Learning models, information processing devices, information processing methods, and methods for generating learning models.

Abstract

DEPCT6618 / 04 / 2566 To provide a program or similar entity that extracts mutations that are important to... Automated genome-based therapy. The program then allows the computer to process the search for training data, which Within it is genomic information obtained by reading the base sequences included in the submission and mutations. Genetic information based on submitted samples is recorded and linked for all genetic tests. This has been done many times in the past and has given rise to learning models (53) for output. Estimates related to genetic mutations are based on submitted genome data. This is obtained by reading the base sequence included in the submitted sample and inputting it through genome data assignment. Let's use genetic mutations as input and define them as output;

Description

Program, learning model, information processing device, information processing method, and learning model generation method

[0001] The present invention relates to a program, a learning model, an information processing device, an information processing method, and a method for generating a learning model.

[0002] Pathological tests, genetic tests, etc. are performed using specimens collected from patients by biopsy, blood sampling, surgery, etc. In genetic tests, a genome analyzer that visualizes the base sequence of nucleic acids read using a sequencer has been proposed (Patent Document 1).

[0003] International Publication No. 2016-175330

[0004] It is known that the effectiveness of anticancer drugs can vary greatly depending on the mutation status of the base sequence. For clinicians in charge of treating cancer patients, information on mutations that contributes to determining treatment strategies is important.

[0005] However, the genome analysis device disclosed in Patent Document 1 cannot automatically extract clinically important mutations.

[0006] The program acquires training data that records genome data obtained by reading the base sequences contained in samples and genetic mutations associated with the samples for multiple genetic tests conducted in the past, and causes the computer to execute a process of generating a learning model that, when genome data obtained by reading the base sequences contained in the samples is input and the genetic mutations are output, outputs a prediction regarding genetic mutations based on the samples.

[0007] In one aspect, an object of the present invention is to provide a program or the like that automatically extracts clinically important mutations based on the base sequence read from a sample.

[0008] 1 is an explanatory diagram illustrating a processing flow using a genome analysis system. FIG. 1 is an explanatory diagram illustrating a method for generating a learning model. FIG. 1 is an explanatory diagram illustrating an overview of an integrated DB. FIG. 1 is an explanatory diagram illustrating an overview of genome data. FIG. 2 is an explanatory diagram illustrating a configuration of a genome analysis system. FIG. 2 is an explanatory diagram illustrating a record layout of a teacher data DB. FIG. 3 is an explanatory diagram illustrating a record layout of an integrated DB. FIG. 4 is an explanatory diagram illustrating a record layout of a report DB. FIG. 5 is an explanatory diagram illustrating a learning model. FIG. 6 is an explanatory diagram illustrating an example of a report. FIG. 7 is an explanatory diagram illustrating an example of a comment field. FIG. 8 is an explanatory diagram illustrating an example of a comment field. FIG. 9 is an explanatory diagram illustrating an example of a nonsynonymous somatic mutation field. FIG. 10 is an explanatory diagram illustrating an example of a germline mutation field. FIG. 11 is an explanatory diagram illustrating an example of an analysis field. FIG. 12 is a flowchart illustrating the processing flow of a program. FIG. 13 is an explanatory diagram illustrating an example of an RNA field. FIG. 14 is an explanatory diagram illustrating an example of an RNA field. FIG. 15 is an explanatory diagram illustrating the record layout of a change history DB. FIG. 16 is an explanatory diagram illustrating the record layout of a report DB according to a third embodiment. FIG. 17 is a flowchart illustrating the processing flow of a program for outputting an additional report. FIG. 18 is an explanatory diagram illustrating the record layout of an expert DB. FIG. 19 is an explanatory diagram illustrating an example of a screen for selecting participants to a expert panel. FIG. 20 is an explanatory diagram illustrating an example of a screen for confirming a request to participate in an expert panel. FIG. 10 is a flowchart explaining the processing flow of a correction reception subroutine in embodiment 4. FIG. 11 is an explanatory diagram explaining an example of an integrated DB review participation request screen. FIG. 12 is a flowchart explaining the processing flow of a program for updating the integrated DB 52. FIG. 13 is a functional block diagram of an information processing device at a stage of predicting clinically significant gene mutations from genome data. FIG. 14 is a functional block diagram of an information processing device at a stage of creating a report based on gene mutations and the integrated DB 52. FIG. 15 is an explanatory diagram explaining the configuration of a genome analysis system in embodiment 7.

[0009] 1 is an explanatory diagram illustrating the flow of processing using a genome analysis system 10. A genome refers to the entire genetic information of one individual, in this case, one human.

[0010] A sample is collected from a patient. It is desirable to collect samples from both the tumor and normal tissue. The tumor sample is collected by biopsy or surgery of the lesion. In the following explanation, a sample collected from the tumor will be referred to as a tumor sample. Except for patients with blood abnormalities such as blood cancer, normal tissue samples are often collected by blood sampling. In the case of a patient with blood cancer, a tumor sample is collected from the blood, and normal tissue samples are collected from other normal tissues.

[0011] Nucleic acids, i.e., DNA (Deoxyribonucleic Acid) or RNA (Ribonucleic Acid), are extracted from each specimen. In the following explanation, the case where DNA is extracted will be described as an example. The base sequence of the DNA is read by a reader 31, and genome data is created. Details of the genome data will be described later. In the following explanation, the reader 31 will be described as a next-generation sequencer as an example, but the reader 31 may also be a DNA microarray or any other device or equipment that reads base sequences.

[0012] Genomic data is input into a learning model 53. A prediction of clinically significant gene mutations is output from the learning model 53. A draft report is automatically created based on the output gene mutations and an integrated database (DB) 52 that integrates information collected from medical literature and the like. Details of the learning model 53 and the integrated DB 52 will be described later.

[0013] Note that predictions of genetic mutations may be output from the learning model 53 regardless of whether they have clinical significance. In this case, clinically significant mutations are extracted based on the genetic mutations output from the learning model 53 and the integrated DB 52, and a draft report is automatically created.

[0014] An expert panel consisting of experts such as oncologists and geneticists reviews the draft report and makes revisions as necessary to complete the report. The clinician in charge of the patient's treatment determines the treatment plan based on the report. Details of the draft report and the report will be described later. Note that review by the expert panel does not necessarily have to be performed. In this case, the clinician determines the treatment plan based on the draft report output from the integrated DB 52.

[0015] 2 is an explanatory diagram illustrating a method for generating the learning model 53. A pathological examination is performed using a tumorous specimen. A portion containing tumor cells is excised from the tumorous specimen. DNA from the tumorous specimen is extracted from the excised specimen. DNA from the normal specimen is extracted from the normal specimen. The DNA from the normal specimen and the DNA from the tumorous specimen are input into a reading device 31, and genome data is created.

[0016] Based on the results of pathological examinations, genomic data, and other test values, experts will determine whether the tumor is benign or malignant, whether it is a primary cancer, the amount of tumor in the tumor sample, and which drugs may be effective, and create diagnostic data.

[0017] The genome data and the diagnostic data are associated and recorded in a teacher data DB 51 (see FIG. 5 ). Details of the teacher data DB 51 will be described later. Supervised machine learning is performed based on the teacher data DB 51 to generate a learning model 53. The learning model 53 is a trained model that outputs a prediction regarding a genetic mutation associated with a sample when genome data obtained by reading the base sequence contained in the sample is input.

[0018] 3 is an explanatory diagram illustrating an overview of the integrated DB 52. The integrated DB 52 is a DB that integrates medical information on gene mutations acquired from multiple information sources by associating the medical information with the source of the medical information. The information sources are various medical information DBs 58, such as DBs that publish medical papers, DBs that publish information on clinical trials of drugs or treatments by countries or research institutions, and DBs that accumulate public information such as medical press releases issued by companies or universities.

[0019] The medical information DB 58 may be a DB that is publicly available free of charge or a DB that is publicly available for a fee. When a DB that is publicly available for a fee is used, a license process is carried out, such as concluding an appropriate license agreement between the provider of the paid DB and the provider of the integrated DB 52.

[0020] Each medical information DB 58 stores medical information in a different format and updates the information at different times. The integrated DB 52 is created by crawling each medical information DB 58, collecting information, and creating a database.

[0021] Crawling is performed as needed to create an updated integrated DB 52. Each integrated DB 52 is version-managed in a manner that allows the update date or update date and time to be determined. The integrated DB 52 will be described in detail later.

[0022] Each integrated DB 52 may be configured to record the difference from the previous version or the difference from any version, so that an integrated DB 52 at any point in time can be constructed as needed. By recording the difference, the recording capacity of the integrated DB 52 can be saved.

[0023] FIG. 4 is an explanatory diagram outlining genome data. Pre-processing is performed on the specimen. Specifically, DNA is extracted from the specimen as described above. The extracted DNA is then subjected to processes such as purification, fragmentation, and amplification. Through fragmentation, the DNA is cut into fragments of a length suitable for reading by the reader 31 used in the subsequent process.

[0024] The reader 31 sequentially reads the base sequence of each fragmented DNA. Information about the base sequence read from one DNA fragment is called a read. The read also records a quality score that indicates the reliability of the reading for each base.

[0025] Each read is mapped to a reference sequence, such as the Japanese Reference Genome (JRG) or the International Human Genome Reference Sequence, and the mapping results are recorded in a file, for example, in BAM, SAM, or CRAM format.

[0026] The mapping results and differences from the reference sequence, that is, information on the positions of mutations in the genome of the specimen relative to the reference sequence and the details of the mutations, are recorded in a file in, for example, VCF format or BCF format.

[0027] VCF files contain many mutations of little clinical significance, such as intron mutations that do not encode genetic information and synonymous mutations that do not result in changes to the encoded amino acids. Therefore, highly specialized knowledge is required to read information from VCF files that can be used to determine treatment strategies, etc.

[0028] When a FASTQ format file and a reference sequence are provided, they can be converted into BAM format, SAM format, CRAM format, or VCF format files using known analytical techniques. The data in the FASTQ format, BAM format, SAM format, CRAM format, VCF format, and BCF format described above are collectively referred to as genome data. The genome data may be in any format other than the formats exemplified here.

[0029] For example, the reading device 31 outputs a file in FASTQ format, and an analysis device (not shown) converts it into a file in BAM format and VCF format. The reading device 31 may have a built-in analysis device and directly output files in BAM format and VCF format. Alternatively, the information processing device 20 (see FIG. 5 ), which will be described later, may acquire a file in FASTQ format or BAM format and convert it into a VCF format.

[0030] When CNA (Copy Number Alteration) analysis is performed, genome data obtained from multiple normal tissue samples collected from a patient is compared with genome data obtained from a tumor tissue sample.

[0031] The CNA analysis may be performed using a panel of normals (PON) technique. When using PON, genome data is created and stored, for example, in BAM or SAM format, for normal tissue samples collected from multiple individuals. Analysis is performed by comparing the genome data obtained from tumor tissue samples collected from patients with the stored genome data.

[0032] 5 is an explanatory diagram illustrating the configuration of genome analysis system 10. Genome analysis system 10 includes information processing device 20, reading device 31, and data server 32.

[0033] The information processing device 20 includes a control unit 21, a main memory device 22, an auxiliary memory device 23, a communication unit 24, and a bus. The control unit 21 is an arithmetic and control device that executes the program of this embodiment. The control unit 21 is configured with one or more central processing units (CPUs), multi-core CPUs, graphics processing units (GPUs), or the like. The control unit 21 is connected to each hardware unit that configures the information processing device 20 via the bus.

[0034] The main memory device 22 is a storage device such as an SRAM (Static Random Access Memory), a DRAM (Dynamic Random Access Memory), a flash memory, etc. The main memory device 22 temporarily stores information required during processing performed by the control unit 21 and programs currently being executed by the control unit 21.

[0035] The auxiliary storage device 23 is a storage device such as an SRAM, a flash memory, or a hard disk. The auxiliary storage device 23 stores the teacher data DB 51, the integrated DB 52, the learning model 53, the draft report DB 55, the report DB 56, the programs to be executed by the control unit 21, and various data required for executing the programs. The teacher data DB 51, the integrated DB 52, the learning model 53, the draft report DB 55, and the report DB 56 may be stored in an external large-capacity storage device connected to the information processing device 20, a data server 32, or the like.

[0036] The communication unit 24 is an interface that performs communication between the information processing device 20 and a network.

[0037] As described above, the reading device 31 is any device or equipment that reads base sequences, such as a next-generation sequencer, a DNA microarray, or the like. Genomic data created based on the base sequences read by the reading device 31 is recorded in the data server 32. The control unit 21 can acquire the genome data recorded in the data server 32 via the communication unit 24 and a network. Note that the control unit 21 may acquire the genome data directly from the reading device 31 without going through the data server 32.

[0038] The information processing device 20 of this embodiment is a general-purpose personal computer, tablet, mainframe, or virtual machine running on a mainframe. The information processing device 20 may be configured with hardware such as multiple personal computers, tablets, or mainframes. The information processing device 20 may be configured with a quantum computer. The information processing device 20 may be integrated with the reading device 31. The information processing device 20 may be realized by so-called cloud computing.

[0039] 6 is an explanatory diagram illustrating the record layout of the teacher data DB 51. The teacher data DB 51 is a DB that records genome data and diagnostic data in association with each other. FIG. 6 shows one record of the teacher data DB 51.

[0040] The teacher data DB 51 has a specimen field, a genome data field, and a diagnosis data field. The specimen field has a normal specimen field and a tumor specimen field. The genome data field has a normal genome field and a tumor genome field. Note that the teacher data DB 51 does not necessarily have to have a normal genome field.

[0041] The diagnostic data field includes a nonsynonymous somatic mutation field, a germline mutation field, and a tumor content field. The nonsynonymous somatic mutation field includes a gene field and a DNA mutation field. The germline mutation field includes a gene field and a DNA mutation field. The training data DB 51 includes one record for each set of training data. Note that the diagnostic data field does not necessarily have to include a tumor content field.

[0042] The normal specimen field records the site where the normal specimen was collected. The tumor specimen field records the site where the tumor specimen was collected. The normal genome field records the file name of the genome data obtained from the normal specimen. The tumor genome field records the file name of the genome data obtained from the tumor specimen.

[0043] The subfields of the nonsynonymous somatic mutation field record nonsynonymous somatic mutations contained in the tumor genome, i.e., genes with somatic mutations that cause changes in amino acids encoded in the DNA base sequence, and the details of the mutations. Somatic mutations refer to mutations that do not occur in the normal genome but occur in the tumor genome. In other words, nonsynonymous somatic mutations are mutations related to the characteristics of the tumor.

[0044] For example, the first line of the nonsynonymous somatic mutation field in Figure 6 indicates that the 5164th base of the ARID1A (AT-rich interactive domain 1A) gene is mutated from C (cytosine) to T (thymine). Similarly, the second line indicates that the 743rd base of the TP53 gene is mutated from G (guanine) to A (adenine).

[0045] The subfields of the germline mutation field record the genes with mutations contained in the normal genome and the details of the mutations. For example, the first line of the germline mutation field in Figure 6 indicates that the 1791st base of the BRAF gene is mutated from T to G.

[0046] The nonsynonymous somatic mutation field and germline mutation field record any number of genes that need to be recorded in the training data among the gene mutations detected from the specimen.

[0047] In some cases, instead of collecting normal samples to obtain genome data, a reference sequence such as the reference genome sequence of Japanese people is used. In this case, the results regarding germline mutations are estimated results.

[0048] The diagnostic data field may include a synonymous somatic mutation field for recording synonymous somatic mutations, or may include a somatic mutation field instead of a nonsynonymous somatic mutation field for recording both synonymous and nonsynonymous somatic mutations.

[0049] The tumor content field records the tumor content of the sample collected from the tumor site. The tumor content is calculated, for example, based on the number of heterozygous SNPs (Single Nucleotide Polymorphisms). The tumor content may also be calculated based on the allele frequency recorded in the BAM file or SAM file, or the allele frequency calculated from the data recorded in the BAM file or SAM file.

[0050] The tumor content may be calculated based on the ratio of the number of nucleated cells observed in a pathological examination to the number of tumor cells, or based on the area occupied by tumor cells in a microscopic field of view. The definition of tumor content is arbitrary, but it is desirable that a consistent definition be used for all of the training data included in the training data DB 51.

[0051] 7 is an explanatory diagram illustrating the record layout of the integrated DB 52. The integrated DB 52 is a DB that integrates medical information on genetic mutations acquired from multiple information sources in association with the sources from which the medical information was acquired. The integrated DB 52 has a version field, a genome mutation field, and a knowledge data field.

[0052] The version field records the version of the integrated DB 52. In this embodiment, the integrated DB 52 is managed by update date. The genome mutation field has a specimen field, a gene field, and a mutation content field. The knowledge data field has a carcinogenicity field, a clinical significance field, a corresponding drug field, a corresponding disease field, a level field, and a basis information field. The integrated DB 52 has one record for each piece of medical information related to a gene mutation.

[0053] The specimen field records the site where the specimen was collected. The gene field records the gene in which the mutation was detected. In a record that records medical information related to a combination of multiple mutations, multiple genes are recorded in the gene field.

[0054] The mutation content field records the content of the mutation, such as a non-synonymous somatic mutation or a germline mutation. Note that information about synonymous somatic mutations that do not result in a change in the encoded amino acid may also be recorded in the integrated DB 52.

[0055] The carcinogenicity field records the level of carcinogenicity of the genomic mutation. The clinical significance field records the clinical significance of the genomic mutation. The knowledge data field may have only one of the carcinogenicity field and the clinical significance field.

[0056] The corresponding drug field records a drug that is effective when administered to a patient with a genomic mutation. The corresponding drug field may also record a drug currently undergoing clinical trials. The corresponding disease field records a disease corresponding to the genomic mutation. The level field records the level of importance of the genomic mutation. The evidence information field records information for accessing the evidence information, such as literature that is the basis for the information described in the record, database name, or an ID (identifier) ​​uniquely assigned to the information.

[0057] In each subfield of the knowledge data field, "-" means that there is no corresponding information.

[0058] 8 is an explanatory diagram illustrating the record layout of the report DB 56. The report DB 56 is a DB that records information about samples and diagnostic data based on the samples in association with each other. FIG. 8 shows one record of the report DB 56.

[0059] The report DB 56 has a specimen ID field, a specimen field, a genome data field, an integrated DB Ver. field, a diagnosis data field, and an expert ID field. The specimen field has a normal specimen field and a tumor specimen field. The genome data field has a normal genome field and a tumor genome field.

[0060] The diagnosis data field includes a nonsynonymous somatic mutation field, a germline mutation field, and a tumor content field. The nonsynonymous somatic mutation field includes a diagnosis data field and a knowledge data field. The diagnosis data field includes a gene field and a DNA mutation field. The knowledge data field includes a carcinogenicity field, a clinical significance field, a corresponding drug field, a corresponding disease field, a level field, and a basis information field.

[0061] The germline mutation field has a diagnosis data field and a knowledge data field. The diagnosis data field has a gene field and a DNA mutation field. The knowledge data field has a clinical significance field, a level field, and a basis information field. The report DB 56 has one record for one set of specimens.

[0062] The specimen ID field records a specimen ID that is uniquely assigned to a set of specimens. The specimen ID is linked to a patient in cooperation with an electronic medical record system or the like. The normal specimen field records the site from which the normal specimen was collected. The tumor specimen field records the site from which the tumor specimen was collected. The normal genome field records the file name of the genome data obtained from the normal specimen. The tumor genome field records the file name of the genome data obtained from the tumor specimen. The integrated DB Ver. field records the version of the integrated DB 52 used when creating the report record.

[0063] The subfields of the diagnostic data field in the nonsynonymous somatic mutation field record the genes with nonsynonymous somatic mutations and the details of the mutations. Each subfield of the knowledge data field records medical information related to the genetic mutation recorded in the diagnostic data field. The information recorded in each subfield is the same as the information recorded in the subfield of the same name in the integrated DB 52 described using Figure 7, so further explanation will be omitted.

[0064] The subfields of the diagnostic data field in the germline mutation field record the genes with germline mutations and the details of the mutations. Each subfield of the knowledge data field records medical information related to the gene mutation recorded in the diagnostic data field. The information recorded in each subfield is the same as the information recorded in the subfield of the same name in the integrated DB 52 described using Figure 7, so further explanation is omitted.

[0065] The expert ID field records an expert ID uniquely assigned to each expert who constitutes the expert panel that reviewed the draft report automatically created by the control unit 21 using a program described below. One expert ID may be assigned to an expert group that includes multiple experts.

[0066] The record layout of the report draft DB 55 is the same as the record layout of the report DB 56 explained using FIG. 8 except that it does not have an expert ID field, so illustration and detailed explanation will be omitted.

[0067] Fig. 9 is an explanatory diagram illustrating the learning model 53. The learning model 53 is a neural network including an input layer 531, an intermediate layer 532, and an output layer 533. Fig. 9 illustrates an example in which the learning model 53 is a CNN. Note that the convolutional layer and the pooling layer are not shown.

[0068] The inputs to the learning model 53 are genomic data of the tumor, genomic data of the normal, the site from which the tumor sample was taken, and the site from which the normal sample was taken. The genomic data is, for example, a tensor of piled-up alignment information, and includes components such as base sequence, strand information, base quality, and map quality. The base sequence may be expressed as the count of each of A, T, G, and C bases. The data input to the learning model 53 is input to the input layer 531 via a repetition of convolutional layers and pooling layers (not shown).

[0069] The output of the learning model 53 is, for example, the probability of each item of diagnostic data. Specifically, it is the probability that each clinically significant mutation occurs and the probability that the tumor content is a predetermined value. For example, in Figure 9, the top output node outputs the probability that a somatic mutation has occurred in which the 6952nd base of the BRCA gene has mutated from C to T, and the second output node outputs the probability that a germline mutation has occurred in which the 6952nd base of the BRCA gene has mutated from C to T.

[0070] Since somatic cells contain alleles, the somatic cells of the specimen contain the "6952nd base of the BRCA gene" derived from the father and the "6952nd base of the BRCA gene" derived from the mother. Therefore, somatic cell mutations include cases where both the paternal and maternal genes are mutated, where only the paternal gene is mutated, and where only the maternal gene is mutated.

[0071] For example, the output of learning model 53 may be scores of HomoRef, Hetero, and HomoAlt, which are indices used in variant callers for genome analysis such as deepvariant.

[0072] The bottom output node in Figure 9 outputs the probability of a tumor content of 10 percent. The output nodes include nodes that output the probability of any tumor content, for example, in increments of 10 percent.

[0073] When genome data and a specimen collection site are input to the input layer 531, the learning model 53 outputs the probability that each clinically significant mutation has occurred and that a predetermined tumor content is present to the output layer 533. In the learning stage, the control unit 21 performs supervised machine learning by calculating parameters of the intermediate layer 532 using a backpropagation method or the like, using a teacher data DB 51 that records the genome data and the specimen collection site in association with diagnostic data regarding the presence or absence of clinically significant mutations and tumor content.

[0074] Supervised machine learning can be performed using any method, such as logistic regression, SVM (Support Vector Machine), random forest, CNN, RNN, or XGBoost (eXtreme Gradient Boosting).

[0075] The learning model 53 may be generated using any computer. The generated learning model 53 is transmitted to the information processing device 20 via a network or the like and recorded in the auxiliary storage device 23. Semi-supervised learning may be used instead of supervised learning.

[0076] 10 is an explanatory diagram illustrating an example of a report 60. The report 60 is created by formatting the information recorded in the records of the report DB 56 and the information recorded in the electronic medical record into a format that is easy for the user to view. The report 60 includes a bibliographic information section 61, a comment section 62, a nonsynonymous somatic mutation section 63, a germline mutation section 64, and an analysis section 65.

[0077] The bibliographic information field 61 includes an ID field 611, a patient information field 612, a specimen field 613, a pathological tissue diagnosis field 614, and a specimen number field 615. The ID field 611 displays a patient ID uniquely assigned to the patient. The patient information field 612 displays the patient's gender and age. Note that the patient information field 612 does not necessarily have to be displayed.

[0078] The normal and tumor specimens used in the genome analysis are displayed in the specimen column 613. In Fig. 10, "FFPE (Formalin Fixed Paraffin Embedded) Lung" means formalin-fixed, paraffin-embedded lung tissue.

[0079] The pathological tissue diagnosis field 614 displays findings from a pathological diagnosis obtained by observing the specimen under a microscope. The specimen number field 615 displays a specimen number uniquely assigned to the specimen. The information displayed in the bibliographic information field 61 is obtained from the electronic medical record system using the specimen ID of the report record described with reference to FIG. 8 as a key.

[0080] 11A, 11B, and 11C are explanatory diagrams illustrating examples of the comment field 62. FIGS. 11A to 11C each show an example of the comment field 62 displayed in a different report. FIG. 11A shows the comment field 62 in a "Pathologic" report, i.e., a report on a specimen in which a germline mutation that is definitely pathogenic was found. The gene and mutation location in which the pathogenic germline mutation occurred, the reasons for this, and advice on future actions regarding the germline mutation are displayed.

[0081] Figure 11B shows an example of a report comment for a sample with low tumor content, i.e., a possible problem with the quality of the tumor sample. Figure 11C shows an example of a comment for a sample in which a cancerous mutation was found in the tumor sample. Information about the gene in which the cancerous somatic mutation occurred and clinical trials related to that gene is displayed.

[0082] The text displayed in the comment field 62 is created by combining standard phrases using a known method based on the information recorded in the diagnosis field of the report DB 56. By selecting and displaying standard phrases related to gene mutations with high pathogenicity or carcinogenicity among multiple gene mutations occurring in the specimen, even clinicians with little knowledge of genetic testing can quickly grasp highly important information.

[0083] 12 is an explanatory diagram illustrating an example of the nonsynonymous somatic mutation column 63. In FIG. 12, an example of the nonsynonymous somatic mutation column 63 is shown, which is displayed based on the nonsynonymous somatic mutation field in the report record illustrated in FIG. 8.

[0084] The nonsynonymous somatic mutation field 63 includes a gene field 631, a cytoband field 632, a DNA mutation field 633, an amino acid mutation field 634, an allele frequency field 635, and a knowledge data field 636. The gene field 631, the DNA mutation field 633, and the knowledge data field 636 each display information recorded in the nonsynonymous somatic mutation field.

[0085] The cytoband field 632 displays the location of the gene on the chromosome. The amino acid mutation field 634 displays amino acid mutations resulting from DNA mutations. The allele frequency field 635 displays, for example, allele frequencies recorded in a BAM file or a SAM file, or allele frequencies calculated from data recorded in a BAM file or a SAM file.

[0086] The total number of somatic mutations and the total somatic mutation frequency, including somatic mutations not listed in the nonsynonymous somatic mutation column 63, are displayed above the nonsynonymous somatic mutation column 63. The total number of somatic mutations and the total somatic mutation frequency can be obtained from a VCF format file.

[0087] 13 is an explanatory diagram illustrating an example of the germ cell mutation column 64. In FIG. 13, an example of the germ cell mutation column 64 is shown that is displayed based on the germ cell mutation field in the report record illustrated in FIG. 8.

[0088] The germ cell mutation field 64 includes a gene field 641, a cytoband field 642, a DNA mutation field 643, an amino acid mutation field 644, a normal site allele frequency field 647, a tumor site allele frequency field 648, and a knowledge data field 645. The gene field 641, the DNA mutation field 643, and the knowledge data field 645 each display information recorded in the germ cell mutation field.

[0089] The cytoband field 642 displays the location of the gene on the chromosome. The amino acid mutation field 644 records amino acid mutations caused by DNA mutations. The normal portion allele frequency field 647 displays the allele frequency of the normal portion recorded in a BAM format or SAM format file, for example. The tumor portion allele frequency field 648 displays the allele frequency of the tumor portion recorded in a BAM format or SAM format file, for example.

[0090] 14 is an explanatory diagram illustrating an example of the analysis field 65. The analysis field 65 includes an estimated tumor content field 651 and a mutation frequency correlation coefficient field 652. The estimated tumor content field 651 displays the estimated tumor content based on the output of the learning model 53.

[0091] The mutation frequency correlation coefficient column 652 displays the correlation coefficient between the gene mutation frequency in a sample collected from a normal area and the gene mutation frequency in a sample collected from a tumor area. When the correlation coefficient is high, the same base is likely to be mutated in the normal area and the abnormal area, and the samples are determined to be from the same patient. When the correlation coefficient is lower than the threshold, sample mix-up or contamination is suspected.

[0092] It is not necessary to display the mutation frequency correlation coefficient column 652. For example, when analysis is performed without using normal tissue samples, the mutation frequency correlation coefficient column 652 is unnecessary.

[0093] When the user selects each of the fields described using Figures 10 to 14 by, for example, right-clicking, the control unit 21 displays the information recorded in the basis information field of the report record. The control unit 21 may display a link to the basis information based on the basis information field, or may display the basis information itself. The user can confirm the reliability of the report 60 by viewing the basis described in the report 60.

[0094] The report 60 may also display contact information for the expert panel that conducted the review. The user can ask questions or ask for advice to the expert panel based on the report 60.

[0095] The report may include information such as the pretreatment performed on the sample, the number of reads of the base sequence obtained by the reader 31, or the mapping depth to the reference sequence, etc. A clinician familiar with genetic testing can judge the reliability of the report based on this information.

[0096] 15 is a flowchart illustrating the flow of program processing. The control unit 21 acquires genome data from the data server 32 based on a report creation request (step S501). The control unit 21 creates a new record in the report draft DB 55 and records data in the specimen ID field, specimen field, and genome data field (step S502).

[0097] The control unit 21 inputs the acquired genome data into the learning model 53 and acquires the predicted probability of each node in the output layer 533 (step S503). The control unit 21 extracts genetic mutations for which a probability equal to or greater than a predetermined threshold is output from the nodes related to the genetic mutations in the output layer 533 (step S504). The threshold may be a different value for each genetic mutation or a fixed value.

[0098] The control unit 21 determines the tumor content in the specimen based on the node with the highest probability among the nodes related to tumor content in the output layer 533 (step S505). The control unit 21 records the mutation extracted in step S504 in the diagnostic data field of the nonsynonymous somatic mutation field or germline mutation field of the report draft record created in step S502, and records the tumor content determined in step S505 in the tumor content field (step S506).

[0099] The tumor content may be calculated by a program separate from the program shown in Figure 15. In this case, step S505 is not necessary.

[0100] The control unit 21 searches the integrated DB 52 using the sample collection site and gene mutation recorded in the draft report record as keys, and acquires knowledge data from the knowledge data field of the extracted record (step S507).The control unit 21 records the acquired knowledge data in the report record (step S508).

[0101] The control unit 21 determines whether processing of all gene mutations recorded in the draft report record has been completed (step S509). If it is determined that processing has not been completed (NO in step S509), the control unit 21 returns to step S507. If it is determined that processing has been completed (YES in step S509), the control unit 21 creates a draft of the report 60 described using FIG. 10 based on the report record and records it in the auxiliary storage device 23 or the data server 32 (step S510).

[0102] The experts who are members of the expert panel review the draft of the report 60 at expert meetings held regularly or irregularly, and revise it as necessary. The expert meetings may be held by having the experts gather in one room, or by video conference or telephone conference. The expert meetings may also be held electronically using a chat system or the like.

[0103] The expert panel may refer to genomic data in FASTQ, BAM, VCF, or other formats as needed. The expert panel may also refer to micrographs taken during the pathology examination. The expert panel may collect information from the pathologist who performed the pathology examination or the clinician in charge of the patient.

[0104] The control unit 21 accepts the revisions decided at the expert meeting (step S511). The control unit 21 records a report record in which the information recorded in the draft report record has been revised in the report DB 56 (step S512). The control unit 21 records the expert ID uniquely assigned to the expert who performed the review in the expert ID field of the report record. The control unit 21 then ends the process.

[0105] The control unit 21 may notify the clinician that a report has been created by email or any other means. The control unit 21 may upload the report to an electronic medical record system. The control unit 21 may notify the clinician that a new report is available when the clinician logs in to the genome analysis system 10.

[0106] 15, the control unit 21 may accept a specification of the date of the integrated DB 52 for creating the report 60. If a specification of a date is accepted, the control unit 21 acquires knowledge data using the latest integrated DB 52 on the specified date in step S507. In step S510, the control unit 21 records a draft report based on the latest information on the specified date.

[0107] For example, when verifying the validity of a treatment plan decided in the past, a draft report based on the latest information on that date can be created by specifying the date on which the medical procedure was performed and running the program described using Figure 15.

[0108] Data may be added to the training data DB 51 based on the information recorded in the report DB 56, post-treatment information, post-medication information, etc., and the training model 53 may be retrained. By adding data that has been reviewed by an expert to the training data, the accuracy of the training model 53 can be improved.

[0109] This embodiment provides a learning model 53 that automatically extracts clinically important mutations based on the base sequence read from a sample. By using the learning model 53, even a doctor who does not have advanced specialized knowledge of genetic testing can determine the presence or absence of clinically important gene mutations.

[0110] According to this embodiment, a genome analysis system 10 can be provided that uses the integrated DB 52 to present medical information related to gene mutations to users. Research in the field of genetic testing is fast-paced, with new findings frequently published, making it difficult for individual physicians to keep up with the latest information. Medical information is provided based on the integrated DB 52, and the evidence for that information is also presented, allowing physicians to check the evidence as needed and provide appropriate medical care to patients.

[0111] By having the expert panel review the draft report and reflecting any revisions made by the expert panel, it is possible to provide a genome analysis system 10 that creates a highly reliable report 60. By having the expert panel review the report, it is possible to create the report 60 based on new information that is not included in the training data DB 51.

[0112] If the clinician has expertise in genetic testing, they may omit the expert panel review and use the draft report as is for report 60. The patient or clinician may obtain the draft report and genomic data and seek the opinion of a specialist of their choice.

[0113] Second Embodiment This embodiment relates to a genome analysis system 10 that analyzes the base sequence of RNA in addition to DNA. Explanation of parts common to the first embodiment will be omitted.

[0114] In this embodiment, the specimen collected from the tumor is divided into three portions. One portion is used for pathological examination, one portion is used for DNA analysis, and the last portion is pre-processed to extract RNA, and the base sequence of the RNA is read by the reader 31 and analyzed in the same manner as DNA.

[0115] By analyzing RNA, information regarding genetic abnormalities expressed in tumors can be obtained. Genetic abnormalities expressed in tumors include, for example, fusion genes in which multiple DNA fragments are fused by translocation or gene rearrangement, or exon skipping, in which a portion of DNA is lost when transcribed into RNA. In the report 60 of this embodiment, an RNA column 66 displaying information obtained by analyzing RNA is displayed between, for example, a nonsynonymous somatic mutation column 63 and a germline mutation column 64.

[0116] 16A and 16B are explanatory diagrams illustrating an example of the RNA column 66. FIGS. 16A and 16B each show an example of the RNA column 66 displayed in a different report. FIG. 16A shows an example of the RNA column 66 for a specimen in which no abnormalities were found in the RNA. FIG. 16B shows an example of the RNA column 66 for a specimen in which a fusion gene and exon skipping were found.

[0117] 16B includes a gene column 661, a mutation column 667, a cytoband column 662, a number of reads column 668, and a knowledge data column 666. The gene column 661 displays the gene from which the RNA was transcribed.

[0118] The mutation column 667 displays the RNA mutation. For example, the top line of Figure 16B displays that a fusion gene between the PAX3 gene and the FOXO1 gene was detected. The bottom line of Figure 16B displays that exon 1 skipping of the MET gene was detected.

[0119] The cytoband column 662 displays the position of the gene on the chromosome. The read count column 668 displays the number and percentage of reads in which a mutation was detected among the reads read by the reader 31. The information displayed in the read count column 668 is read from a FASTQ format file. The knowledge data column 666 displays information obtained from the integrated DB 52.

[0120] According to this embodiment, it is possible to provide a genome analysis system 10 that detects abnormalities in genes expressed in tumors and displays the results in a report 60.

[0121] [Embodiment 3] This embodiment relates to genome analysis system 10 that outputs an additional report indicating changes to a previously output report 60 when integrated DB 52 is updated. Explanation of parts common to embodiment 1 will be omitted.

[0122] 17 is an explanatory diagram illustrating the record layout of the change history DB. The change history DB is a DB that records gene mutations recorded in the integrated DB 52 in association with the change dates on which the knowledge data were changed. The change history DB has a genome mutation field and a change date field.

[0123] The genome mutation field has a tumor specimen field, a gene field, and a mutation content field. The change date field has any number of subfields, such as a first change date field, a second change date field, etc. The change history DB has one record for each piece of medical information recorded in the integrated DB 52.

[0124] The tumor specimen field records the site where the specimen was collected. The gene field records the gene in which the mutation was detected. In a record that records medical information related to a combination of multiple mutations, multiple genes are recorded in the gene field.

[0125] The first change date field records the date on which the record related to the gene mutation recorded in the genome mutation field was recorded in the integrated DB 52. The second change date field and subsequent fields record the dates on which the medical information recorded in the integrated DB 52 was changed.

[0126] Fig. 18 is an explanatory diagram illustrating the record layout of the report DB 56 according to the third embodiment. The report DB 56 according to the present embodiment has a confirmation date field added to the report DB 56 according to the first embodiment described with reference to Fig. 8. The confirmation date field records the date on which the update status of the integrated DB 52 was confirmed.

[0127] 19 is a flowchart illustrating the processing flow of the program for outputting an additional report. The control unit 21 acquires a report record recorded in the report DB 56 (step S521). The control unit 21 acquires the site from which the sample was collected, recorded in the normal sample field and the tumor sample field (step S522). The control unit 21 acquires the confirmation date, recorded in the confirmation date field (step S523).

[0128] The control unit 21 acquires the genetic mutations recorded in the gene field of the nonsynonymous somatic mutation field or the germline mutation field (step S524). The control unit 21 searches the change history DB and extracts records using the sample collection site acquired in step S522 and the genetic mutation acquired in step S524 as keys. The control unit 21 compares the date recorded in the change date field of the extracted record with the confirmation date acquired in step S523, and determines whether the knowledge data has been changed since the confirmation date (step S525).

[0129] If it is determined that the knowledge data has not been changed (NO in step S525), the control unit 21 returns to step S524. If it is determined that the knowledge data has been changed (YES in step S525), the control unit 21 searches the latest integrated DB 52 and extracts records using the site from which the sample was collected obtained in step S522 and the genetic mutation obtained in step S524 as keys. The control unit 21 obtains knowledge data from the extracted records (step S526).

[0130] The control unit 21 records the knowledge data acquired in step S526 in the knowledge data field of the report record (step S527). The control unit 21 may also create a copy of the report record and record the knowledge data acquired in step S526 therein.

[0131] The control unit 21 determines whether or not the processing of all mutations recorded in the report record acquired in step S521 has been completed (step S528). If it is determined that the processing has not been completed (NO in step S528), the control unit 21 returns to step S524.

[0132] If it is determined that the process has ended (YES in step S528), the control unit 21 determines whether there is any genetic mutation for which knowledge data was determined to have been changed in step S525 (step S529). If it is determined that there is any genetic mutation (YES in step S529), the control unit 21 notifies the clinician that the report has been changed (step S530). The notification can be made by any means, such as email or messenger.

[0133] The control unit 21 notifies the expert panel in step S530, and after accepting corrections based on the review results, may notify the clinician or the hospital. If it is determined that there is no gene mutation for which knowledge data has been determined to have been changed (NO in step S529) or after step S530 is completed, the control unit 21 determines whether to end the process (step S531).

[0134] If it is determined not to end (NO in step S531), the control unit 21 returns to step S521. If it is determined to end (YES in step S531), the control unit 21 ends the process.

[0135] According to this embodiment, it is possible to provide a genome analysis system 10 that outputs an additional report when new medical information related to a previously created report is made public. Clinicians can receive additional information about drugs, clinical trials, or treatments that are expected to be effective for patients undergoing treatment and reflect this information in their treatment plans.

[0136] The control unit 21 may accept designation of a report 60 that does not require additional information. A clinician can designate that no additional report is required for a report 60, such as a report about a patient whose treatment has been completed. In step S521, the control unit 21 excludes reports that do not require additional information from the reports to be acquired, thereby avoiding the creation of unnecessary additional reports.

[0137] [Fourth Embodiment] This embodiment relates to a genome analyzing system 10 that provides incentives to experts who participate in an expert panel. Explanation of parts common to the first embodiment will be omitted.

[0138] 20 is an explanatory diagram illustrating the record layout of the expert DB. The expert DB is a DB that records expert IDs uniquely assigned to experts participating in expert panels, their fields of expertise, and their points in association with each other.

[0139] The expert DB has an expert ID field, an area of ​​expertise field, and a points field. The expert ID field records an expert ID. The area of ​​expertise field records the area of ​​expertise of the expert. The points field records the points awarded to the expert.

[0140] Experts can earn points each time they participate in an expert panel and review a draft report. Experts can exchange their accumulated points for, for example, cash vouchers, report request vouchers that can be used when requesting the creation of report 60, or learning model vouchers that can be used when requesting genetic analysis using learning model 53. Points can provide experts with an incentive to participate in the expert panel.

[0141] The points may be set, for example, to 5 points per review. The points to be awarded to each expert may be determined, for example, by a leader of the expert panel, based on the amount of comments or the content of opinions made during the expert review. The points to be awarded per review may also be determined based on the frequency of participation in the expert panel.

[0142] 21 is an explanatory diagram illustrating an example of a screen for selecting participants to an expert panel. The screen shown in FIG. 21 is displayed on an information device, such as a personal computer, tablet, or smartphone, used by a secretariat staff member of the expert panel. The information device used by the secretariat staff member is connected to an information processing device 20 via a network.

[0143] The screen for selecting participants to the expert panel includes a sample information field 74, a filter condition field 75, a search again button 76, a candidate list 77, a confirm button 78, and a request send button 79. The sample information field 74 displays information about the sample to be reviewed by the expert panel.

[0144] The narrowing down criteria field 75 displays items used when narrowing down the experts. The user can select the narrowing down criteria by selecting the check box displayed at the beginning of each item. The narrowing down criteria field 75 may also have a field for accepting free keywords. The candidate list 77 displays a list of candidates for experts to participate in the expert panel.

[0145] The user sets desired conditions using the search condition field 75 and selects the search again button 76. The set conditions are sent to the information processing device 20. The control unit 21 extracts experts that meet the set conditions and sends them to the information device used by the user.

[0146] A list of experts that meet the set conditions is displayed in the candidate list 77. The user uses the check boxes displayed on the right side of the candidate list 77 to select the experts to invite to participate in the expert panel.

[0147] If the number of experts displayed in the candidate list 77 is too large or too small, the user can change the settings in the search criteria field 75 as appropriate and perform the search again. If the user selects the confirm button 78, a list of the selected experts is displayed. If the user selects the send request button 79, the list of the selected experts is sent to the information processing device 20.

[0148] The control unit 21 associates the sample ID with the expert ID of the selected expert and stores them in the auxiliary storage device 23. The control unit 21 sends an email containing a uniform resource locator (URL) to each expert.

[0149] 22 is an explanatory diagram illustrating an example of a screen for confirming a request to participate in an expert panel, which is displayed on an information device used by an expert when the expert accesses a website indicated by a URL.

[0150] The screen for confirming a request to participate in an expert panel includes a request list 72 and a participation button 71. The request list 72 displays a list of expert panels for which experts are requested to participate. For each expert panel, information such as the sample collection site, patient information, and the medical institution that has requested the creation of the report 60 is displayed.

[0151] The expert looks at the request list 72 and selects the participation button 71 for the expert panel in which he or she wishes to participate. The control unit 21 sets up an electronic conference room in which the expert who selected the participation button 71 will participate and uploads the draft report. The participants review the report in the electronic conference room. A pre-designated leader summarizes the conclusion and closes the electronic conference room. Note that electronic conference systems have been widely used for a long time, so detailed explanation of the processing performed by the control unit 21 will be omitted.

[0152] After the electronic conference room ends, the control unit 21 awards points to the experts who participated in the expert panel. Specifically, the control unit 21 extracts records of the experts who participated in the expert panel from the expert DB and adds points to the points field.

[0153] 23 is a flowchart illustrating the processing flow of the correction reception subroutine according to the fourth embodiment. The correction reception subroutine is a subroutine that receives participation of experts in the expert panel and awards points to the participating experts. The correction reception subroutine is started in place of step S511 of the program according to the first embodiment described with reference to FIG. 15.

[0154] The control unit 21 creates an expert panel participation request screen as described using Figure 22 for each expert registered in the expert DB, and sends an email containing the URL to notify the participation request (step S541).

[0155] The control unit 21 can determine which experts to request to review which draft reports based on the fields of expertise recorded in the field of expertise field of the expert DB. For example, for expert panels related to cases in which tumor specimens were collected from the respiratory system and cases requested by the respiratory department, the control unit 21 notifies experts who have the respiratory system registered in the field of expertise field of a request to participate.

[0156] The control unit 21 may select experts registered in the expert DB for each category and notify them of the participation request. Alternatively, the control unit 21 may notify all experts registered in the expert DB of the participation request. The control unit 21 accepts participation in the expert panel by accepting selection of the participation button 71 by the expert (step S542). The control unit 21 sets up an electronic conference room in which participants are registered in each expert panel (step S543). The control unit 21 transmits access information for the electronic conference room to each participant.

[0157] The control unit 21 uploads the draft report to the electronic conference room and makes it available for viewing by the participants (step S544). Participants communicate with other participants through the electronic conference room and review the draft report.

[0158] The pre-designated leader summarizes the conclusion and performs an operation to end the electronic conference room. The control unit 21 accepts the end operation (step S545). The control unit 21 closes the electronic conference room (step S546). The control unit 21 extracts records related to the experts who participated in the expert panel from the expert DB and adds points to the points field (step S547). The control unit 21 ends the process.

[0159] According to this embodiment, it is possible to provide a genome analysis system 10 that provides incentives for participation in an expert panel. By distributing revenues obtained from learning model usage fees, report creation fees, etc. to experts in the form of points, it is possible to provide a genome analysis system 10 that makes it easy to secure experts to participate in the expert panel.

[0160] Since experts themselves can decide whether or not to participate in each expert panel, it is possible to provide a genome analysis system 10 that can gather motivated participants.Since expert reviews are conducted using an electronic conference room, it is possible to provide a genome analysis system 10 that makes it easy for even busy experts to participate in expert panels.

[0161] Fifth Embodiment This embodiment relates to a genome analyzing system 10 that requests an expert to review information recorded in integrated DB 52. Explanation of parts common to the fourth embodiment will be omitted.

[0162] 24 is an explanatory diagram illustrating an example of an integrated DB review participation request screen. The control unit 21 sends an email containing a URL to each expert. When the expert accesses the website indicated by the URL using an information device such as a personal computer or smartphone, the integrated DB review participation request screen shown in FIG. 24 is displayed on the information device.

[0163] The integrated DB review participation request screen includes a request list 73 and a participation button 71. The request list 73 displays a list of medical information for which a specialist is requested to review. For each piece of medical information, the target gene, DNA mutation, and information source are displayed. The target of the integrated DB review may also be information unrelated to a specific gene mutation, as exemplified by No. 3 in FIG. 24.

[0164] The expert can check the request list 73 to determine whether the medical information is related to drugs, diseases, or clinical trials, which are in his or her area of ​​expertise. If the expert wishes to participate in the review, he or she selects the participation button 71. The control unit 21 sets up an electronic conference room in which the experts who selected the participation button 71 can participate, and uploads the draft report. The participants review the report in the electronic conference room. A pre-designated leader summarizes the conclusions and closes the electronic conference room.

[0165] The review may be conducted by a single expert alone, in which case the electronic conference room need not be used.

[0166] The control unit 21 adds a new record to the integrated DB 52 or updates an existing record based on the review result.

[0167] 25 is a flowchart illustrating the processing flow of a program for updating the integrated DB 52. The following description will be given taking as an example a case where the information processing device 20 updates the integrated DB 52. The updating of the integrated DB 52 may be performed by an information device other than the information processing device 20.

[0168] The control unit 21 performs crawling by visiting various medical information DBs 58 to collect new medical information related to gene mutations and to create a database (step S551). The crawling is performed by a program called a crawler or robot. Since crawling has been widely performed in the past, a detailed description will be omitted.

[0169] The control unit 21 selects the medical information collected by crawling and determines whether the information is information about a genetic mutation that has already been recorded in the integrated DB 52 (step S552). If it is determined that the information is information about a genetic mutation that has been recorded in the integrated DB 52 (YES in step S552), the control unit 21 determines whether the content is the same as the information recorded in the integrated DB 52 (step S553).

[0170] If it is determined that the information is not related to a genetic mutation recorded in the integrated DB 52 (NO in step S552), or if it is determined that the content is not identical to the information recorded in the integrated DB 52 (NO in step S553), the control unit 21 records that the medical information being processed is the subject of review (step S554).

[0171] If it is determined that the contents are the same (YES in step S553), or after step S554 is completed, the control unit 21 determines whether or not the processing of the medical information collected in step S551 is completed (step S555). If it is determined that the processing is not completed (NO in step S555), the control unit 21 returns to step S552.

[0172] If it is determined that the process has ended (YES in step S555), the control unit 21 creates an integrated DB review participation request screen, as described using Figure 24, for each expert registered in the expert DB, and sends an email containing the URL to notify the participation request (step S561).

[0173] The control unit 21 accepts participation in the review by accepting selection of the participation button 71 by the expert (step S562). The control unit 21 sets up an electronic conference room in which participants are registered for each review (step S563). The control unit 21 transmits access information for the electronic conference room to each participant.

[0174] The control unit 21 uploads the medical information collected by crawling to the electronic conference room and makes it available for viewing by participants (step S564). Participants communicate with other participants through the electronic conference room and review the medical information.

[0175] A pre-designated leader summarizes the conclusion and closes the electronic conference room. The conclusion may be determined by majority vote of the participating experts. The control unit 21 accepts the closing operation (step S565). The control unit 21 closes the electronic conference room (step S566). The control unit 21 extracts records related to the experts who participated in the review from the expert DB and adds points to the points field (step S567). The control unit 21 updates the integrated DB 52 based on the review results for each medical information (step S568). The control unit 21 ends the process.

[0176] According to this embodiment, it is possible to provide a genome analysis system 10 that automatically collects information to be registered in integrated DB 52 by crawling, and then updates integrated DB 52 after having the information reviewed by experts. By utilizing crawling technology, it is possible to provide a genome analysis system 10 that appropriately reflects new medical information in integrated DB 52.

[0177] By having an expert review the collected medical information before registering it in the integrated DB 52, it is possible to provide a genome analysis system 10 that maintains the reliability of the integrated DB 52 and outputs an accurate report 60.

[0178] By distributing the revenues obtained from the learning model usage fees and report creation fees to experts in the form of points, it is possible to provide a genome analysis system 10 that makes it easy to secure experts to participate in reviews.

[0179] Since experts themselves can decide whether or not to participate in each review, it is possible to provide genome analysis system 10 that can gather motivated review participants. Because reviews are conducted using an electronic conference room, it is possible to provide genome analysis system 10 that makes it easy for even busy experts to participate in reviews.

[0180] 26 is a functional block diagram of an information processing device 20 at a stage of predicting clinically significant gene mutations from genome data. The information processing device 20 has a genome data acquisition unit 81, a genome data input unit 82, and an output unit 83.

[0181] The genome data acquisition unit 81 acquires genome data obtained by reading the base sequence contained in a sample. The genome data input unit 82 inputs the genome data acquired by the genome data acquisition unit 81 to the learning model 53, which receives the genome data and outputs predictions regarding gene mutations. The output unit 83 outputs predictions output from the learning model 53 based on the genome data input by the genome data input unit 82.

[0182] 27 is a functional block diagram of the information processing device 20 at the stage of creating a report based on genetic mutations and the integrated DB 52. The information processing device 20 has a first receiving unit 84, a first output unit 85, a second receiving unit 86, and a second output unit 87.

[0183] The first receiving unit 84 receives genetic mutations detected from a sample. The first output unit 85 outputs a report that records analysis results for the sample and the version of the integrated DB 52 in association with each other, based on the genetic mutations received by the first receiving unit 84, medical information related to the genetic mutations acquired from multiple information sources, and the acquisition date and evidence information of the medical information in an integrated DB 52 that is integrated in association with each other.

[0184] The second receiving unit 86 receives a past date, a report output request for that date, and genetic mutations detected in the sample. The second output unit 87 outputs a report that records the analysis results for the sample and the version of the integrated DB 52 in association with each other, based on the genetic mutations received by the second receiving unit 86 and the integrated DB 52 for that date.

[0185] [Seventh Embodiment] This embodiment relates to an embodiment in which a genome analyzing system 10 according to the present embodiment is realized by operating a general-purpose computer 90 in combination with a program 97. Fig. 28 is an explanatory diagram illustrating the configuration of genome analyzing system 10 according to the seventh embodiment. Explanation of parts common to the first embodiment will be omitted.

[0186] The genome analysis system 10 of this embodiment includes a computer 90 , a reading device 31 , and a data server 32 .

[0187] The computer 90 includes a control unit 21, a main memory device 22, an auxiliary memory device 23, a communication unit 24, a reading unit 29, and a bus. The computer 90 is an information device such as a general-purpose personal computer, a tablet, or a server computer.

[0188] The program 97 is recorded on a portable recording medium 96. The control unit 21 reads the program 97 via the reading unit 29 and stores it in the auxiliary storage device 23. The control unit 21 may also read the program 97 stored in a semiconductor memory 98, such as a flash memory, implemented in the computer 90. Furthermore, the control unit 21 may download the program 97 from another server computer (not shown) connected via the communication unit 24 and a network (not shown) and store it in the auxiliary storage device 23.

[0189] The program 97 is installed as a control program for the computer 90, and is loaded into and executed by the main storage device 22. This causes the computer 90 to function as the information processing device 20 described above.

[0190] The technical features (constituent elements) described in each embodiment can be combined with each other, and by combining them, new technical features can be formed. The embodiments disclosed herein are illustrative in all respects and should be considered not to be limiting. The scope of the present invention is defined by the claims, not by the meaning described above, and is intended to include all modifications within the meaning and scope of the claims.

[0191] 10 Genome analysis system 20 Information processing device 21 Control unit 22 Main memory device 23 Auxiliary memory device 24 Communication unit 29 Reading unit 31 Reading device 32 Data server 51 Teacher data DB 52 Integrated DB 53 Learning model 531 Input layer 532 Intermediate layer 533 Output layer 55 Draft report DB 56 Report DB 58 Medical information DB 60 Report 61 Bibliographic information field 611 ID field 612 Patient information field 613 Specimen field 614 Histopathological diagnosis field 615 Specimen number field 62 Comment field 63 Nonsynonymous somatic mutation field 631 Gene field 632 Cytoband field 633 DNA mutation field 634 Amino acid mutation field 635 Allele frequency field 636 Knowledge data field 64 Germline mutation field 641 Gene column 642 Cytoband column 643 DNA mutation column 644 Amino acid mutation column 645 Knowledge data column 647 Normal part allele frequency column 648 Tumor part allele frequency column 65 Analysis column 651 Estimated tumor content column 652 Mutation frequency correlation coefficient column 66 RNA column 661 Gene column 662 Cytoband column 666 Knowledge data column 667 Mutation column 668 Number of reads column 71 Participate button 72 Request list 73 Request list 74 Specimen information column 75 Narrowing condition column 76 Re-search button 77 Candidate list 78 Confirm button 79 Request send button 81 Genome data acquisition unit 82 Genome data input unit 83 Output unit 84 First reception unit 85 First output unit 86 Second reception unit 87 Second output unit 90 Computer 96 Portable recording medium 97 Program 98 Semiconductor memory

Claims

DEPCT6618 / 04 / 25661. A program that enables a computer to perform the following processing: searching for training data, in which genome data obtained by read sequences included in the submission and genetic mutations based on the submission are recorded in association for multiple genetic tests previously performed; and generating a learning model to output predictions related to genetic mutations based on the submission, where genome data obtained by read sequences included in the submission is input, and genetic mutations are defined as output.

2. A program under claim 1 where the learning model outputs the predicted location of the mutated base pair.

3. A program under claim 1 or 2 where the learning model outputs the predicted tumor quantification of the submission. 4.The program enables the computer to perform the following processing: retrieving genome data obtained by reading the base sequence included in the submission; inputting the genome data into a learning model, which then outputs predictions related to genetic mutations onto the input genome data; and outputting predictions from the learning model based on the input genome data criteria.The program enables the computer to perform the following processing: exporting a report in which the analysis results related to the submitted sample and the version of the integrated database are recorded in a linked manner when the report export request is received based on the criteria of genetic mutations detected from the submitted sample and the integrated database, which includes the following in a linked manner: medical information related to genetic mutations retrieved from multiple information sources, the date of retrieval, and criteria information of the medical information; and exporting a report in which the analysis results related to the submitted sample and the version of the integrated database are recorded in a linked manner when the date and the report export request on that date are received based on the criteria of genetic mutations detected from the submitted sample and the integrated database on a historical date.

6. The program under claim 5, where the report contains medical information extracted from the integrated database by designating the genetic mutations detected from the submitted sample as the key. 7.Programs under claim 5 or 6, whereby the integrated database is updated by adding medical information related to genetic mutations, additional reports will be exported based on the criteria of genetic mutations detected from submissions and the updated integrated database.

8. Programs under claim 7, where requests for review related to updating the integrated database are submitted to experts, the review results based on the submitted requests are accepted, and incentives based on the review results are recorded in association with the experts. 9.The program enables the computer to perform the following processing: retrieving genome data obtained by reading the base sequence included in the submission; inputting the retrieved genome data into a learning model that outputs predictions related to genetic mutations based on the genome data input; retrieving predictions related to genetic mutations output from the learning model based on the input genome data criteria; and exporting a report in which the analysis results related to the submission and a version of the integrated database are recorded, linked together according to the retrieved prediction criteria, and the integrated database which integrates the following, linkedly: medical information related to genetic mutations retrieved from multiple information sources, the retrieve date, and criteria information of the medical information.10.A program under one of Claims 5 through 9 where a request for report review is passed on to an expert; the review outcome, based on the passed review request, is accepted, and incentives based on the received review outcome are recorded in association with the expert.

11. A program under Claims 8 or 10 where incentives are cash vouchers, report writing request certificates, or learning model certificates.

12. A program under one of Claims 8, 10, and 11 where incentives vary based on the review outcome criteria. 13.The learning model consists of: an input layer, which is where the genomic data obtained by reading the base sequence included in the submission is fed into; an output layer, which outputs predictions related to genetic mutations based on the submission; and an intermediate layer, in which parameters are learned using training data, in which the genomic data obtained by reading the base sequence included in the previously sampled submission and genetic mutations based on the submission are recorded in association. The learning model then directs the computer to perform the function where predictions related to genetic mutations based on the submission are output from the output layer through the computation of the intermediate layer, where the genomic data obtained by reading the base sequence included in the submission is fed into the input layer.14The information processing device consists of: a first receiving unit that receives genetic mutations detected from the submitted samples; a first output unit that outputs a report containing the analysis results related to the submitted samples and the version of the integrated database, linked together according to the genetic mutation criteria received by the first receiving unit and the integrated database. This report integrates the following: medical information related to genetic mutations retrieved from multiple sources, the retrieve date, and medical information criteria; a second receiving unit that receives the historical date, the report export request date, and the genetic mutations detected from the submitted samples; and a second output unit that outputs a report containing the analysis results related to the submitted samples and the version of the integrated database, linked together according to the genetic mutation criteria received by the second receiving unit and the integrated database on that date.15.The information processing equipment consists of: a genome acquisition unit that acquires genome data obtained by reading the base sequence included in the submission; a genome input unit that feeds genome data into the learning model, which outputs predictions related to genetic mutations onto the genome input; and an output unit that outputs the predictions from the learning model based on the genome data criteria entered by the genome input unit.

16. The information processing method enables the computer to perform the following processing: acquiring genome data obtained by reading the base sequence included in the submission; inputting genome data into the learning model, which outputs predictions related to genetic mutations onto the genome input; and outputting predictions from the learning model based on the input genome data criteria.17.The model generation methodology comprises: acquiring training data, which includes the following interconnected records: genome data obtained by read sequences included in previously sampled submissions and genetic mutations based on those submissions; and generating a learning model to output predictions related to genetic mutations based on the submission, where the genome data obtained by read sequences included in the submission is fed in, and genetic mutations are defined as the output.

Images