WO2023146212A1 - Dispositif pour différencier les cancers à l'aide d'un algorithme d'apprentissage profond - Google Patents

Dispositif pour différencier les cancers à l'aide d'un algorithme d'apprentissage profond Download PDF

Info

Publication number
WO2023146212A1
WO2023146212A1 PCT/KR2023/000931 KR2023000931W WO2023146212A1 WO 2023146212 A1 WO2023146212 A1 WO 2023146212A1 KR 2023000931 W KR2023000931 W KR 2023000931W WO 2023146212 A1 WO2023146212 A1 WO 2023146212A1
Authority
WO
WIPO (PCT)
Prior art keywords
cancer
fragment
copy number
arm
number variation
Prior art date
Application number
PCT/KR2023/000931
Other languages
English (en)
Korean (ko)
Inventor
권창혁
Original Assignee
권창혁
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 권창혁 filed Critical 권창혁
Publication of WO2023146212A1 publication Critical patent/WO2023146212A1/fr

Links

Images

Classifications

    • C—CHEMISTRY; METALLURGY
    • C12—BIOCHEMISTRY; BEER; SPIRITS; WINE; VINEGAR; MICROBIOLOGY; ENZYMOLOGY; MUTATION OR GENETIC ENGINEERING
    • C12Q—MEASURING OR TESTING PROCESSES INVOLVING ENZYMES, NUCLEIC ACIDS OR MICROORGANISMS; COMPOSITIONS OR TEST PAPERS THEREFOR; PROCESSES OF PREPARING SUCH COMPOSITIONS; CONDITION-RESPONSIVE CONTROL IN MICROBIOLOGICAL OR ENZYMOLOGICAL PROCESSES
    • C12Q1/00—Measuring or testing processes involving enzymes, nucleic acids or microorganisms; Compositions therefor; Processes of preparing such compositions
    • C12Q1/68—Measuring or testing processes involving enzymes, nucleic acids or microorganisms; Compositions therefor; Processes of preparing such compositions involving nucleic acids
    • C12Q1/6876—Nucleic acid products used in the analysis of nucleic acids, e.g. primers or probes
    • C12Q1/6883—Nucleic acid products used in the analysis of nucleic acids, e.g. primers or probes for diseases caused by alterations of genetic material
    • C12Q1/6886—Nucleic acid products used in the analysis of nucleic acids, e.g. primers or probes for diseases caused by alterations of genetic material for cancer
    • C—CHEMISTRY; METALLURGY
    • C12—BIOCHEMISTRY; BEER; SPIRITS; WINE; VINEGAR; MICROBIOLOGY; ENZYMOLOGY; MUTATION OR GENETIC ENGINEERING
    • C12Q—MEASURING OR TESTING PROCESSES INVOLVING ENZYMES, NUCLEIC ACIDS OR MICROORGANISMS; COMPOSITIONS OR TEST PAPERS THEREFOR; PROCESSES OF PREPARING SUCH COMPOSITIONS; CONDITION-RESPONSIVE CONTROL IN MICROBIOLOGICAL OR ENZYMOLOGICAL PROCESSES
    • C12Q1/00—Measuring or testing processes involving enzymes, nucleic acids or microorganisms; Compositions therefor; Processes of preparing such compositions
    • C12Q1/68—Measuring or testing processes involving enzymes, nucleic acids or microorganisms; Compositions therefor; Processes of preparing such compositions involving nucleic acids
    • C12Q1/6813—Hybridisation assays
    • C12Q1/6827—Hybridisation assays for detection of mutation or polymorphism
    • G—PHYSICS
    • G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
    • G16B—BIOINFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR GENETIC OR PROTEIN-RELATED DATA PROCESSING IN COMPUTATIONAL MOLECULAR BIOLOGY
    • G16B30/00—ICT specially adapted for sequence analysis involving nucleotides or amino acids
    • C—CHEMISTRY; METALLURGY
    • C12—BIOCHEMISTRY; BEER; SPIRITS; WINE; VINEGAR; MICROBIOLOGY; ENZYMOLOGY; MUTATION OR GENETIC ENGINEERING
    • C12Q—MEASURING OR TESTING PROCESSES INVOLVING ENZYMES, NUCLEIC ACIDS OR MICROORGANISMS; COMPOSITIONS OR TEST PAPERS THEREFOR; PROCESSES OF PREPARING SUCH COMPOSITIONS; CONDITION-RESPONSIVE CONTROL IN MICROBIOLOGICAL OR ENZYMOLOGICAL PROCESSES
    • C12Q2537/00—Reactions characterised by the reaction format or use of a specific feature
    • C12Q2537/10—Reactions characterised by the reaction format or use of a specific feature the purpose or use of
    • C12Q2537/165—Mathematical modelling, e.g. logarithm, ratio

Definitions

  • the present invention relates to a cancer detection device, and more particularly, to a cancer detection device using a deep learning algorithm.
  • Human or animal blood supplies necessary nutrients and oxygen to tissues, transports chemicals that play an important role in maintaining the functions of our body, such as hormones, or produces and transports antibodies.
  • cfDNA cell-free circulating nucleic acids
  • ctDNA cell-free circulating tumor nucleic acids
  • cfDNA cell-free DNA
  • ctDNA circulating tumor DNA
  • ultralow-pass whole-genome sequencing which is 0.1X to 0.5X ultra-low-pass whole-genome sequencing (ULP-WGS).
  • URP-WGS ultralow-pass whole-genome sequencing
  • Early cancer detection has become more important, and analysis of the size of ctDNA/ctDNA fragments along with the identification of chromosomal abnormalities using whole genomes is attracting more and more attention.
  • changes in the size of cfDNA can increase the accumulation of circulating tumor DNA, and short fragments of less than 166 bp appear in a large number and differ from the normal pattern, so more research is needed.
  • the ratio of pieces of P-arm and Q-arm may be different in specific cancers, it is time to compare these patterns or research using artificial intelligence and deep learning.
  • An object of the present invention is to provide a cancer determination device capable of determining the presence or absence of cancer and determining the origin and stage of cancer using a deep learning algorithm.
  • the cancer discrimination apparatus obtains information on the whole genome sequence from data on cell-free nucleic acids contained in blood, and calculates the distribution of nucleic acid fragment lengths from the obtained sequence information.
  • An analysis unit that extracts the indicated fragment pattern and calculates the copy number variance from the sequence information, inputs the fragment pattern and the copy number variance as input values of the provided deep learning model, and inputs the fragment pattern and the copy number variance It may include a control unit that outputs at least one of whether cancer has occurred and the type of cancer based on the mutation as an output value through the deep learning model.
  • the controller may further input at least one of a pattern different from the normal one of the fragment patterns, a difference between the fragment patterns of the P-arm and the Q-arm, and a fragment angle obtained by angling the fragment length as the input value.
  • the analysis unit extracts the mitochondrial fragment length and the mitochondrial fragment angle obtained by angling the mitochondrial fragment length, and further calculates the P-arm fragment length and the Q-arm fragment length from the sequence information. .
  • control unit may further input two or more data among the mitochondrial fragment length, the mitochondrial fragment angle, and the P-arm fragment length and Q-arm fragment length as the input values.
  • the cancer determination apparatus further includes a display unit, and the control unit displays a user interface capable of identifying a basis for determining the presence of cancer when it is determined that cancer exists based on the output value of the deep learning model. It can be output through the At this time, the control unit determines the difference between a normal case without cancer and a case with cancer in at least one of the fragment pattern and the copy number variation in the form of at least one or a combination of two or more of visual expression and auditory expression. It is possible to control the user interface to output as .
  • the analysis unit when extracting the fragment pattern, the analysis unit removes files with overlapping sequences in the obtained sequence information to organize the sequences, and from the sequence information in which the sequences are arranged, fragments of each of the autosomes and sex chromosomes
  • the length of can be extracted as the fragment pattern.
  • the length of each fragment of the autosome and sex chromosome may be extracted within a range of 75 bp to 440 bp.
  • the analysis unit may calculate the copy number variance using only fragments corresponding to regions that deviate from the average of normal samples in the sequence information where the sequences are arranged.
  • the analysis unit can sequence only reads that are perfectly matched within the sequence information for which the sequences are organized, and extract reads whose GC content and mapping rate are greater than or equal to the reference value from the arranged sequences. .
  • the analysis unit calculates the mitochondrial copy number variation by calculating the ratio of the average read depths of autosomes and mitochondria of the cell-free nucleic acid from the sequence information in which the sequences are arranged.
  • the present invention can provide an effect of determining the presence or absence of cancer using a deep learning algorithm, quickly determining the origin and stage of cancer, and taking action.
  • an artificial intelligence learning approach to biological analytes that can affect cancer, it is possible to detect cancer early and reduce mortality by predicting the interpretation of cancer, the origin and stage of cancer, as well as cancer treatment and Prognosis prediction can also be performed, which can significantly lower medical costs.
  • a method and apparatus for predicting cancer even at a depth as low as 0.5X can contribute to the happiness and welfare of mankind by providing a service available to all citizens at a low price.
  • FIG. 1 is a schematic block diagram of an apparatus for determining cancer according to the present invention.
  • FIG. 2 is a flowchart illustrating a process of determining cancer using the cancer determination apparatus according to the present invention.
  • 3 to 22 are conceptual views for explaining details of the cancer determination process of FIG. 2 .
  • the 'cancer determination device' includes all of various devices capable of providing results to the user by performing calculation processing.
  • the cancer determination device is not only a computer, a terminal, a desktop PC, and a notebook (note book), but also a smart phone, a tablet PC, a cellular phone, and a PCS phone (Personal Communication).
  • Service phone synchronous/asynchronous IMT-2000 (International Mobile Telecommunication-2000) mobile terminal, Palm PC (Palm Personal Computer), Personal Digital Assistant (PDA), etc. may also be applicable.
  • IMT-2000 International Mobile Telecommunication-2000
  • Palm PC Personal Computer
  • PDA Personal Digital Assistant
  • the cancer determination device may receive a request from a client and communicate with a server that performs information processing.
  • the cancer determination apparatus may be a mobile terminal.
  • Mobile terminals described in this specification include mobile phones, smart phones, laptop computers, digital broadcasting terminals, personal digital assistants (PDAs), portable multimedia players (PMPs), navigation devices, and slate PCs.
  • PDAs personal digital assistants
  • PMPs portable multimedia players
  • slate PCs slate PCs.
  • tablet PC ultrabook
  • wearable device eg, watch type terminal (smartwatch), glass type terminal (smart glass), HMD (head mounted display)), etc. may be included there is.
  • the cancer determination device and the computer will be used interchangeably.
  • the cancer determination method is performed by the cancer determination device or computer (or server) as a subject.
  • the deep learning algorithm or artificial intelligence model described below is characterized in that it is created through learning.
  • being made through learning means that a basic artificial intelligence model is learned using a plurality of learning data by a learning algorithm, so that a predefined action rule or artificial intelligence model set to perform a desired characteristic (or purpose) is created. means burden.
  • Such learning may be performed in the device itself in which artificial intelligence according to the present disclosure is performed, or through a separate server and/or system.
  • Examples of the learning algorithm include supervised learning, unsupervised learning, semi-supervised learning, or reinforcement learning, but are not limited to the above examples.
  • An artificial intelligence model may be composed of a plurality of neural network layers.
  • Each of the plurality of neural network layers has a plurality of weight values, and a neural network operation is performed through an operation between an operation result of a previous layer and a plurality of weight values.
  • a plurality of weights possessed by a plurality of neural network layers may be optimized by a learning result of an artificial intelligence model. For example, a plurality of weights may be updated so that a loss value or a cost value obtained from an artificial intelligence model is reduced or minimized during a learning process.
  • the artificial neural network may include a deep neural network (DNN), for example, a Convolutional Neural Network (CNN), a Deep Neural Network (DNN), a Recurrent Neural Network (RNN), a Restricted Boltzmann Machine (RBM), A deep belief network (DBN), a bidirectional recurrent deep neural network (BRDNN), or deep Q-networks, but is not limited to the above examples.
  • DNN deep neural network
  • CNN Convolutional Neural Network
  • DNN Deep Neural Network
  • RNN Recurrent Neural Network
  • RBM Restricted Boltzmann Machine
  • BBN Restricted Boltzmann Machine
  • BBN deep belief network
  • BNN bidirectional recurrent deep neural network
  • Q-networks deep Q-networks
  • the controller 130 may implement artificial intelligence.
  • Artificial intelligence refers to a machine learning method based on an artificial neural network in which a machine learns by mimicking a human's biological neuron.
  • the methodology of artificial intelligence includes supervised learning in which input data and output data are provided together as training data according to the learning method, so that the answer (output data) of the problem (input data) is determined, and only input data is provided without output data.
  • unsupervised learning where the answer (output data) of the problem (input data) is not determined, and whenever an action is taken in the current state, a reward is given in the external environment. , it can be classified as reinforcement learning in which learning proceeds in the direction of maximizing this reward.
  • the methodology of artificial intelligence may be classified according to the architecture, which is the structure of the learning model.
  • the architecture of widely used deep learning technology is Convolutional Neural Network (CNN) and Recurrent Neural Network (RNN). , Transformers, and generative adversarial networks (GANs).
  • the apparatus for determining cancer may include an artificial intelligence model.
  • the artificial intelligence model may be one artificial intelligence model or may be implemented as a plurality of artificial intelligence models.
  • Artificial intelligence models may be composed of neural networks (or artificial neural networks), and may include statistical learning algorithms that mimic biological neurons in machine learning and cognitive science.
  • a neural network may refer to an overall model having a problem-solving ability by changing synaptic coupling strength through learning of artificial neurons (nodes) formed in a network by synaptic coupling. Neurons in a neural network may contain a combination of weights or biases.
  • a neural network may include one or more layers composed of one or more neurons or nodes.
  • the apparatus for determining cancer may include an input layer, a hidden layer, and an output layer.
  • a neural network constituting the device can infer a result (output) to be predicted from an arbitrary input (input) by changing the weight of a neuron through learning.
  • the control unit 130 generates a neural network, trains or learns the neural network, performs an operation based on received input data, and generates an information signal based on the result of the operation.
  • Neural network models include GoogleNet, AlexNet, VGG Network, etc., CNN (Convolution Neural Network), R-CNN (Region with Convolution Neural Network), RPN (Region Proposal Network) ), Recurrent Neural Network (RNN), Stacking-based deep Neural Network (S-DNN), State-Space Dynamic Neural Network (S-SDNN), Deconvolution Network, Deep Belief Network (DBN), Restrcted Boltzman Machine (RBM), It may include, but is not limited to, various types of models such as Fully Convolutional Network, Long Short-Term Memory (LSTM) Network, Classification Network, etc.
  • the controller 130 may perform calculations according to the models of the neural network, ,
  • the neural network may include a deep neural network.
  • Neural networks include CNN (Convolutional Neural Network), RNN (Recurrent Neural Network), perceptron, multilayer perceptron, FF (Feed Forward), RBF (Radial Basis Network), DFF (Deep Feed Forward), LSTM (Long Short Term Memory), GRU (Gated Recurrent Unit), AE (Auto Encoder), VAE (Variational Auto Encoder), DAE (Denoising Auto Encoder), SAE (Sparse Auto Encoder), MC (Markov Chain), HN (Hopfield Network), BM(Boltzmann Machine), RBM(Restricted Boltzmann Machine), DBN(Depp Belief Network), DCN(Deep Convolutional Network), DN(Deconvolutional Network), DCIGN(Deep Convolutional Inverse Graphics Network), GAN(Generative Adversarial Network) ), LSM (Liquid State Machine), ELM (Extreme Learning Machine), ESN (Echo State Network), DRN (De
  • the control unit 130 may include a Convolution Neural Network (CNN) such as GoogleNet, AlexNet, VGG Network, etc., a Region with Convolution Neural Network (R-CNN), a Region Proposal Network (RPN), an RNN ( Recurrent Neural Network), S-DNN (Stacking-based deep Neural Network), S-SDNN (State-Space Dynamic Neural Network), Deconvolution Network, DBN (Deep Belief Network), RBM (Restructed Boltzman Machine), Fully Convolutional Network, LSTM (Long Short-Term Memory) Network, Classification Network, Generative Modeling, eXplainable AI, Continual AI, Representation Learning, AI for Material Design, BERT for natural language processing, SP-BERT, MRC/QA, Text Analysis, Dialog System, Various artificial intelligence structures and algorithms such as GPT-3, GPT-4, Visual Analytics for vision processing, Visual Understanding, Video Synthesis, ResNet data intelligence for Anomaly Detection,
  • CNN Convolution Neural Network
  • a cancer determination device capable of determining the presence or absence of cancer and determining the origin and stage of cancer using a deep learning algorithm will be described in detail.
  • FIG. 1 is a schematic block diagram of an apparatus for determining cancer according to the present invention.
  • the cancer determination apparatus 100 may include a DNA analysis unit 110, a display unit 120, and a control unit 130.
  • the components shown in FIG. 1 are not essential to implement the cancer detection device according to the present invention, so the cancer detection device described herein may have more or fewer components than the components listed above. there is.
  • the DNA analysis unit 110 obtains information on the whole genome sequence from cell-free nucleic acids contained in blood, extracts a fragment pattern representing the distribution of fragment lengths of the nucleic acid from the obtained sequence information, and extracts the obtained sequence. From this information, copy number variation can be calculated.
  • the DNA analysis unit 110 may extract a fragment length pattern of the P/Q-arm and a fragment angle obtained by angling the fragment length.
  • the DNA analysis unit 110 may extract the mitochondrial fragment length and the mitochondrial fragment angle obtained by angling the mitochondrial fragment length, and calculate the P-arm fragment length and the Q-arm fragment length from the sequence information. there is.
  • the DNA analysis unit 110 may extract mitochondrial copy number variation using the obtained sequence information.
  • the DNA analysis unit 110 may be operated and controlled by the control unit 130 .
  • the display unit 120 may output various types of information.
  • the display unit 120 may output a user interface related to cancer determination under the control of the controller 130 .
  • the display unit 120 may be implemented as a touch screen, and in this case, a user input may be received through a touch input.
  • the cancer determination apparatus 100 may further include a memory, and the memory may be electrically connected to the controller 130 .
  • the memory may store basic data for the unit, control data for controlling the operation of the unit, and input/output data.
  • the memory may be various storage devices such as ROM, RAM, EPROM, flash drive, hard drive, and the like.
  • the memory may store data such as various processes for cancer determination by the controller 130 or processes for control.
  • the controller 130 controls overall operations related to cancer determination according to the present invention.
  • the control unit 130 may provide or process appropriate information or functions to a user by processing signals, data, information, etc. input or output through the components described above or by driving an application program stored in a memory.
  • the controller 130 may directly perform an operation related to cancer determination or control an operation related to cancer determination to be performed using the data extracted by the DNA analysis unit 110 .
  • control unit 130 inputs the fragment pattern and copy number variation extracted and calculated by the DNA analysis unit 110 as input values of a deep learning model to which a deep learning algorithm is applied, whether or not cancer has occurred and the type of cancer. At least one of them may be output as an output value.
  • controller 130 may further input (use) the fragment angle as an input value.
  • control unit 130 controls the fragment pattern extracted by the DNA analysis unit 110, the difference between the fragment patterns of the P-arm and the Q-arm, copy number variation, copy number variation using normal and different fragment pattern data, and fragment length.
  • Two or more of the angled fragment angle, mitochondrial fragment length, mitochondrial fragment angle, P-arm fragment length, and Q-arm fragment length may be input to the deep learning model as input values.
  • the accuracy of determining the presence or absence of cancer and determining the type of cancer may improve.
  • the controller 130 may use the P-arm fragment length and the Q-arm fragment length to determine the type of cancer when cancer exists.
  • the controller 130 may output a user interface capable of identifying a basis for determining the presence of cancer.
  • the user interface under the control of the control unit 130, displays a difference between a normal case without cancer and a case with cancer in at least one of a copy number variation and a fragment pattern in at least one of a visual expression and an auditory expression, or Two or more are output in a combined form so that the user can distinguish between a normal case without the presence of cancer and a case with the presence of the cancer.
  • FIG. 2 is a flowchart illustrating a process of determining cancer using the cancer determination apparatus according to the present invention.
  • 3 to 22 are conceptual views for explaining details of the cancer determination process of FIG. 2 .
  • the present invention can provide a cancer discrimination device capable of discriminating not only the occurrence of cancer, but also the origin and stage of cancer by inputting fragment data of cell-free circulating tumor nucleic acid (ctDNA) into the deep learning model.
  • ctDNA cell-free circulating tumor nucleic acid
  • the present invention utilizes fragment lengths, copy number differences, and differences in the ratio of mitochondria and P/Q-arms of cell-free circulating tumor nucleic acid (ctDNA) or cell-free nucleic acid (cfDNA) data floating in plasma, serum, urine, etc. Accuracy can be greatly increased by discriminating the created complex data using multi-dimensional space using deep learning.
  • NGS Next Generation Sequencing
  • the data set used in the deep learning analysis of the present invention includes Copy Number Variations (CNV) shown in FIG. 3 (a) and fragment length shown in FIG. 3 (b) (or a fragment pattern representing the distribution of fragment lengths) may be used.
  • CNV Copy Number Variations
  • fragment length shown in FIG. 3 (b)
  • fragment pattern representing the distribution of fragment lengths
  • this method can be applied not only to distinguish the presence or absence of cancer, but also to methods and devices for predicting and discriminating the origin and stage of cancer.
  • cancer detection device described in the present invention cancer can be detected early to reduce cancer mortality, and by continuously monitoring surgical patients, recurrence can be prevented at an early stage, and medical expenses can be significantly reduced by applying it to prognosis and treatment. and reduce mortality.
  • the cancer discrimination apparatus of the present invention analyzes at least one of plasma and serum from blood and cell-free nucleic acid data extracted from urine using deep learning to detect cancer or normal cancer at an early stage. can judge
  • the DNA analyzer 110 may obtain whole genome sequencing (WGS) from cell-free nucleic acid (ctDNA) (and/or cfDNA) contained in blood (S210).
  • WGS whole genome sequencing
  • the controller 130 may control the rpm speed of the centrifuge to maximize the amount of ctDNA (or cfDNA).
  • control unit 130 a) in the separation method using only one step, extracts plasma from 400g (acceleration of gravity) to 1000g in 10 minutes, b) in the two-step separation method, after applying the method a)
  • the centrifuge can be controlled to extract plasma at a rotational speed of 6000 g or more.
  • Plasma separated in the centrifuge is moved to the DNA analysis unit 110 and can be used for analysis.
  • the DNA analysis unit 110 may extract a fragment pattern indicating a distribution of nucleic acid fragment lengths from the obtained sequence information (S220).
  • the DNA analyzer 110 may analyze a pattern of ctDNA fragment lengths in a file in which whole genome sequencing (WGS) sequences are arranged.
  • WGS whole genome sequencing
  • the DNA analysis unit 110 may perform sequencing of the whole genome sequence (WGS) file in a preset human reference database.
  • WGS whole genome sequence
  • the DNA analysis unit 110 removes files with overlapping sequences in the sequence information in which the sequences are arranged to organize the sequences, and from the sequence information in which the sequences are arranged, the lengths of fragments of each of the autosomes and sex chromosomes (Fragment pattern) can be calculated by extracting from a minimum of 75 bp to a maximum of 440 bp.
  • the DNA analyzer 110 may calculate the length of a mitochondrial fragment and separate the length of a chromosome fragment into a P-arm and a Q-arm.
  • the DNA analyzer 110 may calculate the difference between the P-arm and the Q-arm for the fragment size of the chromosome.
  • the data may include all data of numerical values and images.
  • control unit 130 independently determines the stage or origin of cancer using deep learning or expands samples to a large number of data through generative adversarial networks (GANs), using deep learning It is also possible to independently determine the stage or origin of cancer.
  • GANs generative adversarial networks
  • the DNA analysis unit 110 may calculate copy number variation (CNV) from sequence information (S230).
  • the DNA analyzer 110 may calculate the copy number variation (copy number) of ctDNA using the sequence information in which sequences of whole genome sequencing (WGS) are organized.
  • WGS whole genome sequencing
  • the DNA analysis unit 110 may calculate the copy number variation using only fragments corresponding to a region that deviate from the average of the normal sample in the sequence information in which whole genome sequencing (WGS) sequences are arranged.
  • WGS whole genome sequencing
  • the DNA analysis unit 110 calculates the difference in copy number variation between autosomes and sex chromosomes using the sequence information for which the sequences are arranged, and the control unit 130 can diagnose cancer through this. .
  • the DNA analysis unit 110 removes ambiguous reads from the sequenced sequence information, sequenced only perfectly matched reads, and in the sorted sequence, GC content and mapping rate (Mappability) Leads above the threshold can be extracted.
  • the DNA analysis unit 110 may calculate a Z-score by dividing each region into bin regions of 3 kb or more (3 kb to 10 MB).
  • the DNA analyzer 110 may determine the average value of the normal group for each region, determine the average value of the cancer sample or the copy number variation of the sample to be known, and generate a data type including all numerical and image data.
  • control unit 130 independently determines the stage or origin of cancer using deep learning or expands samples to a large number of data through generative adversarial networks (GANs), using deep learning It is also possible to independently determine the stage or origin of cancer.
  • GANs generative adversarial networks
  • the DNA analyzer 110 may extract mitochondrial fragment lengths, mitochondrial fragment length patterns, and mitochondrial fragment angles obtained by angling the fragment lengths.
  • the DNA analysis unit 110 may calculate the mitochondrial sequence depth (or fragment length) of ctDNA from the sequenced file of whole genome sequencing (WGS).
  • the DNA analyzer 110 may calculate a ratio of average read depths of autosomes and mitochondria using the sequenced data, and calculate mitochondrial copy number variation.
  • the control unit 130 may input the fragment pattern and copy number variation as input values of the deep learning model, and output at least one of whether cancer has occurred and the type of cancer as an output value (S240).
  • the controller 130 may diagnose cancer by using a DNA fragment length pattern and copy number variation of data generated by a mass sequencing method.
  • the deep learning model includes Random Forest (RF), Support Vector Machine (SVC), eXtra Gradient Boost (XGB), Decision Tree (DC), K-nearest Neighbors (KNN), Gaussian Naive Bayes (GNB), Stochastic Gradient Descent At least one algorithm of (SGD), Linear Discriminant Analysis (LDA), Ridge, Lasso, and Elastic net may be applied.
  • RF Random Forest
  • SVC Support Vector Machine
  • eXtra Gradient Boost XGB
  • DC Decision Tree
  • KNN K-nearest Neighbors
  • GNB Gaussian Naive Bayes
  • SGD Stochastic Gradient Descent At least one algorithm of
  • LDA Linear Discriminant Analysis
  • Ridge Lasso
  • Elastic net may be applied.
  • the control unit 130 performs deep learning on copy number variation, copy number variation using only fragments out of the normal fragment length pattern, fragment length pattern of the entire data, fragment ratio of P/Q-arm, and mitochondrial read depth data. can be identified using
  • the control unit 130 separates copy number variation, chromosome and mitochondrial fragment lengths, P-arm and Q-arm fragment lengths, P/Q-arm ratio, and mitochondrial read depth data using deep learning, respectively. can be identified.
  • control unit 130 provides data combining two or more of copy number variation, chromosome and mitochondrial fragment lengths, P-arm and Q-arm fragment length separation, P/Q-arm ratio, and mitochondrial read depth. may be determined using deep learning.
  • the control unit 130 is a convolutional neural network (CNN)-based algorithm such as Inception, ResNet, NasNet, Efficient, MobileNet, VGG, and a deep artificial neural network (DNN)-based, recurrent neural network (RNN), and machine
  • CNN convolutional neural network
  • DNN deep artificial neural network
  • RNN recurrent neural network
  • SVM random forest, KNN, gradient, boosting
  • controller 130 may determine the presence or absence of cancer using variations of autoencoders (Variational Autoencoder, Convolutional Autoencoder, Vanilla Autoencoder, Stacked Autoencoder, and Denoising Autoencoder).
  • autoencoders Variational Autoencoder, Convolutional Autoencoder, Vanilla Autoencoder, Stacked Autoencoder, and Denoising Autoencoder.
  • control unit 130 inflates the number of samples using variants of GAN (Convolutional GAN, Wasserstein GAN, Deep Convolutional GAN, CycleGAN, LSGAN), which are generative hostile neural networks, for four data, and uses deep learning to It is also possible to perform discrimination.
  • GAN Convolutional GAN, Wasserstein GAN, Deep Convolutional GAN, CycleGAN, LSGAN
  • the controller 130 may inflate a sample by several tens to hundreds of times using various sample expansion methods, such as SMOTE and denoising autoencoder, to determine cancer with an artificial intelligence and deep learning classifier.
  • sample expansion methods such as SMOTE and denoising autoencoder
  • the data or data sets described in this specification may include all data types such as numbers and images.
  • the controller 130 may mark the CNV data that comes out as a value by using regions significantly occurring in existing cancers, and may provide a graph of fragment sizes of normal subjects and actual samples in one report.
  • the controller 130 may determine the origin of cancer and the stage of cancer to find the most relevant cancer by using each data type and composite data. To this end, the controller 130 compares cancers of normal people and individuals to determine each cancer, compares normals and all cancers to determine at least two cancers, or compares normals to all cancers to determine the importance of each cancer. , or by comparing each stage of cancer, the stage can be determined using deep learning.
  • the present invention is a deep learning model used to discriminate recurrence and metastasis of a patient after surgery, and can continuously discriminate cancer using cell-free nucleic acid data.
  • the present invention can detect most cancers, including gastric cancer, breast cancer, colon cancer, pancreatic cancer, lung cancer, liver cancer, colorectal cancer, thyroid cancer, and the like, by using deep learning on composite data.
  • chromosomal differences can be determined using composite data such as fragment size differences and CNV information in non-invasive prenatal diagnosis using deep learning.
  • the present invention can detect most cancers, including mammary tumors, lymphomas, osteosarcomas, mast cell cancers, and the like, in animals, especially companion dogs, puppies, and cows, by using deep learning on composite data.
  • Figure 4(a) shows a read depth of 5X
  • Figure 4(b) shows a read depth of 2.5X
  • Figure 4(c) shows a read depth of 0.5X
  • (d) shows a read depth of 0.1X.
  • the data structure using CNV data, fragment length (size) pattern, mitochondria (MitoFrag) and fragment angulation (FragmentAng) used in deep learning in the cancer detection device is as follows.
  • the DNA analysis unit 110 of the present invention a) acquires ctDNA sequence information (fastq-type files or reads) from the separated plasma, and b) sequences the sequence information into a human reference genome c) removing ambiguous leads from the sequenced data, arranging only perfectly matched leads, and extracting leads with GC content and mapping rate (Mappability) greater than or equal to the reference value, d ) Calculate the Z-score by dividing each region into bin regions of 3 kb or more (3 kb ⁇ 10 MB), e) determine the average value of the normal group for each region, and use the copy number variation of the cancer sample or the sample to know The same graph can be extracted.
  • Figure 5 (a) shows the copy number variation of colorectal cancer
  • Figure 5 (b) shows the copy number variation of breast cancer
  • Figure 5 (c) shows the copy number variation of normal samples without cancer. indicates
  • the DNA analysis unit 110 uses plasma to a) sequence the whole genome sequence (WGS) file in a human reference database, b) remove redundancy of the sequenced data, and c) sequence
  • the length of the autosomal fragment is extracted and calculated from a minimum of 75 bp to a maximum of 440 bp using the data from which redundancy has been removed, and using this, a distribution graph of normal samples and patient samples can be used for analysis.
  • Figure 6 (a) shows the difference in fragment length between cancer (3 CGCRC) and normal (3 CGPLH, 610) samples
  • the graph shown in Figure 6 (b) shows normal 250 samples (620)
  • the lower right graph is a fragment length pattern graph of 250 normal samples (620) and 53 breast cancer samples (630).
  • the graph of FIG. 20 is generated by selecting only the data in the area where the corresponding graph line is clear.
  • the DNA analysis unit 110 indicates the fragment length (eg, the horizontal axis indicates the fragment length, and the vertical axis indicates the specific gravity occupied (distributed) by each fragment length).
  • R164 means a fragment having a length of 164, which can be understood to account for a specific gravity of 0.16% in the normal case.
  • the abscissa axis represents the length of the fragment, and the vertical axis represents the specific gravity occupied (distributed) by the length of each fragment.
  • R164 means a fragment having a length of 164, which can be understood to account for a specific gravity of 0.16% in the normal case.
  • the DNA analyzer 110 may indicate the length (size) of fragments mapped to mitochondria for each distribution as shown in FIG. 7 .
  • FIG. 7 is an example of mitochondria (MitoFrag)
  • FIG. 7 (a) shows the distribution of fragment lengths of normal samples (CGPLH36, CGPLH36)
  • FIG. 7 (b) and (c) show breast cancer samples (CGPLBR24) represents the distribution of fragment lengths of
  • the ability to distinguish the fragment pattern (fragment graph) of FIG. 6 may not be high. Accordingly, the DNA analysis unit 110 may angle the fragment length.
  • the DNA analysis unit 110 may calculate the fragment angle by setting the maximum length of the fragment (mostly 166 bp) to 3, dividing it by a conversion value proportional to the maximum value of 3, and then multiplying by 360.
  • FIG. 8 is a graph depicting fragment angles, and it can be seen that the discrimination power is improved compared to a simple fragment graph (fragment pattern).
  • FIG. 8 is an example of FragmentAng
  • FIG. 8 (a) shows the fragment angle of normal samples (CGPLH36, CGPLH36)
  • FIG. 8 (b) and (c) show the fragment angle of breast cancer sample (CGPLBR24). .
  • FIG. 9 illustrates various embodiments of a deep learning algorithm applied to a deep learning model used for cancer discrimination. This is the model structure finally determined using EffcientNet-B5, which showed the best results.
  • the DNN model and the CNN+DNN model can use a basic model that receives 73 feature values as an input and has 3 layers as 32 outputs.
  • the controller 130 may use fragment length (slice pattern) data.
  • the present invention can compare the accuracy of artificial intelligence algorithms using fragmentation data.
  • FIG. 10 various artificial intelligence algorithms can be used, and as shown in FIG. 10 (a), the accuracy of each algorithm can be compared based on the line 1010.
  • CNV copy number variation
  • P/Q-arm P/Q-arm
  • mitochondria mitochondria
  • FragmentAng and P/Qarm showed almost similar values by separating the same values by region, and the prediction accuracy of the CNV value was much higher than the value obtained by processing the fragment size.
  • MitoAng represents Mitochondria Angle
  • MitoFrag represents Mitochondria Fragmentation
  • FragmentAng represents Fragmentation Angle
  • CNV Copy number variation
  • a fragment pattern described herein may mean a fragment length or fragment size.
  • the controller 130 can classify the arm type according to the value of the P-arm and the Q-arm, since the arm with the abnormality is different depending on the type of arm.
  • the control unit 130 receives data of at least two of the fragment pattern, copy number variation, fragment angle obtained by angling the fragment length, mitochondrial fragment length, mitochondrial fragment angle, P-arm fragment length, and Q-arm fragment length. It can be input as an input value to the deep learning algorithm, and the input data set may be as follows.
  • CNV_Fragment is an accuracy value using EfficientNetB5 with CNV image and Fragmentation size image as input.
  • CNV_MitoFrag is an accuracy value using EfficientNetB5 with CNV image and fragmentation size image of mitochondria as input.
  • CNV_PQarm is an accuracy value using EfficientNetB5, a graph drawn by separating the CNV image and the fragmentation size patterns of P-arm and Q-arm.
  • CNV_PQarm showed the highest median value, and a number of cancer samples that could not be detected by the CNV graph alone were identified by the difference in fragment size of PQarm or P/Qarm.
  • the areas of the P-arm and Q-arm are in Table 2 below.
  • the controller 130 receives data of at least three of the fragment pattern, copy number variation, fragment angle obtained by angling the fragment length, mitochondrial fragment length, mitochondrial fragment angle, P-arm fragment length, and Q-arm fragment length. It can be input as an input value to the deep learning algorithm.
  • FIG. 14 (b) shows an example of gastric cancer (ST32) distinguishable only from the CNV (top) graph (CGPLH is a normal sample), and FIG. 15 (b) shows gastric cancer (ST39) distinguishable only from the fragment (bottom) graph. ) case (CGPLH is a normal sample).
  • the cancer discrimination apparatus of the present invention can classify cancers that cannot be distinguished by CNV as fragments and classify cancers that cannot be distinguished by fragments as CNV by analyzing a combination of CNV and fragments. .
  • Table 1 shows the accuracy of cancer-free based on deep learning
  • Table 2 shows the regions of the P-arm and Q-arm selected around the centrosome.
  • Trisomy 21 and Trisomy 18 have different patterns from normal samples (FIG. 19), as shown in FIGS. 17 and 18, and in samples No. 6 and 8 of FIG. 17, only Trisomy 21 is abnormal and chromosome 18 is normal, but the fragmentation size The patterns are almost identical.
  • FIG. 17 shows the patterns of chr18 and chr21 (1710 is normal) of samples Nos. 6 and 8, which is Trisomy 21, and FIG. 18 shows the patterns of chr18 and chr21 (1810 is normal) of samples No. 9 and 10, which is Trisomy 18.
  • indicates 19 shows normal samples (CGPLH is a normal normal person, 11 is a graph of chr18, 21 of a normal mother).
  • chromosome 18 and chromosome 21 of the patient with Trisomy 18 are clearly different from the normal pattern, but the two chromosomes show an almost similar pattern.
  • FIG. 20(a) shows a total copy number graph using all data of 2 samples of breast cancer and 2 samples of pancreatic cancer, and FIG. represents
  • Fig. 21 is a graph in which fragment patterns are drawn by dividing data of P-arm and Q-arm. In certain cancers, mutations or CNV occur only in the gene of the P-arm region, and the stability of the surrounding chromosomes is broken, so the phenomenon of difference in the graph between cancer rather than normal and terminal cancer rather than early cancer was used.
  • FIG. 21(b) is consistent with each other (normal)
  • FIG. 21 (a) is graphs showing differences (colon cancer and lung cancer).
  • Figure 22 (a) shows the ROC curve using only one factor
  • Figure 22 (b) shows the ROC curve using two factors
  • Figure 22 (c) shows the ROC curve using three factors 22(d) shows a graph using 4 to 5 factors.
  • CNV copy number variation
  • CNVaf optimally selected copy number variation
  • fragment P/Q-arm pattern PQarm or P/Qarm
  • entire fragment pattern Fragment
  • mitochondria mitochondria
  • an object of the present invention is to provide a method for detecting and analyzing circulating tumor DNA in blood, and to provide a device for detecting circulating tumor DNA.
  • the present invention uses the fragment size pattern of ctDNA, difference in P/Q-arm fragment pattern, total copy number, selected copy number, mitochondrial expression amount, fragment size and ratio of P/Q-arm, By combining two or more data as well as the data, cancer and normal can be distinguished with an accuracy of 95.6% and an AUC of 97.
  • the present invention may provide a computer readable medium containing instructions configured to be executed by a processor capable of diagnosing the origin and stage of cancer and detecting circulating tumor DNA.
  • an artificial intelligence learning approach to biological analytes that can affect cancer, it is possible to detect cancer early and reduce mortality by interpreting cancer, predicting the origin and stage of cancer, as well as cancer treatment and prognosis. It can also perform forecasting, which can significantly lower medical costs.
  • the present invention can contribute to the happiness and welfare of mankind by providing a service available to the entire population at a low price with a method and apparatus for predicting cancer even at a depth as low as 0.5X.
  • the operation and function of the cancer determination device described above may be similarly or identically applied to the cancer determination method.
  • the method according to an embodiment of the present invention described above may be implemented as a program (or application) to be executed in combination with a server, which is hardware, and stored in a medium.
  • the aforementioned program is C, C++, JAVA, machine language, etc. It may include a code coded in a computer language of. These codes may include functional codes related to functions defining necessary functions for executing the methods, and include control codes related to execution procedures necessary for the processor of the computer to execute the functions according to a predetermined procedure. can do. In addition, these codes may further include memory reference related codes for which location (address address) of the computer's internal or external memory should be referenced for additional information or media required for the computer's processor to execute the functions. there is. In addition, when the processor of the computer needs to communicate with any other remote computer or server in order to execute the functions, the code uses the computer's communication module to determine how to communicate with any other remote computer or server. It may further include communication-related codes for whether to communicate, what kind of information or media to transmit/receive during communication, and the like.
  • the storage medium is not a medium that stores data for a short moment, such as a register, cache, or memory, but a medium that stores data semi-permanently and is readable by a device.
  • examples of the storage medium include ROM, RAM, CD-ROM, magnetic tape, floppy disk, optical data storage device, etc., but are not limited thereto. That is, the program may be stored in various recording media on various servers accessible by the computer or various recording media on the user's computer.
  • the medium may be distributed to computer systems connected by a network, and computer readable codes may be stored in a distributed manner.
  • Steps of a method or algorithm described in connection with an embodiment of the present invention may be implemented directly in hardware, implemented in a software module executed by hardware, or implemented by a combination thereof.
  • a software module may include random access memory (RAM), read only memory (ROM), erasable programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), flash memory, hard disk, removable disk, CD-ROM, or It may reside in any form of computer readable recording medium well known in the art to which the present invention pertains.

Landscapes

  • Chemical & Material Sciences (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Health & Medical Sciences (AREA)
  • Proteomics, Peptides & Aminoacids (AREA)
  • Organic Chemistry (AREA)
  • Engineering & Computer Science (AREA)
  • Zoology (AREA)
  • Analytical Chemistry (AREA)
  • Physics & Mathematics (AREA)
  • Wood Science & Technology (AREA)
  • Genetics & Genomics (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Biophysics (AREA)
  • Immunology (AREA)
  • Biotechnology (AREA)
  • General Health & Medical Sciences (AREA)
  • Microbiology (AREA)
  • Molecular Biology (AREA)
  • Pathology (AREA)
  • Biochemistry (AREA)
  • General Engineering & Computer Science (AREA)
  • Oncology (AREA)
  • Hospice & Palliative Care (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Evolutionary Biology (AREA)
  • Medical Informatics (AREA)
  • Spectroscopy & Molecular Physics (AREA)
  • Theoretical Computer Science (AREA)
  • Investigating Or Analysing Biological Materials (AREA)

Abstract

La présente divulgation concerne un dispositif de différenciation du cancer à l'aide d'un algorithme d'apprentissage profond, le dispositif comprenant : une unité d'analyse pour obtenir des informations sur une séquence du génome entier à partir de données sur un acide nucléique acellulaire inclus dans le sang, extraire, à partir des informations de séquence obtenues, un modèle de fragment indiquant la distribution de la longueur des fragments d'acide nucléique, et calculer la variation du nombre de copies à partir de l'information sur les séquences; et une unité de commande pour saisir le motif de fragment et la variation du nombre de copies, en tant que valeurs en entrée, dans un modèle d'apprentissage profond préalablement présenté, et produire en sortie au moins l'un des éléments suivants : l'existence ou non d'un cancer et le type d'un cancer, en tant que valeur en sortie, au moyen du modèle d'apprentissage profond, en se fondant sur le motif de fragment et la variation du nombre de copies.
PCT/KR2023/000931 2022-01-26 2023-01-19 Dispositif pour différencier les cancers à l'aide d'un algorithme d'apprentissage profond WO2023146212A1 (fr)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
KR1020220011334A KR20230114952A (ko) 2022-01-26 2022-01-26 암 판별 장치 및 암 판별 방법
KR10-2022-0011334 2022-01-26

Publications (1)

Publication Number Publication Date
WO2023146212A1 true WO2023146212A1 (fr) 2023-08-03

Family

ID=87471885

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/KR2023/000931 WO2023146212A1 (fr) 2022-01-26 2023-01-19 Dispositif pour différencier les cancers à l'aide d'un algorithme d'apprentissage profond

Country Status (2)

Country Link
KR (1) KR20230114952A (fr)
WO (1) WO2023146212A1 (fr)

Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2016094853A1 (fr) * 2014-12-12 2016-06-16 Verinata Health, Inc. Utilisation de la taille de fragments d'adn acellulaire pour déterminer les variations du nombre de copies
KR20170000744A (ko) * 2015-06-24 2017-01-03 사회복지법인 삼성생명공익재단 유전자의 복제수 변이(cnv)를 분석하는 방법 및 장치
KR20170106372A (ko) * 2015-01-13 2017-09-20 더 차이니즈 유니버시티 오브 홍콩 암 검출을 위한 혈장 dna의 크기 및 수 비정상의 이용 방법
KR20180123020A (ko) * 2016-02-03 2018-11-14 베리나타 헬스, 인코포레이티드 카피수 변이를 판정하기 위한 dna 단편 크기의 사용
KR20190036494A (ko) * 2017-09-27 2019-04-04 이화여자대학교 산학협력단 Dna 복제수 변이 기반의 암 종 예측 방법
KR20210067931A (ko) * 2019-11-29 2021-06-08 주식회사 녹십자지놈 인공지능 기반 염색체 이상 검출 방법

Family Cites Families (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
KR101473341B1 (ko) 2014-02-18 2014-12-16 연세대학교 산학협력단 유전자 클래스 결정 방법

Patent Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2016094853A1 (fr) * 2014-12-12 2016-06-16 Verinata Health, Inc. Utilisation de la taille de fragments d'adn acellulaire pour déterminer les variations du nombre de copies
KR20170106372A (ko) * 2015-01-13 2017-09-20 더 차이니즈 유니버시티 오브 홍콩 암 검출을 위한 혈장 dna의 크기 및 수 비정상의 이용 방법
KR20170000744A (ko) * 2015-06-24 2017-01-03 사회복지법인 삼성생명공익재단 유전자의 복제수 변이(cnv)를 분석하는 방법 및 장치
KR20180123020A (ko) * 2016-02-03 2018-11-14 베리나타 헬스, 인코포레이티드 카피수 변이를 판정하기 위한 dna 단편 크기의 사용
KR20190036494A (ko) * 2017-09-27 2019-04-04 이화여자대학교 산학협력단 Dna 복제수 변이 기반의 암 종 예측 방법
KR20210067931A (ko) * 2019-11-29 2021-06-08 주식회사 녹십자지놈 인공지능 기반 염색체 이상 검출 방법

Also Published As

Publication number Publication date
KR20230114952A (ko) 2023-08-02

Similar Documents

Publication Publication Date Title
Hou et al. DeepSF: deep convolutional neural network for mapping protein sequences to folds
Han et al. Advanced applications of RNA sequencing and challenges
Nardone et al. Bioinformatics for the'bench biologist': how to find regulatory regions in genomic DNA
WO2021125744A1 (fr) Procédé et système pour fournir des informations d'interprétation sur des données de pathomique
WO2017116135A1 (fr) Système et procédé d'analyse du génotype au moyen d'informations de variations génétiques sur le génome d'un individu
WO2017135496A1 (fr) Procédé et dispositif destinés à analyser la relation entre drogue et protéine
WO2017116123A1 (fr) Système d'identification de cause d'une maladie au moyen d'informations de variation génétique concernant le génome d'un individu
WO2022005090A1 (fr) Méthode et appareil de fourniture de résultat de diagnostic
Kaufman et al. Gene expression of Caenorhabditis elegans neurons carries information on their synaptic connectivity
WO2022124724A1 (fr) Procédé de prédiction de pronostic basé sur l'apprentissage automatique et dispositif associé
WO2017116139A1 (fr) Système d'analyse de variation bioactive au moyen d'informations de variation génétique concernant le génome d'un individu
Rosen et al. Toward universal cell embeddings: integrating single-cell RNA-seq datasets across species with SATURN
WO2023146212A1 (fr) Dispositif pour différencier les cancers à l'aide d'un algorithme d'apprentissage profond
Zhang et al. A survey on protein–DNA-binding sites in computational biology
CA3200206A1 (fr) Procede de diagnostic du cancer et de prediction du type de cancer base sur l'intelligence artificielle
Athaya et al. Multimodal deep learning approaches for single-cell multi-omics data integration
WO2023167448A1 (fr) Procédé et appareil d'analyse d'image de lame pathologique
WO2020032561A2 (fr) Système et procédé de diagnostic de maladie utilisant de multiples modèles de couleurs et un réseau neuronal
WO2021107471A1 (fr) Procédé de récupération de données médicales
DaSilva et al. DNA-Diffusion: Leveraging Generative Models for Controlling Chromatin Accessibility and Gene Expression via Synthetic Regulatory Elements
WO2022010168A1 (fr) Procédé de prédiction du risque de démence chez un patient atteint de la maladie de parkinson et dispositif de prédiction de risque de démence
WO2023158080A1 (fr) Procédé et dispositif d'analyse de types de complexe majeur d'histocompatibilité (cmh)
WO2024019326A1 (fr) Dispositif et procédé pour prédire une quantité d'acide nucléique sur la base d'une image pathologique et proposer un test génomique optimal
WO2024025136A1 (fr) Appareil et procédé de proposition d'un procédé d'analyse de biomarqueur optimal basé sur une analyse multi-informations
WO2024085660A1 (fr) Dispositif et procédé de prédiction du risque d'incidence d'une maladie

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 23747254

Country of ref document: EP

Kind code of ref document: A1