellular Probes (2001) 13, 363-374 and U.S. Pat. No. 7,348,141), Dual-labeled, self-quenched probe (U.S. Pat. No. 5,876,930), Hybridization probe (Bernard PS, et al., Clin Chem 2000, 46, 147-148), PTOCE (PTO cleavage and extension) method (WO 2012/096523), PCE-SH (PTO Cleavage and Extension-Dependent Signaling Oligonucleotide Hybridization) method (WO 2013/115442) and PCE-NH (PTO Cleavage and Extension-Dependent Non-Hybridization) method (WO 2014/104818) and CER method (WO 2011/037306).

When the signal-generating process is performed in accordance with TaqMan^TM probe method, the signal-generation means may comprise a primer pair, a probe with an interactive dual label and DNA polymerase having 5' to 3' nuclease activity. When the signal-generating process is performed in accordance with PTOCE method, the signal-generation means may comprise a primer pair, PTO (Probing and Tagging Oligonucleotide), CTO (Capturing and Templating Oligonucleotide) and DNA polymerase having 5' to 3' nuclease activity. Either PTO or CTO may be labeled with suitable labels.

According to an embodiment, the signal-generating process is performed in a process involving signal amplification together with target amplification.

According to an embodiment, the signal amplification reaction as the signal-generating process is performed in such a manner that signals are amplified simultaneously with amplification of the target nucleic acid sequence (e.g., real-time PCR). Alternatively, the signal amplification reaction is performed in such a manner that signals are amplified with no amplification of the target nucleic acid molecule [e.g., CPT method (Duck P, et al., Biotechniques, 9:142-148 (1990)), Invader assay (U.S. Pat. Nos. 6,358,691 and 6,194,149)].

A multitude of methods have been known for amplification of a target nucleic acid molecule, including, but not limited to, PCR (polymerase chain reaction), LCR (ligase chain reaction, see Wiedmann M, et al., "Ligase chain reaction (LCR)- overview and applications." PCR Methods and Applications 1994 Feb; 3(4):S51-64), GLCR (gap filling LCR, see WO 90/01069, EP 439182 and WO 93/00447), Q-beta (Q-beta replicase amplification, see Cahill P, et al., Clin Chem., 37(9):1482-5(1991), U.S. Pat. No. 5556751), SDA (strand displacement amplification, see G T Walker et al., Nucleic Acids Res. 20(7):16911696(1992), EP 497272), NASBA (nucleic acid sequence-based amplification, see Compton, J. Nature 350(6313):912(1991)), TMA (Transcription-Mediated Amplification, see Hofmann WP et al., J Clin Virol. 32(4):289-93(2005); U.S. Pat. No. 5888779) or RCA (Rolling Circle Amplification, see Hutchison C.A. et al., Proc. Natl Acad. Sci. USA. 102:1733217336(2005)).

The term "signal" as used herein refers to a measurable output. The term "signal value" as used herein is an expression that quantitatively represents a signal.

The magnitude, change, etc. of the signal may serve as an indicator indicating qualitatively or quantitatively the properties, particularly the presence or absence of a target analyte (a target nucleic acid sequence).

Examples of useful indicators include fluorescence intensity, luminescence intensity, chemiluminescence intensity, bioluminescence intensity, phosphorescence intensity, charge transfer, voltage, current, power, energy, temperature, viscosity, light scatter, radioactive intensity, reflectivity, transmittance and absorbance. The most widely used indicator is fluorescence intensity. The signal change includes generation or extinction of the signal as well as increase or decrease of the signal.

Signals include various signal characteristics from the signal detection, e.g., signal intensity [e.g., RFU (relative fluorescence unit) value or in the case of performing amplification, RFU values at a certain cycle number, at selected cycle numbers or at end-point], signal change shape (or pattern) or C_t value, or values obtained by mathematically processing the characteristics.

According to an embodiment, the term "signal" includes not only signals per se obtained at detection temperatures but also a modified signal provided by mathematically processing the signals.

According to an embodiment of this invention, when an amplification curve is obtained by real-time PCR, various signal values (or characteristics) from the amplification curve may be selected and used for determination of target presence (intensity, C_t value or amplification curve data).

The signal (particularly, the signal intensity) may vary depending upon its detection temperature as well as a signal-generating means employed.

The term "signal-generating means" as used herein refers to a means for providing a signal indicative of a property, specifically the presence or absence of a target analyte to be analyzed.

The term "signal-generating means" as used herein refers to any material used in generation of signals indicating the presence of target nucleic acid sequences, for example including oligonucleotides, labels and enzymes. Alternatively, the term used herein "signal-generating means" can be used to refer to any methods using the materials for signal generation.

A wide variety of the signal-generating means have been known to one of skill in the art. The signal-generating means include both labels per se and oligonucleotides with labels. The labels may include a fluorescent label, a luminescent label, a chemiluminescent label, an electrochemical label and a metal label. The label per se may serve as signal-generating means, for example, an intercalating dye. Alternatively, a single label or an interactive dual label containing a donor molecule and an acceptor molecule may be used as signal-generating means in the form of linkage to at least one oligonucleotide.

The signal-generating means may comprise additional components for generating signals such as nucleolytic enzymes (e.g., 5'-nucleases and 3'-nucleases).

The signal-generating means may comprises generating a signal in a dependent manner on the formation of a duplex; generating a signal using the formation of a duplex in a dependent manner on cleavage of a mediation oligonucleotide specifically hybridized to the target analyte; and generating a signal by cleavage of a detection oligonucleotide.

The term "signal amplification reaction" as used herein refers to a reaction that increases or decreases the signal generated by the signal-generating means.

According to an embodiment, the signal amplification reaction means a reaction that increases (amplifies) the signal generated by the signal-generating means depending upon the presence of a target analyte. This signal amplification reaction may or may not be accompanied with amplification of a target analyte (e.g., target nucleic acid molecule). Particularly, the signal amplification reaction means an amplification of the signal accompanied by amplification of a target analyte.

A dataset obtained by a signal amplification reaction includes cycle numbers.

The term "cycle number" or "cycle" as used herein refers to a unit of changes of conditions in a plurality of measurements accompanied with changes of conditions. For example, the changes of conditions include changes in temperature, reaction time, reaction number, concentration, pH and/or replication number of a target nucleic acid molecule sequence. Therefore, the cycle may include temperature, time or process cycle, unit operation cycle and reproductive cycle.

As one example, when a substrate decomposition capacity by an enzyme is analyzed depending on concentrations of the substrate, a plurality of measurements　for the decomposition capacity by the enzyme is carried out with varying substrate concentrations. The increases in the substrate concentration may correspond to the changes of conditions and a unit of the increases may correspond to a cycle.

As another example, an isothermal amplification allows for a plurality of measurements for a sample in the course of reaction time under isothermal conditions and the reaction time may correspond to the changes of conditions and a unit of the reaction time may correspond to a cycle. For example, where a 5-minute interval, such as 5, 10, 15 min and the like, is set as one reaction time, the cycle can be expressed by 5-minute cycle, 10-minute cycle, 15-minute cycle, and the like. Alternatively, regarding a 5-minute interval as one unit, a 5-minute cycle may be represented by 1 cycle, a 10-minute cycle by 2 cycle, a 15-minute cycle by 3 cycle, and the like.

As still another example, in the case of the melting analysis or the hybridization analysis, the signal change may be measured as the temperature changes within a certain range of temperature, the temperature may correspond to the changes of conditions, and a unit of the temperature (e.g., measurement temperature) may correspond to a cycle. For example, where a 0.5℃ interval, such as 40℃, 40.5℃, 50℃, 50.5℃ and the like, is set as one reaction time, the cycle can be expressed by 40℃ cycle, 40.5℃ cycle, 50℃ cycle, 50.5℃ cycle and the like. Alternatively, regarding a 0.5℃ interval as one unit, a 40℃ cycle may be represented by 1 cycle, a 40.5℃ cycle number by 2 cycle, and a 50℃ by 3 cycle, and the like.

More particularly, when repeating a series of reactions or repeating a reaction with a time interval, the term "cycle" or "cycle number" refers to a unit of the repetition.

For example, in a polymerase chain reaction (PCR), a cycle or cycle number refers to a reaction unit comprising denaturation of a target molecule, annealing (hybridization) between the target molecule and primers and primer extension. The increases in the repetition of reactions may correspond to the changes of conditions and a unit of the repetition may correspond to a cycle or cycle number.

The dataset obtained by a signal amplification reaction include a plurality of data points, each data point having a cycle number and a signal value at the cycle number.

The term used herein "signal value" means either signal value actually measured at each cycle number of the signal-generating process (e.g., actual value of fluorescence intensity processed by signal amplification reaction) or its modification. The modification may include mathematically processed value of measured signal value (e.g., intensities). Examples of mathematically processed value of actually measured signal value (i.e., signal value of a raw dataset) may include, but are not limited to, a value obtained by adding a selected constant to the measured signal value, by subtracting a selected constant from the measured signal value, by multiplying the measured signal value by a selected constant, or by dividing the measured signal value by a selected constant; a logarithmic value of the measured signal value; or a derivative of the measured signal value. The term used herein "signal" is intended to encompass the term "signal value" and therefore these terms will be used interchangeably.

The signal value as used herein refers to a value obtained by absolutely or relatively quantifying the magnitude of a signal initially detected at the cycle number in the detector. The signal value is also referred to as a "zero-order signal value", a "raw signal value", or an "original signal value" in order to distinguish it from the first-order change value or the second-order change value. The unit of the signal value may vary depending on the type of signal generation reaction used. For example, when a signal value is obtained at each cycle number by a real-time PCR amplification reaction, the signal value may be represented by RFU (Relative Fluorescence Unit).

The term "data point" as used herein means a coordinate value comprising a cycle number and a signal value at the cycle number. The term "data" means all information that constitutes a dataset. For example, each of the cycle numbers and the signal values is a data.

Data points obtained by the signal-generating process, in particular the signal amplification reaction, can be represented as coordinate values in a two-dimensional rectangular coordinate system. In the coordinate values, the X-axis represents the cycle number, and the Y-axis represents the signal value measured or processed at the cycle number.

The term "dataset" as used herein refers to a set of data points. For example, the dataset may be a set of data points directly obtained by a signal amplification reaction performed in the presence of the signal-generating means, or it may be a set of data points modified from the original data points. The dataset may be all or part of a plurality of data points obtained by a signal amplification reaction or modified data points thereof. In addition, the dataset may be a set of data points including a cycle number and an n^th-order change value at the cycle number.

The dataset may be plotted, giving an amplification curve.

As used herein, the term "amplification curve" refers to a curve obtained by a signal amplification reaction. The amplification curve includes a curve obtained in the presence of an analyte in a sample, or a curve (or line) obtained in the absence of an analyte in a sample. According to one embodiment, the dataset used in the present invention is a raw dataset that has not undergone mathematical processing. According to another embodiment, the dataset used in the present invention is a mathematically processed dataset, for example a baseline-subtracted dataset, to remove background signals in a raw dataset. The baseline-subtracted dataset can be obtained by a variety of methods known in the art (e.g., U.S. Patent No. 8,560,247).

According to one embodiment, the method of the present invention further comprises performing a signal-generating process (e.g., signal amplification reaction) to obtain datasets prior to the step (a).

The dataset used herein includes a dataset obtained by a signal-generating means at a detection temperature. The dataset used herein may be any one of datasets obtained by detection at different detection temperatures in a signal amplification reaction, or any one of datasets obtained by detection using different signal detection means.

The datasets obtained by detection at different detection temperatures refers to dataset obtained, for example, by detecting changes in signal values at different detection temperatures (e.g., at least two detection temperatures at each cycle number) during a signal amplification reaction using a single signal-generating means in a single reaction vessel. For example, two or more datasets can be obtained from the signal amplification reaction by detection at different detection temperatures, according to the MuDT1 technology (WO 2015/147412) or the MuDT2 technology (WO 2016/093619) developed by the present inventor, and the dataset used herein may be one of the datasets above.

The datasets obtained by using different signal detection means refers to dataset obtained, for example, by detecting changes in signal values using different detecting means (e.g., optical modules) in a signal amplification reaction. For example, in a multiplex real-time PCR using two or more signal-generating means (e.g., fluorescent labels) for detection of two or more target nucleic acid sequences, datasets may be obtained by detecting signals from different signal-generating means using appropriate channels containing different optical modules, and the dataset used herein may be one of the datasets above.

Step (b): Providing a normality-representing value at each cycle number 120

Afterwards, a normality-representing value is provided at each cycle number of the dataset by using the signal values.

In this step, a "normality-representing value dataset" including a plurality of data points having a cycle number and a normality-representing value at a cycle number is obtained.

The normality-representing value refers to a value representing the extent (degree) of normality of a signal value at a cycle number. The normality-representing value is calculated for each cycle number and allocated to each cycle number.

The normality-representing value may be expressed in a variety of ways, as long as it represents a normality of a signal.

In an embodiment, the normality-representing value is expressed as a numerical value, in which a larger normality-representing value indicates a higher extent (degree) of normality. In this case, a small normality-representing value may indicate that the cycle number having the normality-representing value is highly likely to exhibit an abnormal signal; whereas a large normality-representing value may indicate that the cycle number having the normality-representing value is highly likely to exhibit a normal signal. For example, a particular cycle number with a normality-representing value of 1000 is likely to represent a normal signal, while another cycle number with a normality-representing value of -2000 is likely to represent an abnormal signal. A typical example of the normality-representing value as expressed above includes a normality score as will be explained below.

In an alternative embodiment, the normality-representing value is expressed as a numerical value, in which a smaller normality-representing value indicates a higher extent (degree) of normality. In this case, a large normality-representing value may indicate that the cycle number having the normality-representing value is highly likely to exhibit an abnormal signal; whereas a small normality-representing value may indicate that the cycle number having the normality-representing value is highly likely to exhibit a normal signal. For example, a particular cycle number with a normality-representing value of 1000 is likely to represent an abnormal signal, while another cycle number with a normality-representing value of -2000 is likely to represent a normal signal.

The normality-representing value at each cycle number of the dataset may be provided by various methods.

In an embodiment, the normality-representing value at said each cycle number is provided by calculating a change value at said each cycle number from signal values at 2-5 consecutive cycle numbers comprising said each cycle number, particularly 2 consecutive cycle numbers comprising said each cycle number. For example, the normality-representing value at the 10^th cycle number is provided by calculating a change value at the 10^th cycle number from signal values at 2-5 consecutive cycle numbers comprising the 10^th cycle number.

The term "change", "change value", or "value of change" as used herein in connection with the signal value at each cycle number means the quantity (or degree) of change of the signal value at a particular cycle number. Since the "change", "change value", or "value of change" is calculated from a certain cycle number, it can be also expressed as "change value of a signal value at a cycle number", or for brevity, as "change value at a cycle number" or modification thereof. The "n^th-order change value" as used herein means the quantity (or degree) of change in (n-1)^th-order change value at a particular cycle number. Since the n^th change value is calculated from a certain cycle number, it can be also expressed as "n^th change value of a signal value at a cycle number", or for brevity, as "n^th change value at a cycle number" or modification thereof.

Specific examples of the change value include a first-order change value (e.g., a first-order difference, a first-order difference quotient, and a first-order derivative), a second-order change value (e.g., a second-order difference, a second-order difference quotient, and a second-order derivative), a third-order change value (e.g., a third-order difference, a third-order difference quotient, and a third-order derivative), and the like.

Thus, the normality-representing value at said each cycle number may be provided by calculating a first-order change value, a second-order change value, a third-order change value, or the like, of signal values at 2-5 consecutive cycle numbers comprising said cycle number.

In an embodiment, the normality-representing value at said each cycle number may be provided by modifying the calculated change value of signal values at 2-5 consecutive cycle numbers comprising said cycle number. The modification may include any mathematical modification, calculation, processing or operation. A mathematical modification, calculation, processing or operation may include addition, subtraction, multiplication or division between change values at different cycle numbers. A mathematical modification, calculation, processing or operation may include addition, subtraction, multiplication or division of a change value with other factors (values).

The number of signal values to be used for calculating a change value includes, without limitation, 2, 3, 4, or 5.

Where the number of signal values is 2, 3, 4, or 5, the normality-representing value at said each cycle number may be provided by calculating a change value using signal values at two, three, four, or five consecutive cycle numbers comprising said cycle number, respectively.

A representative example of the normality-representing value is a normality score.

Although the present specification describes a normality score for a clearer understanding of the normality-representing value, it will be appreciated by those skilled in the art that other normality-representing values may be applied instead of the normality-score.

The normality score as an example of the normality-representing value will be described in detail below.

Normality Score

A normality score is obtained by (i) calculating a second-order change value at each cycle number of the dataset; and (ii) calculating a normality score at each cycle number of the dataset by using the second-order change value; wherein the calculation of the normality score is performed by a mathematical operation that represents sign change between and magnitudes of the second-order change values at two consecutive cycle numbers.

(i) Calculating a second-order change value at each cycle number

First, a second-order change value is calculated at each cycle number of the dataset obtained in the step (a). The calculation of the second-order change value is performed by calculating a first-order change value at each cycle number of the dataset by using two signal values at two consecutive cycle numbers, and then calculating a second-order change value at each cycle number of the dataset by using the two first-order change values at two consecutive cycle numbers.

In this step, a "first-order change value dataset" (including a plurality of data points having a cycle number and a first-order change value at the cycle number) and a "second-order change value dataset" (including a plurality of data points having a cycle number and a second-order change value at the cycle number) are obtained from a raw dataset (including a plurality of data points having a cycle number and a signal value at the cycle number) or a modified dataset thereof.

The term "signal value" used for calculation of a first-order change value and a second-order change value is a signal value, except for values representing the relationship with other signal values (e.g., a change value, such as a first-order change value, a second-order change value, and the like), and is also referred to as a "zero-order signal value", "raw signal value", or "original signal value".

The term "immediately adjacent cycle number" as used herein in connection with any particular cycle number refers to a cycle number that is contiguous to the particular cycle number, i.e., a cycle number that immediately precedes or immediately follows the particular cycle number. For example, in a typical dataset in which the cycle number increases by 1, the cycle number immediately adjacent to the 4^th cycle number is either the 3^rd cycle number or the 5^th cycle number.

In addition, the term "two immediately adjacent cycle numbers", "immediately adjacent two cycle numbers", or "consecutive cycle numbers" as used herein means two cycle numbers immediately adjacent to each other, i.e., two consecutive cycle numbers. For example, in a typical dataset in which the cycle number increases by 1, two immediately adjacent cycle numbers means cycle numbers x and x+1 or cycle numbers x-1 and x.

The term "change", "change value", or "value of change" as used herein in connection with the signal value at each cycle number means the quantity (or degree) of change of the signal value at a particular cycle number. Since the "change", "change value", or "value of change" is calculated from a certain cycle number, it can be also expressed as "change value of a signal value at a cycle number", or for brevity, as "change value at a cycle number" or modification thereof. The "n^th-order change value" as used herein means the quantity (or degree) of change in (n-1)^th-order change value at a particular cycle number. Since the n^th change value is calculated from a certain cycle number, it can be also expressed as "n^th change value of a signal value at a cycle number", or for brevity, as "n^th change value at a cycle number" or modification thereof. In particular, the zero-order change means a raw signal value. According to the above definition, the first-order change value means the quantity (or degree) of change of a zero-order change value (i.e., signal value) at a certain cycle number, and the second-order change value means the quantity (or degree) of change of a first-order change value (i.e., the quantity (or degree) of change of a signal value) at a certain cycle number. As will be described below, it will be appreciated that a first-order change value and a second-order change value at a specific cycle number encompass, in a broad sense, those calculated at other cycle number other than the specific cycle number. The term "change value" may be used interchangeably with the term "rate of change".

In this step, a second-order change value at each cycle number is obtained using a signal value at each cycle number. Specifically, a second-order change value is obtained by calculating a first-order change value at each cycle number of the dataset by using two signal values at two consecutive (immediately adjacent) cycle numbers, and then calculating a second-order change value at each cycle number of the dataset by using the two first-order change values at two consecutive (immediately adjacent) cycle numbers.

The calculation for a second-order change value may be performed in a same or different manner as that for a first-order change value.

The change value may be any one selected from the group consisting of a difference, a difference quotient and a derivative, and thus the n^th-order change value (e.g., a first-order change value and a second-order change value) may be an n^th-order difference, an n^th-order difference quotient, or an n^th-order derivative.

The change value, in particular, a difference (difference value), a difference quotient (difference quotient value) and a derivative (derivative value), may be calculated or obtained by a number of methods known in the art.

For example, the difference as used herein may be obtained by calculating a difference between signal values at two immediately adjacent cycle numbers. The difference quotient as used herein may be obtained by dividing the difference by an interval between two immediately adjacent cycle numbers. The derivative as used herein may be obtained by subjecting signal values at 2, 3, 4, or more data points to a least squares method, or by determining the tangent line (slope) at each cycle number in an amplification curve.

The symbols y(x), D(x), D'(x) and D"(x) used to calculate a second-order change value will have the following meanings:

The symbol y(x) means a signal value at the x^th cycle number; D'(x) means a first-order change value at the x^th cycle number; D"(x) means a second-order change value at the x^th cycle number. In the above definition, x is an integer of 1 or more, which denotes a particular cycle number in a dataset. For example, when a dataset obtained consists of 45 cycle numbers, each of the cycle number may be distinguished by use of the designations, such as 1^st cycle number (cycle number 1), 2^nd cycle number (cycle number 2)... 45^th cycle number (cycle number 45).

The second-order change value as used herein may be any one selected from the group consisting of a second-order difference, a second-order difference quotient, and a second-order derivative.

The second-order difference may be obtained by calculating a first-order difference at each cycle number (including a plurality of data points having a cycle number and a first-order difference at the cycle number) by using two signal values at two immediately adjacent cycle numbers, and then calculating a second-order difference at each cycle number (including a plurality of data points having a cycle number and a second-order difference at the cycle number) by using two first-order differences at two immediately adjacent cycle numbers.

The differences as used herein may be calculated in various ways known in the art.

In an embodiment, the second-order difference (or a second-order difference quotient) is calculated by a forward difference method or a backward difference method known in the art.

For example, the first-order difference may be calculated from two signal values by either a forward difference method or a backward difference method, and then the second-order difference may be calculated from two first-order differences by either a forward difference method or a backward difference method. A forward difference method may be applied to the calculation of both the first-order difference and the second-order difference; a backward difference method may be applied to the calculation of both the first-order difference and the second-order difference; a forward difference method may be applied to the calculation of the first-order difference and a backward difference method may be applied to the calculation of the second-order difference; or a backward difference method may be applied to the calculation of the first-order difference and a forward difference method may be applied to the calculation of the second-order difference.

According to the forward difference method, the second-order difference may be calculated by using the following Equations I and II sequentially or by using the following Equation III alone:

Equation I

D'(x) = y(x+1) - y(x)

Equation II

D"(x) = D'(x+1) - D'(x)

Equation III

D"(X) = y(x+2) - 2 * y(x+1) + y(x)

wherein y(x), y(x+1) and y(x+2) represent signal values at the x^th cycle number, the (x+1)^th cycle number and the (x+2)^th cycle number, respectively; D'(x) represents a first-order difference at the x^th cycle number; and D"(x) represents a second-order difference at the x^th cycle number.

According to the backward difference method, the second-order difference may be calculated using the following Equations IV and V sequentially or by using the following Equation VI alone:

Equation IV

D'(x) = y(x) - y(x-1)

Equation V

D"(x) = D'(x) - D'(x-1)

Equation VI

D"(x) = y(x) - 2 * y(x-1) + y(x-2)

wherein y(x), y(x-1) and y(x-2) represent signal values at the x^th cycle number, the (x-1)^th cycle number and the (x-2)^th cycle number, respectively; D'(x) represents a first-order difference at the x^th cycle number; and D"(x) represents a second-order difference at the x^th cycle number.

For example, for the forward difference method, if y(1), y(2), y(3), y(4) and y(5) are 1, 8, 30, 100 and 500, respectively, D'(1), D'(2), D'(3) and D'(4) will be 7 (=8-1), 22 (=30-8), 70 (=100-30) and 400 (=500-100), respectively, and D"(1), D"(2) and D"(3) will be 15 (=22-7), 48 (=70-22) and 330 (=400-70), respectively; whereas, for the backward difference method, if y(1), y(2), y(3), y(4) and y(5) are 1, 8, 30, 100 and 500, respectively, D'(2), D'(3), D'(4) and D'(5) will be 7 (=8-1), 22 (=30-8), 70 (=100-30) and 400 (=500-100), respectively and D"(3), D"(4) and D"(5) will be 15 (=22-7), 48 (=70-22) and 330 (=400-70), respectively.

It is noted that D'(x) may be calculated by other equation, e.g., D'(x) = y(x) - y(x+1) or D'(x) = y(x-1) - y(x); and D"(x) may be calculated by other equation D"(x) = D'(x) - D'(x+1), D"(X) = y(x) - 2 * y(x+1) + y(x+2), D"(x) = D'(x-1) - D'(x), or D"(x) = y(x-2) - 2 * y(x-1) + y(x).

In the forward difference method, both the end cycle number and the cycle number immediately before the end cycle number do not have D"(x) calculated. For example, where the signal amplification reaction is performed up to 45 cycle numbers, D"(45) and D"(44) cannot be calculated due to absence of signal values at the 46^th cycle number or the 47^th cycle number.

Likewise, in the backward difference method, both the first cycle number and the cycle number immediately after the first cycle number (second cycle number) do not have D"(x) calculated. For example, where the signal amplification reaction is performed up to 45 cycle numbers, D"(1) and D"(2) cannot be calculated due to absence of signal values at the -1^th cycle number or the 0^th cycle number.

As such, the functions such as y(x), D(x), D'(x) and D"(x) are not calculated for a non-existing or undefinable cycle number.

It is noted that the results obtained by the forward difference method and the backward difference method have a certain interconvertibility with respect to cycle numbers.

Specifically, the D'(x) obtained by the forward difference method is identical to the D'(x+1) obtained by the backward difference method. For example, the first-order difference at the 1^st cycle number obtained by the forward difference method is the same as that at the 2^nd cycle number obtained by the backward difference method. Therefore, in view of such interconvertibility, one of skill in the art can readily convert the result of the forward difference method into the result of the backward difference method, or vice versa. For example, where the first-order difference is obtained by the forward difference method, a first-order difference calculated at the (x+1)^th cycle number may be regarded as a first-order difference at the x^th cycle number so as to convert into the result of the backward difference method.

Furthermore, the D"(x) obtained by the forward difference method is identical to the D"(x+2) obtained by the backward difference method. For example, the second-order difference at the 1^st cycle number obtained by the forward difference method is the same as that at the 3^rd cycle number obtained by the backward difference method.

Therefore, in view of such interconvertibility, one of skill in the art can readily convert the result of the forward difference method into the result of the backward difference method, or vice versa. For example, where the second-order difference is obtained by the forward difference method, a second-order difference calculated at the (x+2)^th cycle number may be regarded as a second-order difference at the x^th cycle number so as to convert it into the result of the backward difference method.

The calculation of the second-order differences by a backward difference method is illustrated in Examples of the present application. However, those skilled in the art can employ a forward difference method instead of the backward difference method such that only cycle numbers are altered to obtain the same result as the backward difference method. Therefore, it will be appreciated by one of skill in the art that the first-order difference and the second-order difference at a specific cycle number may encompass those calculated at other cycle numbers, depending upon the calculation method used.

(ii) Calculating a normality score at each cycle number

Next, a normality score is calculated at each cycle number of the dataset by using the second-order change value.

Herein, a "normality score dataset" including a plurality of data points having a cycle number and a normality score at a cycle number is obtained.

The calculation of the normality score is performed by a mathematical operation that represents sign change between and magnitudes of the second-order change values at two consecutive cycle numbers.

The term "normality score" (abbreviated as 'NS') as used herein refers to a numerical value that represents both the sign change between the second-order change values at the two immediately adjacent cycle numbers and the magnitudes of the second-order change values at the two immediately adjacent cycle numbers. The normality score may be a numerical value representing the extent of normality of signals obtained by a signal amplification reaction. A small normality score indicates that the cycle number having the normality score is highly likely to exhibit an abnormal signal. For example, a particular cycle number with a normality score of 1000 is likely to represent a normal signal, while another cycle number with a normality score of -2000 is likely to represent an abnormal signal.

The normality score is used not only to select a candidate cycle number among all cycle numbers in a dataset but also to determine an abnormal signal by comparison between the normality score before alteration of signal value (also referred to as "pre-alteration normality score") and the normality score after alteration of single value (also referred to as "post-alteration normality score").

The term "normality score at a cycle number" as used herein refers to a normality score which is calculated using a second-order change value at a particular cycle number and a second-order change value at a cycle number immediately adjacent to the particular cycle number, and which is assigned (allocated) to the particular cycle number. It can be written in various expressions, such as a normality score obtained from the signal value at a cycle number, a normality score obtained from the second-order change value of the signal value at a cycle number, a normality score calculated at a cycle number, or a variation thereof. It is to be understood that a normality score at a specific cycle number may encompass one calculated at other cycle numbers in a broad sense.

The immediately adjacent cycle number used to calculate the normality score at the x^th cycle number includes the cycle number immediately before or after the x^th cycle number. For example, in a typical dataset in which the cycle number increases by 1, the immediately adjacent cycle number of the x^th cycle number is the cycle number x+1 or cycle number x-1.

The normality score is obtained by a mathematical operation that represents sign change between, and magnitudes of, the second-order change values at two immediately adjacent cycle numbers. As described above, the second-order change value may be, for example, a second-order difference, a second-order difference quotient, or a second-order derivative.

According to an embodiment of the present invention, the normality score at the x^th cycle number is calculated by a mathematical operation that represents sign change between and magnitudes of the second-order change values at the x^th cycle number and the (x+1)^th cycle number.

According to another embodiment of the present invention, the normality score at the x^th cycle number is calculated by a mathematical operation that represents sign change between and magnitudes of the second-order change values at the x^th cycle number and the second-order change value at the (x-1)^th cycle number.

The sign change used herein indicates a circumstance where a second-order change value at the x^th cycle number has a positive sign (positive value, +) and a second-order change value at the (x+1)^thor(x-1)^th cycle number has a negative sign (negative value, -). A mathematical operation that represents the sign change between the second-order change values is any operation that provides a negative normality score when a second-order change value at the x^th cycle number has a positive sign and a second-order change value at the (x+1)^thor(x-1)^th cycle number has a negative sign; and provides a positive normality score when a second-order change value at the x^th cycle number and a second-order change value at the (x+1)^thor(x-1)^th cycle number both have the same sign (e.g., all positive values or all negative values).

The magnitudes of the second-order change values as used herein indicate a combination of both the magnitude of the second-order change value at the x^th cycle number and the magnitude of the second-order change value at the (x+1)^thor(x-1)^th cycle number.

Examples of the mathematical operation that represents sign change between and magnitudes of the second-order change values include a mathematical magnification (e.g., multiplication) of the second-order change values, a mathematical ratio (e.g., division) of the second-order change values, and the like.

According to one embodiment, the normality score is calculated by multiplying the second-order change values at the two immediately adjacent cycle numbers. The multiplication of the second-order change values can effectively represent sign change between and magnitudes of the second-order change values.

More specifically, the normality score may be calculated by the following Equation VII or VIII:

Equation VII

NS(x) = D"(x) * D"(x+1)

wherein NS(x) represents a normality score at the x^th cycle number; D"(x) represents a second-order change value at the x^th cycle number; D"(x+1) represents a second-order change value at the (x+1)^th cycle number; and x is an integer of 1 or more.

Equation VIII

NS(x) = D"(x) * D"(x-1)

wherein NS(x) represents a normality score at the x^th cycle number; D"(x) represents a second-order change value at the x^th cycle number; D"(x-1) represents a second-order change value at the (x-1)^th cycle number; and x is an integer of 2 or more.

In the Equations VII and VIII, D"(x-1), D"(x) and D"(x+1) indicate second-order difference at cycle number x-1, x and x+1, respectively.

For example, if D"(1), D"(2), D"(3) and D"(4) is -100, 50, 350, -400, respectively, NS(1), NS(2) and NS(3) will be -5000 (=-100*50), 17500 (=50*350) and -14000 (=350*-400), respectively, for the Equation VII; while NS(2), NS(3) and NS(4) will be -5000 (=-100*50), 17500 (=50*350) and -14000 (=350*-400), respectively, for the Equation VIII.

It is noted that the normality score at the end cycle number cannot be calculated upon using the Equation VII; whereas the normality score at the first cycle number cannot be calculated upon using the Equation VIII.

It is further noted that the normality scores obtained by the Equation VII and the normality scores obtained by the Equation VIII have certain intercovertibility with respect to cycle numbers.

Specifically, the NS(x) obtained by the Equation VII is the same as the normality score at the NS(x+1) obtained by the Equation VIII. For example, the normality score at the 4^th cycle number obtained by the Equation VII is the same as the normality score at the 5^th cycle number obtained by the Equation VIII.

Therefore, in view of such intercovertibility, one of skill in the art can readily convert the normality scores obtained by the Equations VII into the normality scores obtained by the Equation VIII, or vice versa.

The calculation of normality scores by the Equation VII is exemplified in Examples of the present application. However, those skilled in the art will be able to calculate the normality score at each cycle number using the Equation VIII instead of the Equation VII and then adjust the cycle number, i.e., subtract 1 from the calculated cycle number, thereby obtaining the same result as that by the Equation VII. Therefore, it is to be understood that the normality score at a specific cycle number may encompass the normality score calculated at other cycle number, depending upon its calculation method.

Meanwhile, the calculation of the normality score using the Equation VII or Equation VIII may be combined with the calculation of the second-order change value using the forward or backward difference method in various ways.

For example, the calculation of the second-order difference using the forward difference method may be combined with the calculation of the normality score using Equation VII; the calculation of the second-order difference using the forward difference method may be combined with the calculation of the normality score using Equation VIII; the calculation of the second-order difference using the backward difference method may be combined with the calculation of the normality score using Equation VII; or the calculation of the second-order difference using the backward difference method may be combined with the calculation of the normality score using Equation VIII.

However, it should be noted that, in such various combinations of either the forward difference method or backward difference method and either the Equation VII or Equation VIII, the cycle number having the same normality score will vary with certain regularity, and thus the cycle number indicative of an abnormal signal will also vary with certain regularity.

For example, where a combination of the forward difference method and the Equation VII yields a normality score of -63 at the 2^nd cycle number, a combination of the forward difference method and the Equation VIII will yield the same normality score at the 3^rd cycle number, a combination of the backward difference method and the Equation VII will yield the same normality score at the 4^th cycle number; and a combination of the backward difference method and the Equation VIII will yield the same normality score at the 5^th cycle number. Thus, the cycle number may vary depending upon the calculation methods of the normality score.

The present inventors have verified that a cycle number theoretically determined to be indicative of an abnormal signal by a combination of the backward difference method and the Equation VII exhibits an abnormal signal in a visual inspection on an actual dataset. This proves that the method of present invention using a combination of the backward difference method and the Equation VII is very accurate and effective in determining a cycle number indicative of an abnormal signal. However, considering the above-described rules of cycle number change, one skilled in the art will be able to use other combinations. For example, where a combination of the forward difference method and the Equation VII is used, the cycle number indicative of an abnormal signal may be adjusted by adding cycle number 2 to the result. Therefore, it will be appreciated by one of skill in the art that such various combinations are within the scope of the present invention.

As described above, depending upon the combination of either the forward or the reverse difference method and either the Equation VII or the Equation VIII, the first-order difference, the second-order difference and the normality score corresponding to each cycle number may vary with a certain regularity. Therefore, it should be understood that the first-order difference, the second-order difference and the normality score at a specific cycle number may encompass those at other cycle numbers.

The calculation of the normality score by the combination of the backward difference method and the Equation VII can be also found in Examples of the present application.

Step (c): Selecting a candidate cycle number(s) for an abnormal signal by the normality-representing value 130

In this step, a candidate cycle number(s) for an abnormal signal is selected by the normality-representing value; wherein the normality-representing value at the candidate cycle number(s) corresponds to a pre-alteration normality-representing value 130. In particular, where the normality-representing value is a normality score, a candidate cycle number(s) for an abnormal signal is selected by the normality score.

The term "candidate cycle number" as used herein refers to a cycle number that is likely to or expected to exhibit an abnormal signal.

The present inventors have found that the normality-representing value such as a normality score suggested by the present invention is highly associated with abnormal signals, i.e., the normality-representing value such as a normality score can be used as an indicator for an abnormal signal. According to our findings, where the normality-representing value is a normality score and the normality score is calculated by Equation VII or VIII, the smaller the normality score calculated at a specific cycle number is, the higher the likelihood that the cycle number exhibits an abnormal signal is. Thus, the normality-representing value such as a normality score is used to find a candidate cycle number for an abnormal signal.

The expression "selecting a candidate cycle number(s) for an abnormal signal by the normality-representing value" means that a candidate cycle number(s) is selected based on its normality-representing value, i.e., the normality-representing value serves as a criterion for selecting a candidate cycle number.

The selection of the candidate cycle number may be performed based on the size (magnitude) of the normality-representing value, in particular the normality score. For example, the normality score "70" is deemed to be larger in size compared to the normality score "-80".

The selection of the candidate cycle number may be accomplished by various methods, as follows:

In an embodiment, where a larger normality-representing value indicates a higher extent of normality at a cycle number, the candidate cycle number to be selected is a cycle number having a normality-representing value smaller than a threshold.

The selection of the candidate cycle number may be performed by applying a threshold to the normality-representing values at all cycle numbers; and selecting a cycle number(s) having a normality-representing value smaller than the threshold.

The number of the candidate cycle number(s) to be selected may be 0 (zero) or more. For example, if there is no cycle number having a normality-representing value smaller than the threshold, the number of the candidate cycle number to be selected is 0 (zero). If there is only one cycle number having a normality-representing value smaller than the threshold, the number of the candidate cycle number to be selected is 1 (one). If there are a plurality of cycle numbers having a normality-representing value smaller than the threshold, the number of the candidate cycle number to be selected is more than 1 (one).

In view of the fact that a cycle number having a small normality-representing value is likely to exhibit an abnormal signal, the threshold used to select a candidate cycle number may have a signal value less than zero. The threshold may be determined empirically or experimentally. The threshold may be a RFU (relative fluorescence unit) of -100, -200, -300, -400, -500, -600, -700, -800, -900, -1000, -2000, -3000, -4000 or less, including every value in between these numbers, for example.

In another embodiment, where a larger normality-representing value indicates a higher extent of normality at a cycle number, the candidate cycle number to be selected is a cycle number having a normality-representing value which is smaller than a threshold and which is a minimum.

The selection of the candidate cycle number is performed by applying a threshold to the normality-representing value s at all cycle numbers; and selecting a cycle number having a normality-representing value which is smaller than the threshold and is a minimum.

The number of the candidate cycle number(s) to be selected may be 0 (zero) or 1 (one). For example, if there is no cycle number having a normality-representing value smaller than the threshold, the number of the candidate cycle number to be selected is 0 (zero). If there are one or more cycle numbers having a normality-representing value smaller than the threshold, the number of the candidate cycle number to be selected is 1 (one).

In still another embodiment, where a larger normality-representing value indicates a higher extent of normality at a cycle number, the candidate cycle number to be selected is a cycle number having a negative sign and a minimum normality-representing value.

The selection of the candidate cycle number is performed without using a threshold, i.e., by selecting a cycle number having a negative sign and a minimum normality-representing value, as a candidate cycle number.

The number of the candidate cycle number(s) to be selected may be 0 (zero) or 1 (one). For example, if there is no cycle number having a negative sign, the number of the candidate cycle number to be selected is 0 (zero). If there are one or more cycle numbers having a negative sign, the number of the candidate cycle number to be selected is 1 (one).

In any of the embodiments as described above, if there is no cycle number satisfying a defined criterion (e.g., having a normality-representing value smaller than a threshold; having a normality-representing value which is smaller than a threshold and which is a minimum; or having a negative sign and a minimum normality-representing value), no candidate cycle number is selected in a dataset. This indicates that the dataset to be analyzed by the present invention contains no cycle number indicative of an abnormal signal. In this case, an indication that the dataset obtained in step (a) does not contain an abnormal signal, i.e., an indication that the dataset consists of normal signals, may be displayed.

In an alternative embodiment, where a smaller normality-representing value indicates a higher extent of normality of a signal value at a cycle number, the candidate cycle number is (i) a cycle number having a normality-representing value larger than a threshold, wherein the threshold is selected from values more than 0; (ii) a cycle number having a normality-representing value which is larger than a threshold and which is a maximum, wherein the threshold is selected from values larger than 0; or (iii) a cycle number having a positive sign and a maximum normality-representing value.

As described above, upon selecting a candidate cycle number, the normality-representing value at the candidate cycle number is adopted as a pre-alteration normality-representing value at the candidate cycle number.

The term "pre-alteration normality-representing value" (e.g., pre-alteration normality score) as used herein refers to a normality-representing value (e.g., a normality score) corresponding to the selected candidate cycle number among the normality-representing values or normality scores provided in step (b). The pre-alteration normality-representing value (e.g., the normality score) is one which has been calculated at a particular cycle number in step (b) and is allocated to the particular cycle number. In other words, the pre-alteration normality-representing value (e.g., the normality score) is one which has been calculated at a particular cycle number using signal values before being altered by the signal alteration scheme in step (d) (i.e., using unaltered signal values or pre-alteration signal values).

The term "pre-alteration normality-representing value" (e.g., pre-alteration normality score) is distinguished from a "post-alteration normality-representing value" (e.g., post-alteration normality score) as described below.

The pre-alteration normality-representing value (e.g., the pre-alteration normality score) at the candidate cycle number is not newly calculated or generated in this step, but is provided or acquired spontaneously or automatically upon the selection of the candidate cycle number.

Specifically, the normality-representing values (e.g., the normality scores) at all cycle numbers are provided in the step (b), and the normality-representing value (e.g., the normality score) at the candidate cycle number is designated as or allocated to the pre-alteration normality-representing value (e.g., the pre-alteration normality score) at the candidate cycle number. Therefore, the normality-representing value (e.g., the normality score) at the candidate cycle number corresponds to a pre-alteration normality-representing value (e.g., a pre-alteration normality score).

The pre-alteration normality-representing value (e.g., the pre-alteration normality score) at the candidate cycle number is used to determine whether the candidate cycle number is indicative of an abnormal signal, by comparing it with a "post-alteration normality-representing value" (e.g., post-alteration normality score) at the candidate cycle number.

Step (d): Altering the signal value at the candidate cycle number and further providing a normality-representing value 140

In this step, the signal value at the candidate cycle number(s) is altered, and a normality-representing value at the candidate cycle number(s) is further provided by using the altered signal value; wherein the normality-representing value further provided at the candidate cycle number(s) corresponds to a post-alteration normality-representing value 140.

The modifiers "pre-alteration" and "post-alteration" are used herein to distinguish two elements, one obtained before altering the signal value at the candidate cycle number and another obtained after altering the signal value at the candidate cycle number. The modifier "pre-alteration" may be used interchangeably with "unaltered", and the modifier "post-alteration" may be used interchangeably with "altered".

In this regards, the term "pre-alteration signal value" refers to a signal value at a certain cycle number (particularly, a candidate cycle number) which has not yet been altered; whereas the term "post-alteration signal value" refers to a signal value at a certain cycle number which has been altered. The term "pre-alteration cycle number" refers to a cycle number having the pre-alteration signal value; whereas the term "post-alteration cycle number" refers to a cycle number having the post-alteration signal value. The terms "pre-alteration first-order difference", "pre-alteration second-order difference" and "pre-alteration normality-representing value" (e.g., "pre-alteration normality score") refer to a first-order difference, a second-order difference and a normality-representing value (e.g., a normality score) calculated at the pre-alteration cycle number, respectively; whereas the terms "post-alteration first-order difference", "post-alteration second-order difference" and "post-alteration normality-representing value" (e.g., "post-alteration normality score") refer to a first-order difference, a second-order difference and a normality-representing value (e.g., a normality score) calculated at the post-alteration cycle number, respectively. Further, the term "pre-alteration dataset" refers to a dataset consisting of pre-alteration signal values; whereas the term "post-alteration dataset" refers to a dataset comprising one or more post-alteration signal values.

In this step, the signal value at the candidate cycle number may be altered in various ways, including signal correction methods known in the art

The signal value at the candidate cycle number may be altered by a signal alteration scheme (method) as described herein.

The term "signal alteration scheme (method)" as used herein refers to a means or rule for altering the signal value at the candidate cycle number into another appropriate signal value. The signal alteration scheme (method) may be, for example, a means for determining an alteration direction (e.g., whether the signal value at the candidate cycle number is increased or decreased) and an alteration degree (e.g., how much the signal value at the candidate cycle number is increased or decreased) for a signal value at a candidate cycle number. Alternatively, the signal alteration scheme (method) may be a means for determining a post-alteration signal value (larger or smaller than the pre-alteration signal value) into which the pre-alteration signal value at the candidate cycle number is altered.

The signal alteration scheme according to the present invention will be described in detail as below.

In an embodiment, the signal alteration scheme comprises altering the signal value at the candidate cycle number such that its absolute value is decreased. That is, the signal value at the candidate cycle number is altered by the signal alteration scheme such that its absolute value is decreased.

According to an embodiment, such alteration does not encompass the alteration where the sign of the signal value is changed.

For example, assuming that the signal value at the candidate cycle number is "+10", such alteration of the signal value comprises, for example, the alteration into +9, +8, +7, +6, +5, +4, +3, +2, +1, and so on, but does not comprise, for example, the alteration into -1, -2, -3, -4, -5, -6, -7, -8, -9 and so on. Similarly, where the signal value at the candidate cycle number is "-10", the alteration of the signal value comprises, for example, the alteration into -9, -8, -7, -6, -5, -4, -3, -2, -1, and so on, but does not comprise, for example, the alteration into +1, +2, +3, +4, +5, +6, +7, +8, +9 and so on.

According to an embodiment, the absolute value of the signal value at the candidate cycle number is decreased to 50% or less, 40% or less, 30% or less, 20% or less, 10% or less, or 5% or less of those before alteration.

In a particular embodiment, the signal alteration scheme comprises altering the signal value at the candidate cycle number by the signal alteration scheme such that its absolute value is larger than or equal to a relatively small absolute value of signal values at two cycle numbers immediately adjacent to the candidate cycle number. That is, the signal value at the candidate cycle number is altered such that its absolute value is larger than or equal to a relatively small absolute value of signal values at two cycle numbers immediately adjacent to the candidate cycle number.

For example, assuming that the signal value at the candidate cycle number is "11" and two signal values at cycle numbers immediately adjacent to the candidate cycle number are "6" and "8", the signal value "11" at the candidate cycle number may be altered into any signal value which is smaller than "11" and which is larger than or equal to "6". Assuming that the signal value at the candidate cycle number is "-11" and two signal values at cycle numbers immediately adjacent to the candidate cycle number are "-6" and "-8", the signal value "-11" at the candidate cycle number may be altered into any signal value which is larger than "-11" and which is smaller than or equal to "-6".

In an embodiment, the signal alteration scheme comprises altering the signal value at the candidate cycle number by the signal alteration scheme such that its absolute value is smaller than a relatively large absolute value of signal values at two cycle numbers immediately adjacent to the candidate cycle number

In a further particular embodiment, the signal alteration scheme comprises altering the signal value at the candidate cycle number such that its absolute value is close to a relatively small absolute value of signal values at two cycle numbers immediately adjacent to the candidate cycle number. That is, the signal value at the candidate cycle number is altered by the signal alteration scheme such that its absolute value is close to a relatively small absolute value of signal values at two cycle numbers immediately adjacent to the candidate cycle number.

For example, assuming that the signal value at the candidate cycle number is "11" and two signal values at cycle numbers immediately adjacent to the candidate cycle number are "6" and "8", the signal value "11" at the candidate cycle number may be altered into a value close to "6", e.g., 6.1, 6.2, and the like.

In a further particular embodiment, the signal alteration scheme comprises altering the signal value at the candidate cycle number such that its value is close to a signal value at the cycle number immediately before the candidate cycle number.

In an embodiment, when the signal value at the candidate cycle number is altered considering a reference signal value such as (i) a relatively small absolute value of signal values at two cycle numbers immediately adjacent to the candidate cycle number or (ii) a signal value at the cycle number immediately before the candidate cycle number, a post-altered signal value is within 150%-50%, within 140%-60%, within 130%-70%, within 120%-80%, within 110%-90% or within 105%-95% of the reference signal value. In an embodiment, a post-altered signal value is 150% or less, 140% or less, 130% or less, 120% or less, 110% or less, or 105% or less of the reference signal value. In an embodiment, a post-altered signal value is 50% or more, 60% or more, 70% or more, 80% or more, 90% or more, or 95% or more of the reference signal value.

In an embodiment, the signal alteration scheme provides a post-altered signal value which is larger than or equal to the reference signal value in a dataset representing a signal-increasing curve or a post-altered signal value which is smaller than or equal to the reference signal value in a dataset representing a signal-decreasing curve.

In an embodiment, an alteration by a signal alteration scheme does not encompass the alteration where the sign of the signal value is changed.

In another embodiment, the signal alteration scheme comprises altering the signal value at the candidate cycle number such that a first-order change value at the candidate cycle number is decreased. That is, the signal value at the candidate cycle number is altered by the signal alteration scheme such that a first-order change value at the candidate cycle number is decreased.

In still another embodiment, the signal alteration scheme comprises decreasing the signal value at the candidate cycle number, if the second-order change value at the candidate cycle number is a negative number, or increasing the signal value at the candidate cycle number, if the second-order change value at the candidate cycle number is a positive number.

The sign of the second-order change value at the candidate cycle number is related to the increase or decrease of the signal value at the candidate cycle number.

The second-order change value of a positive number indicates that the first-order change value at the candidate cycle number is larger than that at a cycle number immediately before the candidate cycle number, implying that the signal value at the candidate cycle number is relatively larger than that at a cycle number immediately before the candidate cycle number. Conversely, the second-order change value of a negative number at the candidate cycle number indicates that the first-order change value at the candidate cycle number is smaller than that at a cycle number immediately before the candidate cycle number, implying that the signal value at the candidate cycle number is relatively smaller than that at a cycle number immediately before the candidate cycle number. Therefore, the alteration direction (increase/decrease) of the signal value at the candidate cycle number may be determined based on a sign (positive sign or negative sign) of a second-order difference at the candidate cycle number. The second-order change value at the candidate cycle may be a second-order difference, a second-order difference quotient, or a second-order derivative of the signal value, in particular a second-order difference of the signal value, as described above.

In the embodiment using a sign of a second-order difference at the candidate cycle number, the signal alteration scheme comprises decreasing the signal value at the candidate cycle number, if the second-order change value at the candidate cycle number is a negative number, or increasing the signal value at the candidate cycle number, if the second-order change value at the candidate cycle number is a positive number.

In a particular embodiment using a sign of a second-order difference at the candidate cycle number, if the second-order change value at the candidate cycle number is a positive number, the signal alteration scheme in the step (d) comprises decreasing the signal value at the candidate cycle number such that it is larger than or equal to a relatively small signal value out of signal values at two cycle numbers immediately adjacent to the candidate cycle number.

In a particular embodiment using a sign of a second-order difference at the candidate cycle number, if the second-order change value at the candidate cycle number is a negative number, the signal alteration scheme in the step (d) comprises increasing the signal value at the candidate cycle number such that it is smaller than or equal to a relatively large signal value out of signal values at two cycle numbers immediately adjacent to the candidate cycle number.

Specifically, if the second-order change value at the candidate cycle number is a positive number, the signal value at the candidate cycle number may be altered into another value which is smaller than the signal value at the candidate cycle number and which is larger than or equal to a relatively small one of the two cycles immediately adjacent to the candidate cycle number. For example, where the cycle number having the signal value of "11" is selected as a candidate cycle number among three consecutive signal values "6", "11" and "8" and the second-order change value at the candidate cycle number is a positive number, the signal value "11" at the candidate cycle number may be altered into any signal value which is smaller than "11" and which is larger than or equal to "6". Examples of the signal value into which the signal value at the candidate cycle number is altered (post-alteration value) may include be 10, 9, 8, 7, or 6, including every value in between these numbers. The post-alteration value may be arbitrarily selected within the range.

Alternatively, if the second change value at the candidate cycle number is a negative number, the signal value at the candidate cycle number may be altered into another value which is larger than the signal value at the candidate cycle number and which is smaller than or equal to a relatively large one of the two cycles immediately adjacent to the candidate cycle number. For example, where the cycle number having the signal value of "3" is selected as a candidate cycle number among three consecutive signal values "6", "3" and "8" and the second-order change value at the candidate cycle number is a negative number, the signal value "3" at the candidate cycle number may be altered into any signal value which is larger than "3" and which is smaller than or equal to "8". Examples of the signal value into which the signal value at the candidate cycle number is altered (post-alteration value) may include 4, 5, 6, 7, or 8, including every value in between these numbers. The post-alteration value may be arbitrarily selected within the range.

In still another embodiment, where a normality score is used as an example of the normality-representing value, the signal value at the candidate cycle number is altered into a post-alteration signal value, which is obtained by (i) obtaining a post-alteration first-order change value at the candidate cycle number by using a pre-alteration first-order change value at the candidate cycle number and a pre-alteration normality score at the candidate cycle number; and (ii) obtaining a post-alteration signal value at the candidate cycle number by using the post-alteration first-order change value at the candidate cycle number.

According to the embodiment, an absolute value of the post-alteration first-order change value at the candidate cycle number is smaller than that of the pre-alteration first-order change value at the candidate cycle number. Where a normality score is used as an example of the normality-representing value, a pre-alteration first-order change value at the candidate cycle number and a pre-alteration normality score at the candidate cycle number may be used to set up a formula providing such a smaller absolute value of the post-alteration first-order change value. In an embodiment, a smaller absolute value of the post-alteration first-order change value can provide a post-alteration signal value to be close to a signal value at the cycle number immediately before the candidate cycle number.

According to the embodiment, the pre-alteration first-order change value may be calculated using two signal values at two consecutive cycle number. The pre-alteration first-order change value may be a pre-alteration first-order difference. The pre-alteration first-order difference may be calculated by a forward difference method as in Equation I or a backward difference method as in Equation IV.

According to a particular embodiment, where a normality score is used as an example of the normality-representing value, the pre-alteration normality score at the candidate cycle number may be calculated by Equation VII or VIII.

The post-alteration signal value at the candidate cycle number may be calculated by summing the pre-alteration signal value at a cycle number immediately adjacent to the candidate cycle number and the post-alteration first-order change value at the candidate cycle number.

This embodiment can be further implemented by using the following Equations IX and X.

The signal value at the candidate cycle number (pre-alteration signal value) may be altered into a signal value (post-alteration signal value) determined by Equations IX and X.

Equation IX

wherein D'_post(c) represents a post-alteration first-order change value at the candidate cycle number; D'_pre(c) represents a pre-alteration first-order change value at the candidate cycle number; NS_pre(c) represents a pre-alteration normality score at the candidate cycle number; k is a number of 1 or more; and c represents the candidate cycle number;

Equation X

y_post(c) = y_pre(c-1) + D'_post(c)

wherein y_post(c) represents a post-alteration signal value at the candidate cycle number; y_pre(c-1) represents a pre-alteration signal value at the (candidate cycle number-1); D'_post(c) represents the post-alteration first-order change value at the candidate cycle number, as determined by Equation IX.

A post-alteration first-order change value at the candidate cycle number (D'_post(c)) may be calculated by an equation, D'_post(c) = y_post(c) - y_post(c-1), and thus y_post(c) may be expressed by an equation, y_post(c) = y_post(c-1) + D'_post(c). In view that the signal value at the (c-1)^th cycle number is not altered, y_post(c-1) may be considered to be equal to y_pre(c-1). Therefore, y_post(c) may be calculated by the sum of y_pre(c-1) and D'_post(c), as shown in Equation X.

In the above Equation IX, k may a number of around 1, particularly 1.

An example of the signal alteration scheme using Equations IX and X will be described below:

In the case of altering the signal value at the c^th cycle number (as a candidate cycle number) by using Equations IX and X, a pre-alteration first-order difference at the candidate cycle number (D'_pre(c)) and a pre-alteration normality score (NS_pre(c)) are applied to the Equation IX, thereby obtaining a post-alteration first-order difference at the candidate cycle number (D'_post(c)).

For example, assuming a certain dataset in which a candidate cycle number (e.g., 10^th cycle number) has a signal value of "8916.60" (in RFU), a pre-alteration first-order difference of "128.81" and a pre-alteration normality score of "-20235.34", the post-alteration first-order difference at the candidate cycle number will be calculated by Equation IX as follows:

Then, the post-alteration first-order difference at the candidate cycle number (D'_post(c)) and the signal value at the cycle number immediately before the candidate cycle number (y_pre(c-1)) are applied to the Equation X, thereby obtaining a post-alteration signal value at the candidate cycle number (y_post(c)).

For example, assuming that the signal value at the 9^th cycle number is 8787.79, the post-alteration signal value at the candidate cycle number can be calculated by Equation X as follows: y_post(10) = 8787.79 + 0.90 = 8788.69.

As a result, the signal value "8916.60" at the candidate cycle number may be altered into another signal value "8788.69" according to Equations IX and X above.

The signal alteration scheme according to this embodiment of the present invention is illustrated in Fig. 4. As shown in Fig. 4, the signal value at the candidate cycle number (y_pre(10)) is replaced with a post-alteration signal value at the candidate cycle number (y_post(10)).

As such, the use of Equations IX and X allows for providing a post-alteration signal value at the candidate cycle number in an automatic manner, thereby enabling the original signal value at the candidate cycle number to be automatically replaced with another signal value, i.e., post-alteration signal value.

After the signal value at the candidate cycle number is altered as described above, a normality-representing value at the candidate cycle number is further provided by using the altered signal value (post-alteration signal value). In particular, where a normality score is used as an example of the normality-representing value, a normality score at the candidate cycle number is further provided by using the altered signal value (post-alteration signal value)

The normality-representing value or the normality score further provided at the candidate cycle number by using the altered signal value is referred to as a "post-alteration normality-representing value" or a "post-alteration normality score" at the candidate cycle number. Therefore, the normality-representing value or the normality score further provided at the candidate cycle number(s) corresponds to a post-alteration normality-representing value or a post-alteration normality score.

The term "post-alteration normality-representing value" or "post-alteration normality score" as used herein refers to a normality-representing value or a normality score calculated at a particular cycle number using the signal value altered in step (d) (i.e., using post-alteration signal value).

The post-alteration normality-representing value or the post-alteration normality score may be calculated in a same manner to the calculation of the pre-alteration normality-representing value or the pre-alteration normality score in step (b).

For example, the post-alteration normality score as an example of the normality-representing value may be further provided by re-calculating a second-order change value at the candidate cycle number using the post-alteration signal value at the candidate cycle number, and re-calculating a normality score at the candidate cycle number using the second-order change value.

The post-alteration normality-representing value at the candidate cycle number is used to determine whether the candidate cycle number is indicative of an abnormal signal, by comparing it with a "pre-alteration normality-representing value" at the candidate cycle number.

Step (e): Comparing the post-alteration normality-representing value to the pre-alteration normality-representing value 150

In this step, the post-alteration normality-representing value is compared to the pre-alteration normality-representing value at the candidate cycle number(s) 150. In particular, where a normality score is used as an example of the normality-representing value, the post-alteration normality score is compared to the pre-alteration normality score at the candidate cycle number(s).

The term "compare", "comparing" or "comparison" in connection with two normality-representing values refers to a process of determining which of the two normality-representing values is larger. The comparison between the two normality-representing values should be performed by considering the signs of the normality-representing values, i.e., positive sign or negative sign. For instance, a normality-representing value "10" is considered to be larger than a normality-representing value "-80".

The post-alteration normality-representing value used for the comparison is one further provided in step (d); whereas the pre-alteration normality-representing value is one provided in step (b) and adopted in step (c).

Step (f): Determining a cycle number indicative of an abnormal signal 160

In this step, whether the candidate cycle number(s) is indicative of an abnormal signal is determined based on the result of the comparison in step (e) 160.

Specifically, the candidate cycle number is determined to be indicative of an abnormal signal, with a proviso that the post-alteration normality-representing value indicates a higher extent of normality than the pre-alteration normality-representing value at the candidate cycle number.

In an embodiment where a larger normality-representing value indicates a higher extent of normality at a cycle number, the candidate cycle number is determined to be indicative of an abnormal signal, with a proviso that the post-alteration normality-representing value is larger than the pre-alteration normality-representing value at the candidate cycle number. A typical example of the normality-representing value having such characteristics includes a normality score as described herein.

In such case, the small normality-representing value at a certain cycle number is highly associated with an abnormal signal at the cycle number. In other words, the smaller the normality-representing value at a certain cycle number is, the higher the probability that the cycle number exhibits an abnormal signal is. Accordingly, if the magnitude of the normality-representing value at the candidate cycle number becomes larger by the alteration of the signal value according to the present invention, it can be seen that the abnormal signal at the candidate cycle number is altered to a normal signal and thus the candidate cycle number can be determined as a cycle number indicative of an abnormal signal. In contrast, if the magnitude of the normality-representing value at the candidate cycle number becomes smaller by the alteration of the signal value according to the present invention, it can be seen that the normal signal at the candidate cycle number is rather altered to an abnormal signal and thus the candidate cycle number can be determined as a cycle number indicative of a normal signal.

In an alternative embodiment where a smaller normality-representing value indicates a higher extent of normality at a cycle number, the candidate cycle number is determined to be indicative of an abnormal signal, with a proviso that the post-alteration normality-representing value is smaller than the pre-alteration normality-representing value at the candidate cycle number.

In such case, the small normality-representing value at a certain cycle number is highly associated with a normal signal at the cycle number. In other words, the smaller the normality-representing value at a certain cycle number is, the lower the probability that the cycle number exhibits an abnormal signal is. Accordingly, if the magnitude of the normality-representing value at the candidate cycle number becomes smaller by the alteration of the signal value according to the present invention, it can be seen that the abnormal signal at the candidate cycle number is altered to a normal signal and thus the candidate cycle number can be determined as a cycle number indicative of an abnormal signal. In contrast, if the magnitude of the normality-representing value at the candidate cycle number becomes larger by the alteration of the signal value according to the present invention, it can be seen that the normal signal at the candidate cycle number is rather altered to an abnormal signal and thus the candidate cycle number can be determined as a cycle number indicative of a normal signal.

As such, the present invention can distinguish between a cycle number indicative of an abnormal signal and a cycle number indicative of a normal signal, through comparison of the pre-alteration normality-representing value and the post-alteration normality-representing value at the cycle number.

The method of Section I described above may be applied to a plurality of datasets obtained by multiple PCR. For example, when a plurality of datasets are obtained by multiplex PCR, the method of Section I may be applied to each dataset to determine a cycle number indicative of an abnormal signal in each dataset as well as to correct each dataset. Specifically, when a plurality of datasets are obtained by detection at different detection temperatures from a single amplification reaction in accordance with the MuDT1 technology developed by the present inventors (see WO 2015/147412), the method of Sections I is applied to each dataset, thereby determining a cycle number indicative of an abnormal signal in each dataset as well as correcting each dataset.

In addition to the determination of a cycle number indicative of an abnormal signal, the method of the present invention may be applied to correct an abnormal signal in a dataset, i.e., to provide a corrected dataset.

In an embodiment, if the candidate cycle number is determined to be indicative of an abnormal signal in accordance with the present invention, a corrected dataset may be provided by altering the signal value at the candidate cycle number into a suitable signal value.

With respect to the provision of the corrected dataset, if the post-alteration normality-representing value indicates a higher extent of normality than the pre-alteration normality-representing value at the candidate cycle number, the post-alteration dataset may be provided as a corrected dataset; whereas if the post-alteration normality-representing value indicates a lower extent of normality than the pre-alteration normality-representing value at the candidate cycle number, the pre-alteration dataset may be provided instead of the post-alteration dataset while invalidating the alteration of the signal value at the candidate cycle number.

The term "alteration" as used herein refers to simply substituting or replacing a particular value with another value, while the term "correction" refers to substituting or replacing an abnormal value with a suitable normal value. In other words, unlike the alteration, the correction means that the characteristic of a particular value is improved.

Accordingly, the method of the present invention provides a corrected dataset (post-alteration dataset) if it is determined that the candidate cycle number is indicative of an abnormal signal; while it provides an uncorrected dataset (pre-alteration dataset) if it is determined that the candidate cycle number is indicative of a normal signal.

Alternatively, the method of the present invention may be designed to provide any one of a post-alteration dataset and a pre-alteration dataset, or to provide a cycle number indicative of an abnormal signal.

In an embodiment, when a corrected dataset (post-alteration dataset) is provided, a corrected dataset may be provided by further altering signal values at all cycle numbers after the candidate cycle number indicative of an abnormal signal as well as altering the signal value at the candidate cycle number into a suitable signal value. In such case, the alteration of the signal values at all cycle numbers after the cycle number indicative of an abnormal signal may be performed such that the difference between the post-alteration signal value and the pre-alteration signal value at each cycle number is same to that at the cycle number indicative of an abnormal signal.

In an embodiment, the method of the present invention is designed to display only the cycle number indicative of an abnormal signal without displaying the post-alteration dataset.

In another embodiment, the method of the present invention is designed to display only the post-alteration dataset without displaying the cycle number indicative of an abnormal signal.

In still another embodiment, the method of the present invention is designed to display the cycle number indicative of an abnormal signal together with the pre-alteration dataset, or to display the cycle number indicative of an abnormal signal together with the instruction such as 'retest' or 'invalidation'.

The design of such a method of the present invention may be readily adapted by those skilled in the art, and this should also be construed as being within the scope of the present invention.

As described above, the candidate cycle number(s) selected in step (c) of the method of the present invention may be a plurality of candidate cycle numbers.

In this case, the candidate cycle numbers are sequentially or simultaneously subjected to the steps (d)-(f).

Specifically, where a plurality of candidate cycle numbers, e.g., a first candidate cycle number and a second candidate cycle number, are selected in step (c), the signal values at the candidate cycle numbers are separately altered by a signal alteration scheme, the post-alteration normality-representing value is compared to the pre-alteration normality-representing value at each of the candidate cycle numbers, and then whether the each of the candidate cycle numbers is indicative of an abnormal signal is determined.

According to an embodiment, where a plurality of candidate cycle numbers are selected and a post-alteration normality-representing value at one candidate cycle is calculated after the alteration of the signal value at the one candidate cycle, signal values at other candidate cycles are pre-altered signal values, not post-altered signal values at other candidate cycles. Alternatively, post-altered signal values at other candidate cycles may be used instead of pre-altered signal values at other candidate cycles.

According to an embodiment, detection of an abnormal signal in a dataset may be performed several times with changing a way of selecting a candidate cycle number.

II. Detection of Abnormal Signal in Dataset by iterative mode

The method of the present invention may be repeatedly performed to detect additional abnormal signals. The repetition of the steps (a)-(f) allows not only to detect additional abnormal signals, but also to provide a more accurately corrected dataset.

In another aspect of this invention, there is provided a method for detecting an abnormal signal in a dataset in an iterative manner, the method further comprising repeating the steps of: (g) providing a post-alteration dataset after performing the step (f), wherein the post-alteration dataset is obtained by altering the signal value at the cycle number(s) indicative of an abnormal signal; and (h) performing the steps (a)-(f) using the post-alteration dataset instead of the dataset in the step (a).

The method may be also referred to as "iterative mode", "continuous mode", or "loop mode", in that it repeats steps (a)-(f) to detect additional abnormal signals.

According to the iterative mode, additional abnormal signals can be detected as follows:

Step (g): Providing a post-alteration dataset

First, a post-alteration dataset is provided after performing the step (f).

The post-alteration dataset is obtained by altering the signal value at the cycle number(s) determined to be indicative of an abnormal signal.

In step (f) above, a candidate cycle number is determined to be indicative of an abnormal signal, with a proviso that the post-alteration normality-representing value indicates a higher extent of normality than the pre-alteration normality-representing value at the candidate cycle number. In this case, the signal value at the candidate cycle number may be altered to provide a post-alteration dataset. In contrast, if the post-alteration normality-representing value indicates a lower extent of normality than the pre-alteration normality-representing value at the candidate cycle number, the post-alteration dataset is not provided and the iterative mode no longer proceeds.

The post-alteration dataset may be obtained by altering the signal value at the cycle number indicative of an abnormal signal in a same manner as in step (d).

The repetition of steps (a)-(f) is performed using the newly provided post-alteration dataset instead of the dataset of step (a).

When performing steps (a)-(f), the candidate cycle number may be selected in a same manner or in a different way. For example, in a first performance of steps (a)-(f), a cycle number having a normality-representing value smaller than a threshold may be selected as a candidate cycle number; in a second performance of steps (a)-(f), a cycle number having a normality-representing value which is smaller than a threshold and which is a minimum may be a cycle selected as a candidate cycle number.

In an embodiment, the post-alteration dataset is obtained by further altering signal values at all cycle numbers after the cycle number indicative of an abnormal signal.

The phrase "all cycle numbers after the cycle number indicative of an abnormal signal" refers to a cycle region ranging from a cycle number immediately after the cycle number indicative of an abnormal signal to an end cycle number. For example, in the case of a dataset consisting of a total of 45 cycle numbers, all cycle numbers after 25^th cycle number indicates a region ranging from 26^th cycle number to 45^th cycle number.

In an embodiment, the alteration of the signal values at all cycle numbers after the cycle number indicative of an abnormal signal is performed such that the difference between the post-alteration signal value and the pre-alteration signal value at each cycle number is same to that at the cycle number indicative of an abnormal signal.

For example, assuming that the difference between the pre-alteration signal value and the post-alteration signal value at the cycle number indicative of an abnormal signal is "10", the signal value at each cycle number after the cycle number indicative of an abnormal signal may be lowered by a difference of "10".

In another embodiment, the alteration of the signal values at all cycle numbers after the cycle number indicative of an abnormal signal is performed such that the first-order difference at each cycle number remains unchanged. For example, if the signal value at the 10^th cycle number has been altered, the post-alteration signal value at the 11^th cycle number may be calculated by summing the post-alteration signal value at the 10^th cycle number and the pre-alteration first-order difference at the 11^th cycle number, such that the post-alteration first-order difference is identical to the pre-alteration first-order difference at the 11^th cycle number. In the same way, the signal value at the 12^th cycle number may be calculated by summing the post-alteration signal value at the 11^th cycle number and the pre-alteration first-order difference at the 12^th cycle number.

In particular, where there are a plurality of cycle numbers indicative of an abnormal signal, the alteration of the signal values at and after any one of the cycle numbers is performed in consideration of the alteration of the signal values at and after other cycle number performed earlier.

For example, when there are 10^th and 27^th cycle numbers indicative of an abnormal signal, all signal values at and after the 10^th cycle number may be altered and then all signal values at and after the 27^th cycle number may be altered. In this case, the alteration of the signal value at the 10^th cycle number may also affect signal values at all cycle numbers (including 27^th cycle number) after the 10^th cycle number. For example, when the signal value at the 10^th cycle number is altered from "300" into "100", the signal value at 27^th cycle number is also lowered by a difference of "200". Therefore, the subsequent alteration at the 27^th cycle number should be performed in consideration of the alteration of the signal value at and after the 10^th cycle number. For example, when the signal value at and after the 27^th cycle number is altered, the signal values may be further altered from the values affected by the alteration of the 10^th cycle number.

Step (h): Performing the steps (a)-(f)

Next, the steps (a)-(f) are repeated using the post-alteration dataset instead of the dataset in the step (a).

The repetition of the steps (a)-(f) may be performed by providing a normality-representing value at each cycle number of the post-alteration dataset and selecting a candidate cycle number(s) for abnormal signals by the normality-representing value, wherein the normality-representing value at the candidate cycle number(s) corresponds to a pre-alteration normality-representing value; altering the signal value at the candidate cycle number(s), and further providing a normality-representing value at the candidate cycle number(s) by using the altered signal value; wherein the normality-representing value further provided at the candidate cycle number(s) corresponds to a post-alteration normality-representing value; comparing the post-alteration normality-representing value to the pre-alteration normality-representing value at the candidate cycle numbers; and determining the candidate cycle number(s) to be indicative of an abnormal signal, with a proviso that the post-alteration normality-representing value indicates a higher extent of normality than the pre-alteration normality-representing value at the candidate cycle number.

Where the normality-representing value is a normality score, the steps (a)-(f) may be performed by calculating a second-order change value at each cycle number of the post-alteration dataset; calculating a normality score at each cycle number of the post-alteration dataset by using the second-order change value; selecting a candidate cycle number(s) for abnormal signals by the normality score, wherein the normality score calculated at the candidate cycle number(s) corresponds to a pre-alteration normality score; altering the signal value at the candidate cycle number(s), and re-calculating a normality score at the candidate cycle number(s) by using the altered signal value; wherein the normality score re-calculated at the candidate cycle number(s) corresponds to a post-alteration normality score; comparing the post-alteration normality score to the pre-alteration normality score at the candidate cycle numbers; and determining the candidate cycle number(s) to be indicative of an abnormal signal, with a proviso that the post-alteration normality-representing value indicates a higher extent of normality than the pre-alteration normality-representing value at the candidate cycle number.

The repetition of the steps (a)-(f) is the same as that described in Section I, except for selecting a new candidate cycle number using a new post-alteration dataset in place of the pre-alteration dataset.

The repetition of the steps (a)-(f) may be terminated, if there is no cycle number determined to be indicative of an abnormal signal. For example, where one candidate cycle number is selected and the selected candidate cycle number is determined not to be indicative of an abnormal signal, or where a plurality of candidate cycle numbers are selected and all of the selected candidate cycle numbers are determined not to be indicative of an abnormal signal, the repetition of steps (a)-(f) may be terminated.

When the repetition is terminated, all the cycle numbers determined during the repetition are finally determined as a cycle number indicative of an abnormal signal in the original dataset.

In the repetition of steps (a)-(f), normality-representing values at all cycle numbers may be pre-calculated together with the post-alteration normality-representing value at a candidate cycle number. In other words, when the post-alteration normality-representing value at a candidate cycle number is calculated in a first performance of step (d), the normality-representing values at other cycle numbers to be calculated in a second performance of step (d) may also be pre-calculated in the first performance of step (d) with post-altered signal values at all cycle numbers after the candidate cycle number.

According to the iterative mode, each dataset used for selection of a candidate cycle number during repetition of the steps (a)-(f) can be distinguished by use of different terms. For example, if the steps (a)-(f) are performed two or more times, each dataset used in each repetition may be referred to as a "pre-alteration dataset", "first post-alteration dataset", "second post-alteration dataset", and the like.

The "first post-alteration dataset" refers to one in which a signal value at a cycle number determined to be indicative of an abnormal signal upon a first performance of steps (a)-(f) has been replaced with its appropriate post-alteration signal value; the "second post-alteration dataset" refers to one in which a signal value at another cycle number determined to be indicative of an abnormal signal upon a second performance of steps (a)-(f) has been further replaced with its appropriate post-alteration signal value.

The method of Section II described above may be applied to a plurality of datasets obtained by multiple PCR. For example, when a plurality of datasets are obtained by multiplex PCR, the method of Section II may be applied to each dataset to determine a cycle number indicative of an abnormal signal in each dataset as well as to correct each dataset. Specifically, when a plurality of datasets are obtained by detection at different detection temperatures from a single amplification reaction in accordance with the MuDT1 technology developed by the present inventors (see WO 2015/147412), the method of Sections II is applied to each dataset, thereby determining a cycle number indicative of an abnormal signal in each dataset as well as correcting each dataset.

In addition to the determination of a cycle number indicative of an abnormal signal, the method of Section II may be applied to correct an abnormal signal in a dataset, i.e., to provide a corrected dataset.

If the candidate cycle number is determined to be indicative of an abnormal signal in accordance with the present invention, a dataset in which the signal value at the candidate cycle number is altered may be provided as a corrected dataset (e.g., amplification curve).

When the repetition is terminated, the dataset in which the signal values at the cycle numbers determined to be indicative of an abnormal signal before the termination are all altered may be finally provided as a corrected dataset.

The corrected dataset may be used for further analysis. For example, the dataset may be used to determine the presence/absence of a target analyte.

The methods of Sections I and II may be applied to a certain region, e.g., a region of interest in a dataset, not all cycle numbers in a dataset.

In an embodiment, the method of the present invention may be applied only to the cycle numbers within the baseline region in a dataset.

In another embodiment, the method of the present invention may be applied only to the cycle numbers within the exponential region in a dataset.

In still another embodiment, the method of the present invention may be applied only to the cycle numbers within the plateau region in a dataset.

In further another embodiment, the method of the present invention may be applied only to the cycle numbers selected by the user. The cycle number to be analyzed may be selected arbitrarily. For example, the cycle numbers to be analyzed are from the 3^rd cycle number to the end cycle number, from 5^th cycle number to the end cycle number, or from 7^th cycle number to the end cycle number. Alternatively, the cycle numbers to be analyzed are mid-range cycle numbers.

The methods of Sections I and II described above may be used in combination with other methods known in the art, for example, those disclosed in U.S. Patent No. 8,560,247 and U.S. Patent Application Publication No. 2015/0186598. For example, the result of any of the methods described above may be compared with that of any conventional method known in the art, and then the commonly determined cycle number may be ultimately determined as a cycle number indicative of an abnormal signal.

The normality-representing value may encompass the normality score as described above, as well as a value obtained by multiplying two first-order change values at two cycle numbers.

In an embodiment, known methods for correcting an abnormal signal at a cycle number may be applied to the step (d) for altering a signal value at the selected cycle number.

III. Detection of abnormal signal in a dataset (without selection of candidate cycle number)

The detection of an abnormal signal is performed without selection of a candidate cycle number.

In still another aspect of the present invention, there is provided a method for detecting an abnormal signal in a dataset, comprising:

(a) obtaining a dataset for a target analyte by a signal amplification reaction; wherein the dataset comprises a plurality of data points; wherein each of the data points has a cycle number and a signal value at the cycle number; and

(b) performing the following steps (b-1) to (b-4) for all cycle numbers in the dataset:

(b-1) providing a normality-representing value at the cycle number by using the signal values; wherein the normality-representing value is a value representing the extent of normality of a signal value at a cycle number;

(b-2) altering a signal value at the cycle number, and further providing a normality-representing value at the cycle number by using the altered signal value; wherein the normality-representing value further provided at the cycle number corresponds to a post-alteration normality-representing value; and

(b-3) comparing the post-alteration normality-representing value to the pre-alteration normality-representing value at the cycle number; and

(b-4) determining the cycle number to be indicative of an abnormal signal, with a proviso that the post-alteration normality-representing value is larger than the pre-alteration normality-representing value at the candidate cycle number.

Unlike the methods of Sections I and II, the method of Section III is characterized by sequential analysis of all cycle numbers instead of analyzing the selected candidate cycle number.

The method of Section III is not limited to the order of analysis of the cycle numbers, so long as all the cycle numbers are analyzed at least once. In other words, all the cycle numbers in a dataset may be subjected to the method of Section III, e.g., sequentially, randomly or in a regular manner.

In an embodiment, the dataset obtained in the step (a) may be a dataset excluding a cycle region in which an abnormal signal is unlikely to be exhibited. For example, the dataset in the step (a) may a dataset excluding a cycle region from 1^st cycle number to 10^th cycle number, 1^st cycle number to 9^th cycle number, 1^st cycle number to 8^th cycle number, 1^st cycle number to 7^th cycle number, 1^st cycle number to 6^th cycle number, a 1^st cycle number to 5^th cycle number, 1^st cycle number to 4^th cycle number, 1^st cycle number to 3^th cycle number, 1^st cycle number to 2^nd cycle number, or a 1^st cycle number.

Since the method of Section III is similar to the methods of Sections I and II, the common descriptions between them are omitted in order to avoid undue redundancy leading to the complexity of this specification.

IV. Storage medium, Computer program and Device for Signal Extraction

Since the storage medium, the device and the computer program of the prevent invention described hereinbelow are intended to perform the present methods in a computer, the common descriptions between them are omitted in order to avoid undue redundancy leading to the complexity of this specification.

In further another aspect of this invention, there is provided a computer readable storage medium containing instructions to configure a processor to perform a method for detecting an abnormal signal in a dataset, comprising:

(a) receiving a dataset for a target analyte by a signal amplification reaction; wherein the dataset comprises a plurality of data points; wherein each of the data points has a cycle number and a signal value at the cycle number;

(f) determining the candidate cycle number(s) to be indicative of an abnormal signal, with a proviso that the post-alteration normality-representing value indicates a higher extent of normality than the pre-alteration normality-representing value at the candidate cycle number.

In still another aspect of this invention, there is provided a computer readable storage medium containing instructions to configure a processor to perform a method for detecting an abnormal signal in a dataset, the method comprising:

(b-4) determining the cycle number to be indicative of an abnormal signal, with a proviso that the post-alteration normality-representing value indicates a higher extent of normality than the pre-alteration normality-representing value at the candidate cycle number.

In still another aspect of this invention, there is provided a computer program to be stored on a computer readable storage medium to configure a processor to perform a method for detecting an abnormal signal in a dataset, the method comprising:

The program instructions are operative, when performed by the processor, to cause the processor to perform the present method described above. The program instructions for performing the present method may comprise (i) an instruction to calculate a second-order change value at each cycle number in a dataset; (ii) an instruction to calculate a normality-representing value at each cycle number by using the second-order change value; (iii) an instruction to select a candidate cycle number(s) for an abnormal signal by the normality-representing value; (iv) an instruction to alter the signal value at the candidate cycle number(s), and re-calculate a normality-representing value at the candidate cycle number(s) by using the altered signal value; (v) an instruction to compare the post-alteration normality-representing value to the pre-alteration normality-representing value at the candidate cycle number(s); and (iii) an instruction to determine whether the candidate cycle number(s) to be indicative of an abnormal signal.

The present method described above is implemented in a processor, such as a processor in a stand-alone computer, a network attached computer or a data acquisition device such as a real-time PCR machine.

The types of the computer readable storage medium include various storage medium such as CD-R, CD-ROM, DVD, flash memory, floppy disk, hard drive, portable HDD, USB, magnetic tape, MINIDISC, nonvolatile memory card, EEPROM, optical disk, optical storage medium, RAM, ROM, system memory and web server.

The signal values from the signal-generating process may be received through several mechanisms. For example, the signal values may be acquired by a processor resident in a PCR data acquiring device. The signal values may be provided to the processor in real time as the signal values are being collected, or it may be stored in a memory unit or buffer and provided to the processor after the experiment has been completed. Similarly, the signal values may be provided to a separate system such as a desktop computer system via a network connection (e.g., LAN, VPN, intranet and Internet) or direct connection (e.g., USB or other direct wired or wireless connection) to the acquiring device, or provided on a portable medium such as a CD, DVD, floppy disk, portable HDD or the like to a stand-alone computer system. Similarly, the dataset may be provided to a server system via a network connection (e.g., LAN, VPN, intranet, Internet and wireless communication network) to a client such as a notebook or a desktop computer system.

The instructions to configure the processor to perform the present invention may be included in a logic system. The instructions may be downloaded and stored in a memory module (e.g., hard drive or other memory such as a local or attached RAM or ROM), although the instructions can be provided on any software storage medium such as a portable HDD, USB, floppy disk, CD and DVD. A computer code for implementing the present invention may be implemented in a variety of coding languages such as C, C++, Java, Visual Basic, VBScript, JavaScript, Perl and XML. In addition, a variety of languages and protocols may be used in external and internal storage and transmission of data and commands according to the present invention.

In a further aspect of this invention, there is provided a device for detecting an abnormal signal in a dataset, comprising (a) a computer processor and (b) the computer readable storage medium described above coupled to the computer processor.

According to an embodiment, the device further comprises a reaction vessel to accommodate the sample and signal-generating means, a temperature controlling means to control temperatures of the reaction vessel and/or a detector to detect signals at cycle numbers.

The processor may be prepared in such a manner that a single processor can do several performances. Alternatively, the processor unit may be prepared in such a manner that several processors do the several performances, respectively.

According to an embodiment, the processor may be embodied by installing software into conventional devices for detection of target nucleic acid sequences (e.g. real-time PCR device).

The signal values may be received with amplification curves in various fashions. For example, the signal values may be received and collected by a processor in a data collector of the real-time PCR device. Upon collecting the signal values, they may be provided to a processor in a real-time manner, or stored in a memory unit or buffer and then provided to a processor after experiments.

Likely, the signal values may be provided from the real-time PCR device to the computer system such as a desktop computer system via network connection (e.g., LAN, VPN, intranet and internet) or direct connection (e.g., USB and wired or wireless direct connections), or via portable media such as CD, DVD, floppy disk and portable HDD. Alternatively, the signal values may be provided to a server system via network connections (e.g., LAN, VPN, intranet, internet and wireless communication network) connected to a client such as notebook and desktop computer systems.

As described above, the present method may be embodied by an application (i.e., program) supplier-installed or user-direct installed into the computer system, and recorded in a computer readable storage medium.

A computer program embodying the present method may implement all functions for detection of abnormal signal. The computer program may a program comprising program instructions stored on a computer readable storage medium to configure a processor to perform the present method.

The computer program may be coded by using suitable computer languages such as C, C++, JAVA, Visual basic, VBScript, JavaScript, Perl, XML and machine languages. The program codes may include function codes for mathematical functions described above and control codes for implementing process in order by a processor of the computer system.

The codes may further comprise memory reference codes by which additional information or media required in implementing the above-described functions by the processor is referred at location (address) of internal or external memory of the computer system.

When the computer system requires communication with another computer or server in remote for implementing functions of the processor, the codes may further comprise communication-relating codes encoding how the processor is communicated with another computer or server in remote by using communication module (e.g., wired and/or wireless communication module) or what information or media is transmitted.

Functional programs and codes (code segments) for embodying the present invention may be easily inferred or modified by programmers in the art in considering system environments of computers reading storage media and executing programs.

The storage medium network-connected to the computer system may be distributed and computer-readable codes may be stored and executed in a distribution manner. In such case, at least one computer among a plurality of distributed computers may implement a portion of the functions and transmit results of the implementation to at least one computer that may also implement a portion of the functions and transmit results of the implementation to at least one computer.

The storage medium in which application (i.e., program) is recorded for executing the present invention includes a storage medium (e.g., hard disk) contained in application store servers or application provider servers, application provider servers per se, another computer having the program and its storage medium.

The computer system capable of reading the storage medium may include general PC such as desk top or notebook computers, mobile terminals such as Smartphone, Tablet PC, PDA (Personal Digital Assistants) and mobile communication terminals as well as all computing-executable devices.

The present invention will now be described in further detail by examples. It would be obvious to those skilled in the art that these examples are intended to be more concretely illustrative and the scope of the present invention as set forth in the appended claims is not limited to or by the examples.

EXAMPLES

EXAMPLE 1: Detection of Abnormal Signal

According to an embodiment of the present invention, it was verified whether an abnormal signal could be detected in a dataset obtained from a real-time PCR reaction.

<1-1> Obtaining dataset by Real-time PCR reaction

Among the datasets obtained by real-time PCR reactions for analytes, a dataset suspected of having an abnormal signal by visual inspection was selected for applying one embodiment of the present invention. The real-time PCR reactions were performed on a CFX96™ Real-Time PCR Detection System (Bio-Rad Laboratories) with 45 cycles of amplification using a TaqMan probe as a signal-generating means.

The dataset selected (amplification curve) and the signal value (RFU) at each cycle number are shown in Fig. 2 and Table 1, respectively.

Table 1

<1-2> Calculating second-order difference at each cycle number

For the dataset, the first-order difference at each cycle number was calculated by subtraction of the signal values at two immediately adjacent cycle numbers. Then, the second-order difference at each cycle number was calculated by subtraction of the first-order differences at two immediately adjacent cycle numbers. The calculation of the first-order differences and the second-order differences was performed by a backward difference method.

Specifically, the second-order difference at each cycle number was calculated using the following Equations IV and V sequentially or by using the following Equation VI alone.

Equation IV

D'(x) = y(x) - y(x-1)

Equation V

D"(x) = D'(x) - D'(x-1)

Equation VI

D"(x) = y(x) - 2 * y(x-1) + y(x-2)

wherein D'(x) represents the first-order difference at the x^th cycle number; D"(x) represents the second-order difference at the x^th cycle number; y(x) is the signal value at the x^th cycle number; y(x-1) is the signal value at the (x-1)^th cycle number; and y(x-2) represents the signal value at the (x-2)^th cycle number.

<1-3> Calculating normality score (NS) at each cycle number

Afterwards, the normality score (NS) at each cycle number was calculated by multiplying the second-order differences at two immediately adjacent cycle numbers, in accordance with the following Equation VII. The normality score was used as an example of the normality-representing value.

Equation VII

NS(x) = D"(x) * D"(x+1)

wherein NS(x) represents the normality score at the x^th cycle number; D"(x) represents the second-order difference at the x^th cycle number; and D"(x+1) represents the second-order difference at the (x+1)^th cycle number.

The NS at each cycle number as calculated above is shown in Fig. 3 and Table 2.

Table 2

<1-4> Selecting a candidate cycle number and providing a pre-alteration NS

Based on the NS calculated in Example <1-3> (see Fig. 3 and Table 2), a cycle number having a minimum NS was selected as a candidate cycle number. As a result, the 10^th cycle number (NS = -20235.34; signal value = 8916.60) was selected as a candidate cycle number.

Further, the NS "-20235.34" at the candidate cycle number was adopted as a pre-alteration NS, and provided for comparison in Example <1-6>.

<1-5> Altering the signal value at the candidate cycle number and providing a post-alteration NS

The signal value at the candidate cycle number, i.e., 10^th cycle number was altered according to a predetermined signal alteration scheme using the following Equations IX and X.

Specifically, a pre-alteration first order difference (a first-order difference calculated using pre-alteration signal values) at the candidate cycle number and a pre-alteration NS at the candidate cycle number were subjected to the following Equation IX, thereby obtaining a post-alteration first-order difference.

Equation IX

wherein D'_post(c) represents a post-alteration first-order change value at the candidate cycle number; D'_pre(c) represents a pre-alteration first-order change value at the candidate cycle number; NS_pre(c) represents a pre-alteration normality score at the candidate cycle number; and k is a number of 1 or more; and c represents the candidate cycle number.

In the Example, a first-order difference was used as one of the first-order change values.

In the Example, the candidate cycle number was 10^th cycle number; a pre-alteration first-order difference at the candidate cycle number (D'_pre(10)) was 128.81; and a pre-alteration NS at the candidate cycle number (NS_pre(10)) was -20235.34.

A post-alteration first-order difference at the candidate cycle number (D'_post(10)) was calculated by Equation IX (wherein, K=1) using the D'_pre(10) and the NS_pre(10), as follows:

Afterwards, a pre-alteration signal value at the cycle number immediately before the candidate cycle number and the post-alteration first-order difference at the candidate cycle number were subjected to the following Equation X, thereby obtaining a post-alteration signal value at the candidate cycle number.

Equation X

y_post(c) = y_pre(c-1) + D'_post(c)

The post-alteration signal value at the 10^th cycle number can be calculated as follows: y_post(10) = y_pre(9) + D'_post(10).

In the Example, since the pre-alteration signal value at the 9^th cycle number (y_pre(9)) was 8787.79, and the post-alteration first-order difference at the 10^th cycle number (D'_post(10)), calculated by Equation IX, was 0.90, y_post(10) = 8787.79 + 0.90 = 8788.69.

Based on the calculation result, the signal value at the 10^th cycle number (y_pre(10)), "8916.60", was altered into the post-alteration signal value (y_post(10)), "8788.69".

Then, the NS at the candidate cycle number was re-calculated by using the altered signal value. The re-calculated NS at the candidate cycle number was "-97.10". The NS "-97.10" at the candidate cycle number was adopted as a post-alteration NS, and provided for comparison in Example <1-6>.

<1-6> Determining an abnormal signal

According to the embodiment of the present invention, the post-alteration NS at the candidate cycle number is compared to the pre-alteration NS at the candidate cycle, and if the post-alteration normality score is larger than the pre-alteration normality score at the candidate cycle number, the candidate cycle number is determined to be indicative of an abnormal signal.

In this Example, the post-alteration NS "-97.10" was larger than the pre-alteration NS "-20235" at the candidate cycle number (10^th cycle number).

Accordingly, it was determined that the 10^th cycle number is a cycle number indicative of an abnormal signal.

EXAMPLE 2: Detection of Additional Abnormal Signal by Embodiment of Present Invention (I)

We have investigated whether an additional abnormal signal can be detected using a dataset in which an abnormal signal has been altered into a normal signal. To this end, a first post-alteration dataset in which the signal value at the cycle number indicative of an abnormal signal is altered was prepared and subjected to Examples <1-1> to <1-6> instead of the dataset of Example <1-1>.

<2-1> Providing a first post-alteration dataset

For further identifying a cycle number indicative of an abnormal signal, a post-alteration dataset was used instead of the dataset of Example <1-1>.

The post-alteration dataset was obtained as follows:

First, the signal value at the 10^th cycle number (y_pre(10); "8916.60") determined to be indicative of an abnormal signal in Example <1-6> was altered into the post-alteration signal value (y_post(10); "8788.69") (see Fig. 4).

Then, signal values at all cycle numbers after the candidate cycle numbers were further altered (see Fig. 4). In this case, the signal values at the cycle numbers after the candidate cycle number were altered such that the difference between the pre-alteration signal value and the post-alteration signal value at each of the cycle numbers have the same as that at the candidate cycle number. Specifically, a post-alteration signal value at the 11^th cycle number was calculated by adding the pre-alteration first-order difference at the 11^th cycle number (D'_pre(11)) to the post-alteration signal value at the 10^th cycle number (y_post(10)). In this manner, the signal values from the 12^th cycle number to the end cycle number were altered.

The resulting dataset was referred to as a "first post-alteration dataset", which is shown in Table 3.

Table 3

The first post-alteration dataset (solid line) and the dataset before alteration of signal value (pre-alteration dataset of Example <1-1>; dotted line) are depicted in Fig. 5.

<2-2> Calculating second-order difference at each cycle number

For the first post-alteration dataset, the second order difference at each cycle number was calculated as in Example <1-2>.

<2-3> Calculating normality score (NS) at each cycle number

The NS at each cycle number was calculated by multiplying the second-order differences at two immediately adjacent cycle numbers as in Example <1-3>.

The NS at each cycle number as re-calculated above are shown in Fig. 6 and Table 4.

Table 4

<2-4> Selecting a candidate cycle number and providing a pre-alteration NS

Based on the NS re-calculated in Example <2-3> (see Fig. 6 and Table 4), a cycle number having a minimum NS was selected as a candidate cycle number. As a result, the 11^th cycle number (NS = -802.05; signal value = 8763.82) was selected as a candidate cycle number.

Further, the NS "-802.05" at the candidate cycle number was adopted as a pre-alteration NS, and provided for comparison in Example <2-6>.

<2-5> Altering the signal value at the candidate cycle number and providing a post-alteration NS

The signal value at the candidate cycle number, i.e., 11^th cycle number was altered according to a predetermined signal alteration scheme as in Example <1-5>.

As a result, a post-alteration signal value at the 11^th cycle number was calculated as follows: y_post(11) = 8788.69 + 0.85 = 8789.54.

Based on the calculation result, the signal value at the 11^th cycle number (y_pre(11)), "8763.82", was altered into the post-alteration signal value (y_post(11)), "8789.54".

Then, the NS at the candidate cycle number was re-calculated by using the altered signal value. The re-calculated NS at the candidate cycle number was "-0.27". The NS "-0.27" at the candidate cycle number was adopted as a post-alteration NS, and provided for comparison in Example <2-6>.

<2-6> Determining an abnormal signal

According to the embodiment of the present invention, the post-alteration NS at the candidate cycle number was compared to the pre-alteration NS at the candidate cycle number.

In this Example, the post-alteration NS "-0.27" was larger than the pre-alteration NS "-802.05" at the candidate cycle number (11^th cycle number).

Accordingly, it was determined that the 11^th cycle number is a cycle number indicative of an abnormal signal.

EXAMPLE 3: Detection of Additional Abnormal Signal by Embodiment of Present Invention (II)

We have investigated whether an additional abnormal signals can be detected using a dataset in which an abnormal signal has been altered into a normal signal. To this end, a second post-alteration dataset in which the signal value at the cycle number indicative of an abnormal signal is altered was prepared and subjected to Examples <1-1> to <1-6> instead of the dataset of Example <1-1>.

<3-1> Providing a second post-alteration dataset

For further identifying a cycle number indicative of an abnormal signal, a second post-alteration dataset was used instead of the dataset of Example <1-1>.

The second post-alteration dataset was obtained as follows:

First, the signal value at the 11^th cycle number (y_pre(11)), "8763.82", determined to be indicative of an abnormal signal in Example <2-6> was altered into the post-alteration signal value (y_post(11)), "8789.54", for the first post-alteration dataset.

Then, signal values at all cycle numbers after the candidate cycle numbers were further altered. In this case, the signal values at the cycle numbers after the candidate cycle number were altered such that the difference between the pre-alteration signal value and the post-alteration signal value at each of the cycle numbers have the same as that at the candidate cycle number.

The resulting dataset was referred to as a "second post-alteration dataset", which is shown in Table 5.

Table 5

The second post-alteration dataset (solid line) and the first alteration dataset (dotted line) are depicted in Fig. 7.

<3-2> Calculating second-order difference at each cycle number

For the second post-alteration dataset, the second order difference at each cycle number was calculated as in Example <1-2>.

<3-3> Calculating normality score (NS) at each cycle number

The NS at each cycle number as re-calculated above are shown in Fig. 8 and Table 6.

Table 6

<3-4> Selecting a candidate cycle number and providing a pre-alteration NS

Based on the NS re-calculated in Example <3-3> (see Fig. 8 and Table 6), a cycle number having a minimum NS was selected as a candidate cycle number. As a result, the 23^th cycle number (NS = -250.71; signal value = 9564.74) was selected as a candidate cycle number.

Further, the NS "-250.71" at the candidate cycle number was adopted as a pre-alteration NS, and provided for comparison in Example <3-6>.

<3-5> Altering the signal value at the candidate cycle number and providing a post-alteration NS

The signal value at the candidate cycle number, i.e., 23^th cycle number was altered according to a predetermined signal alteration scheme as in Example <1-5>.

As a result, a post-alteration signal value at the 23^th cycle number was calculated as follows: y_post(23) = 9383.66 + 10.76 = 9394.42

Based on the calculation result, the signal value at the 23^th cycle number (y_pre(11)), "9564.74", was altered into the post-alteration signal value (y_post(11)), "9394.42".

Then, the NS at the candidate cycle number was re-calculated by using the altered signal value. The re-calculated NS at the candidate cycle number was "-23682.71". The NS "-23682.71" at the candidate cycle number was adopted as a post-alteration NS, and provided for comparison in Example <3-6>.

<3-6> Determining an abnormal signal

In this Example, the post-alteration NS "-23682.71" was smaller than the pre-alteration NS "-250.71" at the candidate cycle number (23^th cycle number).

Accordingly, it was determined that the 23^th cycle number is a cycle number indicative of a normal signal.

According to an embodiment of the present invention, the method of the present invention may be terminated if the post-alteration NS is smaller than the pre-alteration NS at the candidate cycle number for each dataset. In this Example, a cycle number indicative of an abnormal signal was no longer determined, since the post-alteration NS was smaller than the pre-alteration NS at the candidate cycle number in Example 3.

According to Examples 1 to 3, the 10^th cycle number and the 11^th cycle number were finally determined to be cycle numbers indicative of an abnormal signal.

In addition, the dataset in which the signal values at the 10^th and 11^th cycle numbers have been replaced with respective appropriate post-alteration signal values (second post-alteration dataset) were finally provided as a corrected dataset. The finally provided corrected is shown in Fig. 11. In the figure, the dotted line represents the dataset before correction (pre-alteration dataset), and the solid line represents the finally corrected dataset (second post-alteration dataset).

As shown in Fig. 11, it was revealed that a cycle number indicative of an abnormal signal could be accurately corrected by the present method. Thus, the use of the present method makes it possible not only to determine a cycle number indicative of an abnormal signal in a dataset, but also to correct a dataset containing abnormal signal(s).

Having described a preferred embodiment of the present invention, it is to be understood that variants and modifications thereof falling within the spirit of the invention may become apparent to those skilled in this art, and the scope of this invention is to be determined by appended claims and their equivalents.

Claims

A method for detecting an abnormal signal in a dataset, comprising:

(a) obtaining a dataset for a target analyte by a signal amplification reaction; wherein the dataset comprises a plurality of data points; wherein each of the data points has a cycle number and a signal value at the cycle number;

(b) providing a normality-representing value at each cycle number of the dataset by using the signal values; wherein the normality-representing value is a value representing the extent of normality of a signal value at a cycle number;

(c) selecting a candidate cycle number(s) for an abnormal signal by the normality-representing value; wherein the normality-representing value at the candidate cycle number(s) corresponds to a pre-alteration normality-representing value;

(d) altering the signal value at the candidate cycle number(s), and further providing a normality-representing value at the candidate cycle number(s) by using the altered signal value; wherein the normality-representing value further provided at the candidate cycle number(s) corresponds to a post-alteration normality-representing value;

(e) comparing the post-alteration normality-representing value to the pre-alteration normality-representing value at the candidate cycle number(s); and

(f) determining the candidate cycle number(s) to be indicative of an abnormal signal, with a proviso that the post-alteration normality-representing value indicates a higher extent of normality than the pre-alteration normality-representing value at the candidate cycle number.
The method of claim 1, wherein the normality-representing value at said each cycle number is provided by calculating a change value at said each cycle number from signal values at 2-5 consecutive cycle numbers comprising said each cycle number.
The method of claim 2, wherein the normality-representing value is a normality score, which is obtained by (i) calculating a second-order change value at each cycle number of the dataset; and (ii) calculating a normality score at each cycle number of the dataset by using the second-order change value; wherein the calculation of the normality score is performed by a mathematical operation that represents sign change between and magnitudes of the second-order change values at two consecutive cycle numbers.
The method of claim 3, wherein the calculation of the second-order change value is performed by calculating a first-order change value at each cycle number of the dataset by using two signal values at two consecutive cycle numbers, and then calculating a second-order change value at each cycle number of the dataset by using the two first-order change values at two consecutive cycle numbers.
The method of claim 3, wherein the second-order change value is any one selected from the group consisting of a second-order difference, a second-order difference quotient, and a second-order derivative.
The method of claim 5, wherein the second-order difference or a second-order difference quotient is obtained by a forward or backward difference method.
The method of claim 3, wherein the calculation of the normality score is performed by multiplying the second-order change values at two consecutive cycle numbers.
The method of claim 7, wherein the calculation of the normality score is performed by the following Equation VII or VIII:

Equation VII

NS(x) = D"(x) * D"(x+1)

wherein NS(x) represents a normality score at the x^th cycle number, D"(x) represents a second-order change value at the x^th cycle number, D"(x+1) represents a second-order change value at the (x+1)^th cycle number, and x is an integer of 1 or more;

Equation VIII

NS(x) = D"(x) * D"(x-1)

wherein NS(x) represents a normality score at the x^th cycle number, D"(x) represents a second-order change value at the x^th cycle number, D"(x-1) represents a second-order change value at the (x-1)^th cycle number, and x is an integer of 2 or more.
The method of claim 1, wherein where a larger normality-representing value indicates a higher extent of normality at a cycle number, the candidate cycle number is (i) a cycle number having a normality-representing value smaller than a threshold, wherein the threshold is selected from values less than 0; (ii) a cycle number having a normality-representing value which is smaller than a threshold and which is a minimum, wherein the threshold is selected from values less than 0; or (iii) a cycle number having a negative sign and a minimum normality-representing value; where a smaller normality-representing value indicates a higher extent of normality of a signal value at a cycle number, the candidate cycle number is (i) a cycle number having a normality-representing value larger than a threshold, wherein the threshold is selected from values more than 0; (ii) a cycle number having a normality-representing value which is larger than a threshold and which is a maximum, wherein the threshold is selected from values larger than 0; or (iii) a cycle number having a positive sign and a maximum normality-representing value.
The method of claim 1, wherein the signal value at the candidate cycle number is altered such that its absolute value is decreased.
The method of claim 10, wherein the signal value at the candidate cycle number is altered such that its absolute value is larger than or equal to a relatively small absolute value of signal values at two cycle numbers immediately adjacent to the candidate cycle number.
The method of claim 3, wherein the signal value at the candidate cycle number is altered into a post-alteration signal value obtained by (i) obtaining a post-alteration first-order change value at the candidate cycle number by using a pre-alteration first-order change value at the candidate cycle number and a pre-alteration normality score at the candidate cycle number; and (ii) obtaining a post-alteration signal value at the candidate cycle number by using the post-alteration first-order change value at the candidate cycle number.
The method of claim 3, wherein the signal value at the candidate cycle number is altered into a post-alteration signal value determined by Equations IX and X:

Equation IX

wherein D'_post(c) represents a post-alteration first-order change value at the candidate cycle number; D'_pre(c) represents a pre-alteration first-order change value at the candidate cycle number; NS_pre(c) represents a pre-alteration normality score at the candidate cycle number; k is a number of 1 or more; and c represents the candidate cycle number;

Equation X

y_post(c) = y_pre(c-1) + D'_post(c)

wherein y_post(c) represents a post-alteration signal value at the candidate cycle number; y_pre(c-1) represents a pre-alteration signal value at the (candidate cycle number-1); D'_post(c) represents the post-alteration first-order change value at the candidate cycle number, as determined by Equation IX.
The method of claim 1, wherein when a plurality of candidate cycle numbers are selected in the step (c), the candidate cycle numbers are sequentially or simultaneously subjected to the steps (d)-(f).
The method of claim 1, which further comprises repeating the steps of: (g) providing a post-alteration dataset after performing the step (f), wherein the post-alteration dataset is obtained by altering the signal value at the cycle number(s) indicative of an abnormal signal; and (h) performing the steps (a)-(f) using the post-alteration dataset instead of the dataset in the step (a).
The method of claim 15, wherein the post-alteration dataset is obtained by further altering signal values at all cycle numbers after the cycle number indicative of an abnormal signal.
The method of claim 16, wherein the alteration of the signal values at all cycle numbers after the cycle number indicative of an abnormal signal is performed such that the difference between the post-alteration signal value and the pre-alteration signal value at each cycle number is same to that at the cycle number indicative of an abnormal signal.
The method of claim 15, wherein the repetition of the steps (a)-(f) is terminated, if there is no cycle number determined to be indicative of an abnormal signal.
A method for detecting an abnormal signal in a dataset, comprising:

(a) obtaining a dataset for a target analyte by a signal amplification reaction; wherein the dataset comprises a plurality of data points; wherein each of the data points has a cycle number and a signal value at the cycle number; and

(b) performing the following steps (b-1) to (b-4) for all cycle numbers in the dataset:

(b-1) providing a normality-representing value at the cycle number by using the signal values; wherein the normality-representing value is a value representing the extent of normality of a signal value at a cycle number;

(b-2) altering a signal value at the cycle number, and further providing a normality-representing value; wherein the normality-representing value further provided at the cycle number corresponds to a post-alteration normality-representing value; and

(b-3) comparing the post-alteration normality-representing value to the pre-alteration normality-representing value at the cycle number; and

(b-4) determining the cycle number to be indicative of an abnormal signal, with a proviso that the post-alteration normality-representing value indicates a higher extent of normality than the pre-alteration normality-representing value at the candidate cycle number.
A computer readable storage medium containing instructions to configure a processor to perform a method for detecting an abnormal signal in a dataset, comprising:

(a) receiving a dataset for a target analyte by a signal amplification reaction; wherein the dataset comprises a plurality of data points; wherein each of the data points has a cycle number and a signal value at the cycle number;

(b) providing a normality-representing value at each cycle number of the dataset by using the signal values; wherein the normality-representing value is a value representing the extent of normality of a signal value at a cycle number;

(c) selecting a candidate cycle number(s) for an abnormal signal by the normality-representing value; wherein the normality-representing value at the candidate cycle number(s) corresponds to a pre-alteration normality-representing value;

(d) altering the signal value at the candidate cycle number(s), and further providing a normality-representing value at the candidate cycle number(s) by using the altered signal value; wherein the normality-representing value further provided at the candidate cycle number(s) corresponds to a post-alteration normality-representing value;

(e) comparing the post-alteration normality-representing value to the pre-alteration normality-representing value at the candidate cycle number(s); and

(f) determining the candidate cycle number(s) to be indicative of an abnormal signal, with a proviso that the post-alteration normality-representing value indicates a higher extent of normality than the pre-alteration normality-representing value at the candidate cycle number.
A device for detecting an abnormal signal in a dataset, comprising (a) a computer processor and (b) the computer readable storage medium of claim 20 coupled to the computer processor.
A computer program to be stored on a computer readable storage medium to configure a processor to perform a method for detecting an abnormal signal in a dataset, comprising:

(a) receiving a dataset for a target analyte by a signal amplification reaction; wherein the dataset comprises a plurality of data points; wherein each of the data points has a cycle number and a signal value at the cycle number;

(b) providing a normality-representing value at each cycle number of the dataset by using the signal values; wherein the normality-representing value is a value representing the extent of normality of a signal value at a cycle number;

(c) selecting a candidate cycle number(s) for an abnormal signal by the normality-representing value; wherein the normality-representing value at the candidate cycle number(s) corresponds to a pre-alteration normality-representing value;

(d) altering the signal value at the candidate cycle number(s), and further providing a normality-representing value at the candidate cycle number(s) by using the altered signal value; wherein the normality-representing value further provided at the candidate cycle number(s) corresponds to a post-alteration normality-representing value;

(e) comparing the post-alteration normality-representing value to the pre-alteration normality-representing value at the candidate cycle number(s); and

(f) determining the candidate cycle number(s) to be indicative of an abnormal signal, with a proviso that the post-alteration normality-representing value indicates a higher extent of normality than the pre-alteration normality-representing value at the candidate cycle number.