Evaluation of the relative quantitative effect of somatic point mutations at the individual tumor level for prioritization

Abstract

The technology described in this specification discloses a method or system for analyzing genomic data, calculating predictive factors, and quantitatively evaluating biological effects based on the predictive factors. Biological effects such as the pathogenicity of cancer and the risk that a subject may develop a specific cancer can be determined based on the predictive factors. The predictive factors may include the observed occurrence number of a gene variant divided by the predicted occurrence number of the gene variant. Prediction of drug treatment may include prioritization of gene variants according to selective variant effects and determination of which drug treatment to prioritize. The prediction uses genomic coordinates for each gene variant and nucleotide changes from various databases, and may further include filtering out and excluding duplicate samples from the same subject.

Description

Technical Field

[0001] Cross - reference to related patent applications This patent application claims priority to U.S. Provisional Patent Application No. 63 / 354,438, filed on June 22, 2022, with the title "Assessment of relative quantitative effect of somatic point mutations at the individual tumor level for prioritization", the disclosure of which is incorporated herein by reference.

Background Art

[0002] Background of the Invention Cancer treatment has become more accurate and individualized with respect to genomic mutations in tumors. Cancer cells are affected by driver variants with a spectrum of pathogenic effects. These drivers confer a selective advantage to the tumor. Currently, cancer gene variants are dichotomized into harmful or non - harmful variants. There can be a large number of harmful variants that can be targeted by biological drugs, and in many cases, not all of them can be targeted with respect to side effects, drug availability, and side effects. Currently, there is no method for prioritizing which gene / gene(s) a drug should target.

[0003] The identification of many variants in the human genome that can cause disease has been made possible by next - generation sequencing technology. To distinguish sequence variants that are neutral as a cause from active disease drivers, various prediction tools have been proposed. Multiple types of data have been promisingly shown to be useful for distinguishing disease drivers from neutral variants. These and various other types of data have shown that they have information indicating whether variants within the genome can be pathogenic or are substantially neutral, but no evidence has been presented indicating whether a particular type of data is actually useful and to what extent.

[0004] Therefore, there is a need for tools to assist in the identification of new drivers and the estimation of the different effects of mutations in tumors.

[0005] Therefore, there is a need for technologies that enable better prediction of outcomes, selection of treatment methods, and prioritization of more important variants for tumors. SUMMARY OF THE INVENTION MEANS FOR SOLVING THE PROBLEM

[0006] SUMMARY OF THE INVENTION Aspects of the present disclosure relate to systems and methods for assessing the risk of a disease (e.g., cancer), predicting the treatment response of tumors having specific gene variants, and proposing possible treatment modalities based on the assessed risk.

[0007] In one embodiment, the present disclosure describes a method for quantitatively assessing the biological effect of at least one gene variant of a subject. This method uses a computer system including a processor, a memory, and instructions stored in the memory, which, when executed by the processor, execute a method including a series of steps. This method receives at least one gene variant of a subject. This method analyzes a genomic database to determine a mutation rate for the at least one gene variant. This method determines the observed occurrence number of the at least one gene variant in the database. This method calculates the predicted occurrence number of the at least one gene variant based on the mutation rate and the observed occurrence number. This method calculates a predictor associated with the at least one gene variant based on the mutation rate, the observed occurrence number, and the predicted occurrence number. This method uses the predictor to generate a quantitative assessment of the biological effect of the at least one gene variant. Then, the computer system transmits the predictor and the quantitative assessment to a user device.

[0008] In certain embodiments, the quantitative assessment can include prognosis, the risk of developing cancer, or treatment response. In certain embodiments, the predictor includes a tumor variant amplitude (TVA) equal to the logarithm of the ratio of the observed occurrence of at least one gene variant in a genomic database to the predicted occurrence of the at least one gene variant in the genomic database. In certain embodiments, prior to analyzing the genomic database, the genomic database is filtered to avoid duplication of samples from the same subject and is filtered using at least one of the genomic coordinates of each entry; the nucleotide changes of each entry; the somatic state of each entry; or the type of cancer of each entry.

[0009] In certain embodiments, the quantitative assessment can compare multiple drug therapies for a tumor to gene variants present in the tumor. Based on the comparison, the quantitative assessment can select a drug therapy from among the multiple drug therapies for use in the subject's tumor. In certain embodiments, the quantitative assessment can predict a likely response of the subject's tumor to the selected drug therapy based on the comparison. In certain embodiments, the step of identifying a drug therapy selected from among the multiple drug therapies includes ranking the gene variants based on the classification of the gene variants and based on the TVA. In certain embodiments, the quantitative assessment can include comparing the subject's germline DNA to a database of gene variants and cancer risk and quantifying the risk that the subject will develop cancer based on the comparison. In certain embodiments, the quantitative assessment can further include comparing the subject's tumor DNA to a database of gene variants and tumor mutations and quantifying the subject's prognosis. In certain embodiments, the method can determine a diagnosis using a predictor and an artificial intelligence model.

[0010] In one embodiment, the present disclosure describes a system for quantitatively evaluating the biological effect of at least one genetic variant of interest for use with a user device. The system includes a measurement device, a processor, and a memory accessible by the processor and storing computer program instructions that, when executed by the processor, perform a method. The measurement device measures the occurrence number of at least one genetic variant. The processor analyzes a genomic database to determine a mutation rate for at least one genetic variant. The processor determines the observed occurrence number of at least one genetic variant in the database. The processor calculates the predicted occurrence number of at least one genetic variant based on the mutation rate and the observed occurrence number. The processor calculates a predictor associated with at least one genetic variant based on the mutation rate, the observed occurrence number, and the predicted occurrence number. The processor uses the predictor to generate a quantitative evaluation of the biological effect of at least one genetic variant. The predictor and the quantitative evaluation are transmitted to the user device.

[0011] In one embodiment, the quantitative evaluation can include prognosis, the risk of developing cancer, or treatment response. In one embodiment, the predictor includes a tumor variant amplitude (TVA) equal to the logarithm of the ratio of the observed occurrence number of at least one genetic variant in the genomic database to the predicted occurrence number of at least one genetic variant in the genomic database. In one embodiment, prior to analyzing the genomic database, the processor filters the genomic database to avoid duplication of samples from the same subject, and the processor filters the genomic database using at least one of the genomic coordinates of each entry; the nucleotide change of each entry; the somatic state of each entry; or the cancer type of each entry.

[0012] In one embodiment, the quantitative evaluation compares multiple drug therapies for a tumor with gene variants present in the tumor. Based on the comparison, a drug therapy can be selected from among the multiple drug therapies for use in the subject's tumor. The quantitative evaluation can further include predicting a likely response of the subject's tumor to the selected drug therapy based on the comparison. In one embodiment, the step of identifying a drug therapy selected from among the multiple drug therapies includes ranking the gene variants based on the classification of the gene variants and based on the TVA. In one embodiment, the quantitative evaluation can compare the subject's germline DNA with a database of gene variants and cancer risk, and based on the comparison, quantify the risk that the subject will develop cancer and transmit the risk to a user device. In one embodiment, the quantitative evaluation can include comparing the subject's tumor DNA with a database of gene variants and tumor mutations, and based on the comparison, quantifying the subject's prognosis. In one embodiment, the system can use predictors and artificial intelligence models to determine a diagnosis.

[0013] To better understand the features listed above of the present invention, a more specific description of the present invention, briefly summarized above, can be made by referring to embodiments, some of which are shown in the accompanying drawings. However, it should be noted that the accompanying drawings only show typical embodiments of the present invention, and the present invention can recognize other equally effective embodiments.

Brief Description of the Drawings

[0014]

Figure 1

[0015]

Figure 2

[0016]

Figure 3

[0017]

Figure 4

[0018]

Figure 5

[0019]

Figure 6

[0020]

Figure 7

[0021]

Figure 8

[0022]

Figure 9

[0023]

Figure 10

[0024]

Figure 11

[0025]

Figure 12

[0026]

Figure 13

[0027]

Figure 14

[0028]

Figure 15

[0029]

Figure 16

[0030]

Figure 17

[0031]

Figure 18

[0032]

Figure 19

[0033]

Figure 20

[0034]

Figure 21

[0035]

Figure 22

[0036]

Figure 23

[0037]

Figure 24

[0038]

Figure 25

[0039]

Figure 26

[0040]

Figure 27

[0041]

Figure 28

[0042]

Figure 29

[0043]

Figure 30

[0044]

Figure 31

[0045]

Figure 32

[0046]

Figure 33

[0047]

Figure 34

[0048] Other features of this embodiment will become apparent from the following detailed description.

[0049] Detailed Description In the following detailed description of the preferred embodiments, reference is made to the accompanying drawings, which form a part hereof, and in which are shown by way of illustration specific embodiments in which the invention may be practiced. The subject matter of the present disclosure may be embodied in many different forms and should not be construed as limited to the embodiments set forth herein; rather, these embodiments are provided so that this disclosure will satisfy applicable legal requirements. Indeed, many modifications and other embodiments of the subject matter of the present disclosure described herein will come to mind to one skilled in the art to which the present disclosure pertains having the benefit of the teachings presented in the foregoing description and the related drawings. Accordingly, it is to be understood that the subject matter of the present disclosure is not to be limited to the specific embodiments disclosed and that modifications and other embodiments are intended to be included within the scope of the appended claims.

[0050] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this disclosure belongs. However, any compositions, methods, and materials similar or equivalent to those described herein can be used in the practice or testing of the present invention. All publications mentioned are incorporated herein by reference in their entirety.

[0051] The use of "a", "an", "the", and similar referents in the context of describing the claimed invention (especially in the context of the claims) is to be construed to include the singular and the plural unless otherwise indicated herein or clearly contradicted by context.

[0052] The recitation of ranges of values herein is merely intended to serve as a shorthand method of referring individually to each separate value falling within the range, unless otherwise indicated herein, and each separate value is incorporated herein as if it were individually recited herein.

[0053] The use of the term "about" is intended to indicate a value that is above or below the value recited within a range of approximately + / - 10%; in other embodiments, the value may be within a range of values above or below the recited value within approximately + / - 5%; in other embodiments, the value may be within a range of values above or below the recited value within approximately + / - 2%; in other embodiments, the value may be within a range of values above or below the recited value within approximately + / - 1%. The foregoing ranges are intended to be made clear by the context and are not meant to suggest further limitation. All methods described herein may be performed in any suitable order, unless otherwise indicated herein or unless clearly contradicted by the context. The use of any examples, or exemplary language (e.g., "such as") provided herein is merely intended to better illustrate the invention and does not limit the scope of the invention, particularly unless otherwise claimed. No term herein should be construed as indicating that any non-claimed element is essential to the practice of the invention.

[0054] Cancer variant

[0055] The present disclosure relates to methods and systems for determining which cancer genes are most useful / effective in predicting optimal treatment and outcome, including, for example, reduced tumor size (responsive to drug treatment), remission, and the like.

[0056] Cancer cells are affected by driver variants that have a spectral pathogenic effect. These drivers confer a selective advantage to the tumor. In cancer treatment, the diagnosis of genetic variants in tumor cells is used to select the most appropriate treatment regimen for individual patients. In breast cancer, for example, genetic variations in estrogen receptor expression or human epidermal growth factor receptor 2 (Her2) receptor tyrosine kinase expression determine whether anti-estrogen drugs (tamoxifen) or anti-Her2 antibodies (Herceptin) are incorporated into the treatment plan. In chronic myeloid leukemia (CML), the diagnosis of the Philadelphia chromosome gene translocation that fuses the genes encoding the Bcr and Abl receptor tyrosine kinases indicates that Gleevec (STI571), a specific inhibitor of Bcr-Abl kinase, should be used in the treatment of cancer. In CML patients with such genetic changes, inhibition of Bcr-Abl kinase leads to rapid elimination of tumor cells and remission of leukemia. Furthermore, genetic testing services are currently available, and based on the discovery that certain single nucleotide polymorphisms (SNPs) are associated with the risk of many common diseases, information regarding disease risk is provided to individuals.

[0057] In the present disclosure, as an example, a cancer shared dataset from multiple cancer genomic databases can be combined and two different metrics based on the observed and predicted frequencies of variants based on cancer-specific somatic mutagenesis rates can be applied to 535 cancer genes. The first metric is a binary classifier based on a binomial test, while the second metric, the tumor variant amplitude (TVA), is a continuous metric representing the selective advantage of the variant. The correlation of TVA with many experimental and clinical metrics related to cancer was examined. TVA was superior to all other computational tools in terms of correlation with the experimentally derived functional scores of cancer mutations. It was also highly correlated with drug response, overall survival, and other clinical significance in related cancer genes. This study demonstrates the significant impact of a selective advantage metric based on large-scale cancer datasets to understand the spectral effects of driver variants in cancer.

[0058] Variant scoring techniques

[0059] Cancer cells accumulate somatic variants over time. Some variants confer a selective advantage, providing cancer cells with improved capabilities such as, among other things, enhanced growth, invasion, and metastasis to other organs. Conventionally, genetic variants in cancer have been divided into two different categories: driver variants that affect protein activity and contribute to cancer traits, and passenger variants that do not confer an advantage to cancer cells. Since this dichotomous classification may be too simplistic, a spectrum-based approach has been proposed to evaluate the pathogenicity of variants. Such an approach discriminates variants according to quantitative metrics such as protein stability and selection pressure. The selection pressure approach defines multiple subgroups of variants: disruptive variants by negative selection, passenger variants by neutral selection, potential driver variants by positive selection in the presence of other same-gene driver variants, weak driver variants by moderate positive selection, and strong driver variants by high positive selection. Most pathogenicity scores come with a threshold that provides a dichotomous classification due to the simplicity of this approach and the lack of information regarding the quantitative effects of variants. The continuous scores underlying these classifiers are not suitable for the task of predicting the quantitative effects of variants. Some studies have attempted to directly quantify the effects of variants using different approaches, but each study has its limitations. One of the best-known methods is Envision, a tool based on supervised learning of deep mutational scanning (DMS) datasets. The main limitations of Envision are the small number of sufficient DMS experiments and the mixing of information from different experiments and genes in different ways. Another approach is based on evolutionary selection intensity. The limitations of the present disclosure are mainly the very small sample sizes and separation by cancer type. Some of these quantification tools are superior to classical classifiers in predicting the effects of variants (plural).

[0060] Variant classifiers rely on a variety of features including protein sequence, evolutionary conservation, structural information, biophysical information, 3D protein clusters, biochemical assays, allele frequencies, and the occurrence of tumor variants. Another way to classify variants is to use genome context-specific mutation rates. Mutation rates depend on the genome context and are not constant for a given genomic change. Several methods for estimating mutation rates and avoiding potential biases can be described. Next, a binomial test can be used to identify tumor variants that are more common than expected based on the mutation rate. Variants that occur at a higher rate than expected are more likely to have positive selection in the tumor evolutionary process and are thus more likely to be true drivers of tumorigenesis. Brown et al. (Brown, A. L., Li, M., Goncearenco, A. & Panchenko, A. R. Finding driver mutations in cancer: Elucidating the role of background mutational processes. PLoS Comput. Biol. 15, (2019)(PMID:31034466) identified new drivers using a binomial test based on the trinucleotide context mutation rate. They reported that this approach showed improved performance compared to conventional methods based on variant occurrence. The main limitations of their study were that it was based on the analysis of a small number of tumor samples, included only samples sequenced against normal tissue, used a small validation dataset, and did not compare their results to information from healthy populations at all. The binomial test has not yet been used in large datasets to systematically identify new drivers.

[0061] In this study, the binary method was implemented on a large-scale cancer shared dataset (CSD) of 137,224 tumor samples collected from four different sources (TCGA, ICGC, MSKCC, and GENIE). Using the mutation rate, the number of sequenced samples, and the occurrence of each variant to classify drivers, the relative strength or impact of each variant on cancer cells was quantified. To quantify this relative strength, a predictor named "tumor variant amplitude" (TVA) was developed, which represents the logarithm of the ratio of the actual occurrence and the predicted occurrence of a variant based on the mutation rate. TVA was validated as a quantitative predictor of the relative strength or impact of variants using experimental, pharmacological, and clinical data. A combination of a binomial test to discover new drivers and TVA to measure the impact of variants on a spectral scale yielded a comprehensive and new catalog of many somatic drivers. Each driver among 535 selected COSMIC cancer genes was assigned a ranking of its impact. This catalog can be particularly useful for the long tail of drivers that mutate at much lower frequencies compared to mutation hotspots.

[0062] In certain embodiments, TVA can be used as part of a system to propose treatments based on prioritized dominant variants of samples from patients. The system can access a database of treatments such as drug therapies and can show a set of prioritized drug therapies to a healthcare provider based on variants prioritized by TVA or another predictor. In certain embodiments, artificial intelligence (AI) can utilize the predictor as a feature of a set of features to provide a list of possible diagnoses relevant to a particular patient to a physician. In certain embodiments, an AI module can include a trained model incorporating information related to the predictor as part of a process to classify a disease or as part of a process to propose a treatment for a disease.

[0063] Computer-readable programming

[0064] Many operating systems, including Linux (registered trademark), UNIX (registered trademark), OS / 2 (registered trademark), and Windows (registered trademark), can execute multiple tasks simultaneously and are called multitasking operating systems. Multitasking is the ability of an operating system to execute more than one executable file at the same time. Each executable file runs in its own address space and cannot share any of their memory. Therefore, it is impossible for any program to damage the execution of any other program running on the system. However, programs can only exchange information by going through the operating system (or by reading files stored in the file system).

[0065] Since multiprocess computing is similar to multitask computing, the terms task and process are often used interchangeably, although some operating systems distinguish between the two. The present invention may be a system, method, and / or computer program product at any possible level of technical detail integration. The computer program product may include a computer-readable storage medium (or multiple media), and the computer-readable storage medium (or multiple media) has computer-readable program instructions thereon for causing a processor to execute aspects of the present invention. The computer-readable storage medium may be a tangible device that can hold and store instructions for use by an instruction execution device.

[0066] A computer-readable storage medium may be, for example, an electronic storage device, a magnetic storage device, an optical storage device, an electromagnetic storage device, a semiconductor storage device, or any suitable combination of the foregoing, but is not limited thereto. A non-exhaustive list of more specific examples of computer-readable storage media includes the following: portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), static random access memory (SRAM), portable compact disk read-only memory (CD-ROM), digital versatile disk (DVD), memory stick, floppy (registered trademark) disk, mechanically encoded devices such as punch cards, or raised structures in grooves having instructions recorded thereon, and any suitable combination of the foregoing.

[0067] An example of the system is shown in FIG. 34. Computing device 3400 is shown with a processing device 3404 (e.g., a central processing unit (CPU), but also including a graphics processing unit (GPU) or even multiple processors or cores), an input / output device 3402, a network adapter 3406, and a memory 3410. Network adapter 3406 connects computing device 3400 to a network 3408 which may include a measurement device 3430. In the memory 3410 of computing device 3400, there is data such as measurement data 3412, patient data 3414, drug data 3416, and treatment data 3418. Some data may exist in other locations connected to the network, such as a database of therapeutic treatments or a database of human genes. Also, in the memory 3410 of the computing device, there may be various programs, subroutines or algorithms, such as a classification algorithm 3420, an analysis algorithm 3422, and a comparison algorithm 3434.

[0068] As used herein, a computer-readable storage medium should not be construed to be a transient signal per se, such as a radio wave or other freely propagating electromagnetic wave, an electromagnetic wave propagating through a waveguide or other transmission medium (e.g., an optical pulse passing through an optical fiber cable), or an electrical signal transmitted through an electrical wire. The computer-readable program instructions described herein can be downloaded from a computer-readable storage medium to respective computing / processing devices, or to an external computer or external storage device via a network 3408, such as the Internet, a local area network, a wide area network, and / or a wireless network. The network 3408 can include copper transmission cables, optical transmission fibers, wireless transmission, routers, firewalls, switches, gateway computers, and / or edge servers for data transmission between devices. A network adapter card or network interface 3406 in each computing / processing device receives the computer-readable program instructions from the network and transfers the computer-readable program instructions for storage on a computer-readable storage medium within each respective computing / processing device.

[0069] Computer-readable program instructions for carrying out the operation of the present invention may be source code or object code written in any combination of one or more programming languages, including assembly instructions, instruction set architecture (ISA) instructions, machine instructions, machine-dependent instructions, microcode, firmware instructions, state setting data, configuration data for integrated circuits, or object-oriented programming languages such as Smalltalk, C++, and procedural programming languages such as the "C" programming language or similar programming languages. The computer-readable program instructions may be executed entirely on the user's computer, partially executed on the user's computer as a stand-alone software package, partially executed on the user's computer and partially executed on a remote computer, or executed entirely on a remote computer or server. In the latter scenario, the remote computer may be connected to the user's computer via any type of network, including a local area network (LAN) or a wide area network (WAN), or the connection may be made to an external computer (e.g., via the Internet using an Internet service provider).

[0070] In some embodiments, for example, an electronic circuit including a programmable logic circuit, a field programmable gate array (FPGA), or a programmable logic array (PLA) can execute the computer-readable program instructions by utilizing the state information of the computer-readable program instructions to personalize the electronic circuit in order to carry out aspects of the present invention.

[0071] Aspects of the invention are described herein with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer readable program instructions. These computer readable program instructions may be provided to a processor of a general purpose computer, special purpose computer, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions / acts specified in the block or blocks of the flowchart and / or block diagram.

[0072] These computer-readable program instructions can also be stored in a computer-readable storage medium that can direct a computer, a programmable data processing apparatus, and / or other devices to function in a particular manner, such that the computer-readable storage medium having instructions stored therein includes a manufactured article that includes instructions for implementing the functionality / acts specified in the flowchart and / or block(s) of the block diagram. The computer-readable program instructions can also be loaded onto a computer, other programmable data processing apparatus, or other device, causing a series of operational steps to be performed on the computer, other programmable apparatus, or other device to implement the functionality / acts specified in the blocks of the flowchart and / or block diagram. The flowchart and block diagrams in the figures illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of the present invention. In this regard, each block in the flowchart or block diagram can represent a module, segment, or portion of instructions that includes one or more executable instructions for implementing the specified logical function(s).

[0073] In some alternative implementations, the functions shown in the blocks may occur in orders different than those shown in the figures. For example, two blocks shown in succession may actually be executed substantially simultaneously or in the reverse order depending on the functions involved. Also, it should be noted that each block of the block diagrams and / or flowchart diagrams, as well as combinations of blocks in the block diagrams and / or flowchart diagrams, can be implemented by a dedicated hardware-based system that performs the specified functions or actions, or by a combination of dedicated hardware and computer instructions. Although specific embodiments of the present invention have been described, it will be understood by those skilled in the art that there are other embodiments equivalent to the described embodiments. Therefore, it should be understood that the present invention is not limited by the specifically shown embodiments, but only by the appended claims.

[0074] From the above description, it can be understood that the present invention provides a system, a computer program product, and a method for efficiently performing the described techniques. References to elements in the singular in the claims are not intended to mean "only one" unless explicitly so recited, but rather "one or more." All structural and functional equivalents to the elements of the above exemplary embodiments that are now known or later become known to those skilled in the art are intended to be encompassed by the claims. Elements of the claims herein are not to be construed under the provisions of 35 U.S.C. § 112, paragraph 6, unless the element is expressly recited using the phrase "means for" or "step for."

[0075] From the foregoing written description of the invention, those skilled in the art will be able to make and use what is presently considered to be the best mode. However, those skilled in the art will understand and recognize the existence of alternatives, adaptations, variations, combinations, and equivalents to the specific embodiments, methods, and examples herein. Those skilled in the art will recognize that the disclosure is illustrative only and that various changes can be made within the scope of the invention. Additionally, although a particular feature of the present teachings may be disclosed with respect to only one of a plurality of implementations, such a feature may be combined with one or more other features of one or more other implementations as desired and advantageous for any given or particular function. Further, as long as the terms "including", "includes", "having", "has", "with", or variations thereof are used in either the detailed description or the claims, such terms are intended to be as inclusive as the term "comprising".

[0076] Other embodiments of the present teachings will be apparent to those skilled in the art from consideration of the specification and practice of the teachings disclosed herein. Accordingly, the invention should not be limited by the described embodiments, methods, and examples, but should be limited only by all embodiments and methods within the scope and spirit of the invention. Accordingly, the invention is not limited to the specific embodiments shown herein but is limited only by the following claims.

Examples

[0077] Method

[0078] List of cancer genes

[0079] The analysis focuses on a set of genes from the COSMIC cancer census obtained in April 2021. As an example, the study focuses on 546 genes defined in the COSMIC cancer census as having known somatic pathogenic variants, and their roles are not only as fusion genes. Eleven genes were excluded from the analysis, resulting in the selection of 535 cancer genes. For these genes (MRTFA, NSD3, NCOA4, MALAT1, TENT5C, NSD2, AFDN, KNL1, SSX2, DEK, and NOTCH1), gene exclusion was performed due to missing information such as transcripts and missing positions in hg19. All possible variants for the selected genes were obtained from dbNSFP by the ENSEMBLE coordinates of the genes.

[0080] Data collection

[0081] Data was obtained from four different data sources - TCGA, ICGC, GENIE, and MSKCC. Specific APIs for each source were used to download the data (GENIE and MSKCC were downloaded from the same database). All variants were converted to hg19 coordinates using the hg19 position of the variant and the nucleotide changes from the database, although other genomic coordinate systems may also be used. Preprocessing was performed to filter out and exclude duplicate samples from the same patient, and it was confirmed that the somatic validation status and cancer type for each variant were collected.

[0082] Specific information on variants for all available variants was collected from dbNSFP v4.2a, a database that compiles many variant predictor scores (sequence-based, conservation, variant annotation source, and meta predictors) for many possible transcripts (obtained from VEP, ANNOVAR, and snpEff). Summaries of allele counts and frequencies of each variant in normal populations from gnomAD, ESP6500, and UK10K were also obtained from the dbNSFP database. Preprocessing of dbNSFP was performed to separate columns into different transcripts for each gene.

[0083] Drug Response Information Collection

[0084] Bulk data for "IC50 drug screening" was obtained from the website of Genomics of Drug Sensitivity in Cancer. Bulk mutation data for cell lines was obtained from the website of Cell Model Passports.

[0085] TCGA Clinical Data Collection

[0086] Clinical data for TCGA samples was obtained from the website of cBioPortal. Mutation data for all TCGA samples was obtained via the cBioPortal API.

[0087] Deep Mutation Scanning (DMS) Experiment Data Collection

[0088] PTEN DMS experiment data was obtained from MaveDB, a public repository for datasets from Multiplexed Assays of Variant Effect. TP53 DMS experiment data was obtained from the TP53 UMD database.

[0089] Calculation of Mutation Rate

[0090] For all variants, the trinucleotide context for the plus strand was extracted using the Bio.seq module from the Biopython v1.75 package. The mutation rate for each of the 96 trinucleotides was defined according to the MutaGene mutation rate estimation.

[0091] Transcript selection

[0092] For each gene, transcripts were selected from all possible transcripts according to the COSMIC major transcript selection. If no transcript was selected in COSMIC, the Matched Annotation from NCBI and EMBL-EBI (MANE) transcript was retrieved from BioMart. Grouping of nucleotide changes to amino acid changes was performed according to the VEP HGVS protein sequence name (HGVSP) in the selected transcripts, and only information about the transcripts selected for the gene was retained. For each amino acid change, the mutation rate was calculated as the sum of the mutation rates of all single nucleotide substitutions that result in the given amino acid change. The CSD occurrences of all single nucleotide substitutions that result in the given amino acid change were also summed.

[0093] Calculation of the binomial test

[0094] Based on the number of CSD samples = n (CSD samples in which the gene of the variant was sequenced), the number of CSD variant occurrences = k (number of CSD samples with the variant), and the mutation rate = p (based on the MutaGene estimated rate), a one-sided binomial test was performed for all variants. For variants that are never seen in healthy populations, only the occurrences of samples sequenced compared to the patient's normal tissue were used to avoid misidentifying germline mutations. For all other variants, the occurrences of both samples compared to normal tissue and samples not compared were used. All calculations were performed using SciPy.

[0095] Classifier testing

[0096] For comparison of the MutaGene estimated values and the improved estimated values, similar to Brown et al., a combined benchmark dataset from the MutaGene web server was used. The dataset was downloaded from the MutaGene website (https: / / www.ncbi.nlm.nih.gov / research / mutagene / ), and various parameters including the receiver operating characteristic (ROC) curve, area under the curve (AUC), and maximum Matthews correlation coefficient (MCC) were calculated for (i) the occurrence of MutaGene; (ii) the binomial p-value of MutaGene; (iii) the occurrence of CSD; (iv) the binomial p-value for all CSD occurrences without considering information on the healthy population; and (v) the binomial p-value considering information on the healthy population.

[0097] Deep mutational scanning (DMS) correlation

[0098] The Spearman correlation of the DMS studies of TVA and cancer genes was compared with the correlation between 31 publicly available bioinformatics scores and DMS scores. Thirty of the scores were obtained from dbNSFP, while the EVE score was obtained from the EVE website (evemodel.org). The scores used are shown in Table 3.

[0099] Calculation of tumor variant amplitude

[0100] For all variants, based on the number of CSD samples = n (CSD samples in which the gene of the variant was sequenced), the CSD variant occurrence = k (number of CSD samples with the variant), and the mutation rate = p (based on the estimated rate of MutaGene), the formula:

Number

[0101] Binomial p-value multiple testing correction

[0102] To correct for multiple testing, false discovery rate (FDR) correction was used for all binomial test p-value variants from all selected genes. All calculations were performed using the statsmodels package.

[0103] Filtering of healthy populations

[0104] Regarding variants reported in one of the normal genomic databases - gnomAD, UK10K, or ESP6500, to confirm the somatic state and avoid germline contamination, binomial tests and TVA calculations were performed based only on the occurrence of samples compared to normal tissue. In addition,

Number

[0105] Definition of passenger labels

[0106] In the examination of the driver positions in the catalog, the number of passengers at the same positions (where drivers were present) was analyzed. Passengers were defined as variants that also appeared in the database of healthy populations and were not significant (p-value > 0.1) in the binomial test. Due to the low power of the binomial test for positions with low mutation rates, the condition of statistical significance was not sufficient. Therefore, the condition of appearance in the database of healthy populations was also added because the test might not detect drivers with relatively low TVA values.

[0107] Criteria for the association of drugs and genetic alterations in GDSC

[0108] This analysis focused on pairs of drugs and genetic alterations in the GDSC database that met the following criteria: (i) drug response related to cancer gene variants, (ii) at least 50 cell lines with cancer gene alterations, (iii) an effect size greater than 0.7 (greater than 0.5 indicates a moderate effect size, and greater than 1 indicates a large effect size), (iv) the drug-gene association is statistically significant (false discovery rate (FDR) p-value < 0.1), and (v) the association can be explained by the effect of the drug on the mutant proteins of the related pathway.

[0109] Definition of drug response subgroups

[0110] To compare the responses to drugs between the TVA values of gene variants, samples were binned according to their TVA scores of the gene variants and their presence in the binary catalog of drivers. Non-drivers were defined as variants not in the binary catalog with TVA ≤ 1.5. This larger TVA value for variants not in the catalog increases the number of variants in the non-driver group. Weak drivers were defined as variants in the binary catalog with 1 ≤ TVA < 2. Moderate drivers were defined as variants in the catalog with 2 ≤ TVA < 3. Strong drivers were defined as variants in the catalog with 3 ≤ TVA < 4. Very strong drivers were defined as variants in the catalog with TVA ≥ 4. The last group was required for the KRAS, NRAS, and PIK3CA genes.

[0111] MSI threshold

[0112] In uterine cancer, TCGA samples with POLE variants were defined as samples positive for microsatellite instability (MSI) according to the MSI sensor score. A cut-off of 3.5 was used as suggested in the original paper of the MSI score.

[0113] Definition of POLE driver subgroups

[0114] For the comparison between POLE variants of TVA, samples were binned according to their appearance in the binary catalog of those POLE variants and drivers that had the highest TVA scores. Non-drivers were defined as POLE variants not present in the binary catalog with TVA <= 1.5. Weak drivers were defined as POLE variants in the binary catalog with 1 <= TVA < 2. Moderate drivers were defined as POLE variants in the catalog with 2 <= TVA < 3. Strong drivers were defined as POLE variants in the catalog with 3 <= TVA < 4.

[0115] Definition and calculation of survival analysis subgroups

[0116] In the overall survival analysis, all TCGA samples having more than one unique TP53 variant were excluded. For the comparison of overall survival among patients with TVA values of TP53 variants, samples were binned according to their TP53 TVA scores and their appearance in the binary catalog of drivers. Non-drivers were defined as TP53 variants not present in the binary catalog with TVA <= 1.5. Weak drivers were defined as TP53 variants in the binary catalog with 1 <= TVA < 2. Moderate drivers were defined as TP53 variants in the catalog with 2 <= TVA < 3. Strong drivers were defined as TP53 variants in the catalog with 3 <= TVA < 4. The analysis was performed in R using the survival package and visualized using the survminer package.

[0117] Multivariate overall survival (OS) analysis

[0118] For all TCGA pan-cancer samples with unique TP53 variants, a multivariate analysis of TVA continuous values, diagnostic age, gender, and cancer type was performed. Five samples were filtered out and excluded due to their small sample sizes for their cancer types: testicular germ cell tumor (TGCT), pheochromocytoma and paraganglioma (PCPG), diffuse large B-cell lymphoma (DLBC). The analysis was performed in R using the survival package and visualized using the forestmodel package.

Table 1

Table 2

Table 3-1

Table 3-2

[0119] Results

[0120] (Example 1) Improvement of the binomial test

[0121] An improved application of the binomial test for cancer gene variants was developed to identify pathogenic variants with positive selection. The main improvements were: (i) by using data from healthy populations, providing more accurate predictions than analyses based solely on occurrence in cancer datasets; (ii) by an analysis that allows inclusion of samples not sequenced against normal tissue as a comparison, significantly increasing the sample size; and (iii) by grouping nucleotide changes that result in the same amino acid change, focusing on the impact on proteins rather than genomic changes. The parameters used in the analysis were the occurrence of variants in cancer datasets, the number of samples in cancer datasets, and the estimated mutation rate for the genomic context of each variant. Four different public databases were used to create a cancer shared dataset (CSD) of 137,224 samples, which is about six times more than previously investigated. The mutation rate was based on MutaGene's pan-cancer context-dependent mutation rate estimation. The binomial test was performed in two different ways: (i) for all variants, including the occurrence in all samples of the CSD; (ii) for variants occurring in samples of the CSD, but only including the occurrence in samples of the CSD with a comparison to normal tissue for variants that appear in the healthy genomic database. Additionally, in the second method, variants with an allele frequency greater than 0.0001 in the healthy genomic database were excluded because they may represent normal genomic mutations (see Materials and Methods). This approach was tested against a combined benchmark dataset from the MutaGene web server used by Brown et al. This combined dataset was derived using five different datasets from experimental assays and included a total of 5,277 labeled variants from 58 cancer genes. The CSD occurrence approach outperformed both MutaGene occurrence and MutaGene binomial p-values in all investigated metrics (AUC-ROC: 0.7904 > 0.7083 / 0.7903) (Table 4, Figure 1). The binomial p-value without considering the healthy population improved the prediction accuracy compared to CSD occurrence (AUC-ROC: 0.8025 > 0.7904) (Table 4, Figure 1).The binary p-value considering a healthy population further improved the prediction (AUC-ROC: 0.8102 > 0.8025) (Table 4, Figure 1).

[0122] The combined benchmark dataset also includes germline variants from the BRCA1 and BRCA2 genes in particular. Some cancer genes such as BRCA1 and BRCA2 are called cancer predisposition genes. These genes are more abundant in germline variants compared to somatic variants in cancer. Since the binary approach depends on the estimated somatic mutation rate, it is suitable for somatic variants. Thus, variants from germline cancer genes have low accuracy for evaluating the binomial test method. In fact, when filtering out and excluding the variants of BRCA1 and BRCA2 from the combined benchmark dataset, this method was performed even better (Table 4, Figure 20).

Table 4

[0123] (Example 2) Catalog characteristics of amino acid change drivers

[0124] An improved optimal approach incorporating information from a healthy population was applied to 535 selected cancer genes (see method). With this approach, 10,866 variants suspected of amino acid changes were identified as pathogenic with an FDR-adjusted p-value threshold of 0.1.

[0125] Some tools, such as the first binomial tool and the structural clustering tool, predict the pathogenicity of variants according to the positions of amino acids within genes and mark all different variants at these positions as pathogenic variants. However, since some amino acid variants still retain the characteristics of the reference amino acid, gene positioning is not sufficient to define the pathogenic state of the variant. All variants of amino acid changes in the driver catalog were grouped according to the gene positions, and it was mapped which of the suspicious drivers were at the same positions as other drivers. This analysis shows that most variants (71%, n = 7,669) are unique drivers at their gene positions, and it is shown that one-fifth of these positions also have passenger variants (Table 5). Passengers were defined as variants that appear at least once in the healthy population and appear in tumors as predicted under the null binomial distribution assumption (see methods). For the other 29% of the variants, there may be at least one additional driver per position. In this group, a larger number of drivers per position are associated with fewer passengers found at these positions, suggesting that these positions are very important and less susceptible to change (Table 5). For each position, the number of passengers was calculated from all possible non-driver amino acid changes. This analysis showed that this association is not due to a small number of possible amino acid changes remaining after excluding the drivers at those positions (Table 5).

[0126] Looking globally at the number and types of drivers for each gene, it is shown that tumor suppressor genes (TSGs) had a greater number of drivers compared to the smaller number in oncogenes (OGs) (Figure 2). Regarding the types of variants, TSGs have both missense and nonsense drivers, while oncogenes mainly have missense drivers and, in rare cases, also a small number of nonsense drivers (p-value < 2.2e-16, Pearson's chi-square test) (Figure 2, Figure 21).

[0127] The catalog of variants was queried in relation to a publicly available clinical annotation database. In ClinVar, which is not cancer-specific, approximately three-quarters of the variants identified by this approach were absent: 17% were classified as "pathogenic / likely pathogenic"; 6.6% were classified as "uncertain or conflicting", while only 0.2% (n = 24) were classified as "benign / likely benign" (Figure 3). Most (80%) of these "benign" variants were submitted by a single submitter, suggesting that clinical curation in ClinVar is not well established. This assay confirms the high specificity of the catalog. In the Cancer Genome Interpreter, a cancer-specific source of pathogenic variants consisting of three public sources (ClinVar, OncoKB, and DoCM), 88.8% of the variants identified by this approach were not reported, and 11.2% were classified as pathogenic in cancer (Figure 4). This assay highlights the ability of the approach to identify many new cancer-related somatic pathogenic variants.

Table 5-1

Table 5-2

[0128] (Example 3) Correlation of driver TVA with experimental studies

[0129] The question of whether there is a quantitative relationship between the excessive retention rate of variants and their functional activity was investigated. The p-value is optimally used to measure statistical significance rather than a quantitative measurement. Therefore, using the same parameters as the binomial test, a measure or statistic was defined to measure the selective advantage of variants over cancer cells. This statistic is called the "tumor variant amplitude" (TVA), which is

Number

[0130] Deep mutational scanning (DMS) experiments are a useful source for quantifying the effects of variants. Recent studies have evaluated many variant effect predictors by their statistical correlation with DMS experiments. Data were collected from five DMS studies conducted on cancer genes with many known somatic pathogenic variants, which included large libraries of variants. TP53 was the subject of three of these studies, and PTEN was the subject of the other two studies. Each of these studies differed in the experimental platform used, the properties of the protein of interest, the types of changes included, and the focus on protein domains. All of these differences result in specific limitations in all of the studies (Table 1). Spearman correlations were calculated between DMS experiment scores and both raw TVA and attributed TVA. For PTEN in the Mighell study, only highly reliable variants were used. One TP53 study had three scores representing three different experimental measurements, so each score was used separately. For comparison, Spearman correlations were also calculated for 30 variant predictors from dbNSFP and the recently published Evolutionary model of Variant Effect (EVE) scores (see Materials and Methods). The EVE score is an improvement of the DeepSequence tool and was ranked first in statistical correlation with DMS experiments in a recent comparison of many variant effect predictors. Analysis of all of these studies showed a moderate to strong correlation (ρ = 0.33–0.77, Spearman correlation) between attributed TVA and cancer-related DMS experiment scores, and an even stronger correlation (ρ = 0.38–0.79, Spearman correlation) for raw TVA in all DMS studies (Figure 5). In comparison with the 31 predictors, attributed TVA was ranked first in four of the seven DMS scores examined, whereas in the remaining three scores, this was ranked second, seventh, and ninth (Figure 5). In the Kato study, attributed TVA was ranked second after the EVE score, but raw TVA was much higher than the EVE score.It should be noted that one of the PTEN studies in which TVA was ranked 9th is considered to have low accuracy because it measured the stability of proteins that are not the same as functional activity. The first score of Giacomelli for TP53, in which TVA was ranked 7th, is one of three assays from the same paper known as insufficient screening for nonsense variants and variants located outside the DNA-binding domain (Figure 6). Since the first score is based on cancer cells with wild-type TP53 as compared to the other two scores based on cancer cells with null TP53, there appears to be a difference in the performance of TVA among the three Giacomelli scores. This examines the dominant negative effect of mutant TP53 relative to that of endogenous TP53. The wild-type p53 protein in the cells of the first score is less susceptible to the influence of truncated p53 proteins or p53 proteins having drivers in the tetramerization domain. This reduction occurs because the wild-type p53 protein does not form non-functional tetramers with mutant p53, and thus only forms tetramers of wild-type p53, resulting in false negative values in the first score. Indeed, for missense variants in the DNA-binding domain only, the TVA correlation with Giacomelli's first score is much stronger (ρ = 0.72 > 0.53), and TVA was ranked first compared to all 31 predictors (Figure 5). The score distributions of variants vary among DMS studies. Some are more polarized while others have a broader distribution of values. When the data is polarized to the maximum and minimum values, the dichotomous approach of drivers and passengers is strengthened, while a broad distribution of values is more suitable for the spectral effect approach. For TP53, for example, Kotler's score (Figure 7) is more polarized, while Kato's score (Figure 23) and Giacomelli's score (Figures 6, 24, 25) are more spectral. For spectral scores, the distribution includes one extreme of neutral variants with normal protein function, one extreme of pathogenic variants with abnormal protein function, and many intermediate variants.A good correlation was seen in the gap of the intermediate variant between the two extremes of the TVA and DMS score distributions. This suggests that the relative intermediate occupancy of these variants can be explained by the partial protein functions caused by weak / moderate drivers, while the two extremes represent functional and non-functional protein variants associated with passengers and strong drivers, respectively. These weak to moderate drivers are part of the long tail of drivers that can be discovered by this approach (Figure 22). Some deviation can be found in each MDS assay score and TVA graph (further information and analysis can be found in other figures).

[0131] (Example 4) Overall survival of the TVA subgroup in prognostic genes

[0132] The occurrence of variants in certain cancer genes can function as prognostic indicators. One such gene is the TP53 gene, which has been associated with poor prognosis in various cancer types. To avoid ambiguity, tumors with more than one variant were excluded. All TCGA samples with one unique TP53 variant were divided into four groups: non-driver, weak driver, moderate driver, and strong driver according to their TVA values and binary test catalog labels (see Materials and Methods). These groups were compared to a control group of patients with wild-type TP53 (Table 6, Figure 8). This analysis showed clear overall survival (OS) curves for each TP53 variant group, which correlated well with the intensity of the variant estimated by TVA. Non-drivers and weak drivers had the highest OS among all TP53 groups. Non-drivers showed no significant difference compared to other groups due to a small sample size (n = 32), while patients with weak drivers had a small sample size (n = 77) but had a statistically significant better survival rate compared to patients with moderate and strong drivers (p-values = 0.03 and 0.005, log-rank test respectively). Both non-drivers and weak drivers were equivalent to the OS curve of patients with wild-type TP53 (p-values = 0.88 and 0.46, log-rank test respectively). Patients with moderate drivers had a worse OS compared to weak drivers (p-value = 0.02, log-rank test), wild-type (p-value = 1.6e-14, log-rank test), and non-drivers (p-value = 0.37, log-rank test). Patients with strong drivers had the worst OS among all groups (p-values = 3.5e-14 and 0.0047 for wild-type and weak drivers respectively, log-rank test), with marginal significance compared to moderate drivers (p-value = 0.07, log-rank test).

[0133] The relationship between the continuous values of TVA and OS was investigated for patients with any single TP53 variant. Multivariate analysis was performed, including age at diagnosis, gender, and cancer type. A strong effect of TVA values on OS was found, and among the other variables tested, higher TVA was associated with shorter OS (HR = 1.35, p-value = 0.000478) (Figure 9). Note that Figure 9 uses the hazard ratio, not the odds ratio.

[0134] Similar trends are predicted to be obtained for other genes related to prognosis in specific cancers, but the sample sizes of TCGA data are mostly insufficient due to the tumor type specificity of genes or the low frequency of mutations. For example, patients with low-grade glioma (LGG) with EGFR have poor OS. In the LGG survival analysis, all groups had very small sample sizes (non-driver n = 3, weak driver n = 11, moderate driver n = 16, strong driver n = 0), and non-drivers tended to have a survival curve as good as that of the wild-type EGFR group (p-value = 0.324, log-rank test), which was distinguished from the poor prognosis of weak drivers (p-value = 0.04, log-rank test) and moderate drivers (p-value = 0.08, log-rank test). Both weak drivers and moderate drivers were different compared to the wild-type EGFR group (p-value < 1e-13, log-rank test), and there was no clear distinction between the groups of individual drivers (p-value = 0.198, log-rank test) (Table 2, Figure 31).

Table 6

[0135] (Example 5) Drug sensitivity by TVA

[0136] Rare pathogenic variants have become important in the inter-individual variability of drug response. Identification of these variants and interpretation of their pathogenicity are essential for pharmacogenetic prediction. The Genomics of Drug Sensitivity in Cancer Project (GDSC) is a public database containing information on the responses of a large number of human cancer cell lines to a wide range of anti-cancer drugs. In the analysis, the recently released GDSC2 dataset was used, which is considered an improved and more accurate source compared to the previous version. GDSC2 includes 809 cell lines and 198 compounds examined by 135,242 IC50 calculations. The relationship between genomic features and drug response was analyzed from the GDSC analysis of variance model meeting certain criteria (see Materials and Methods). To further investigate the relationship between TVA of each variant and drug response, all variants were divided into the following subgroups: (i) non-driver (ii) weak driver (iii) moderate driver (iv) strong driver (v) very strong driver and (vi) wild type (see Materials and Methods). Some of the drugs tested in GDSC2 directly affect the proteins translated from the alterations of the related cancer genes, while others indirectly affect through cancer gene pathways (upstream or downstream of the genes).

[0137] The PIK3CA gene encodes the catalytic subunit of PI3K. A strong association was found between the TVA subgroups of PIK3CA variants and the responses to two different PI3K inhibitors (Figures 10 and 11). On the other hand, the TVA subgroups of BRAF variants had different associations with various BRAF inhibitors. For the PLX4720 inhibitor (vemurafenib precursor compound), only the "very strong driver" group had a clearly low IC50, while all other groups were equivalent to each other (Figure 12). The "very strong driver" group included the V600E class I variants, and all other driver groups included both class II and III BRAF variants. This inhibitor is known to act only on class I, RAS-independent monomers and not on class II and III variants. For another BRAF inhibitor, dabrafenib, the TVA subgroups of BRAF variants were associated with drug response, except for two cell lines in the "strong driver" group (Figure 13). Indeed, dabrafenib has been shown to have a partial response against tumors with BRAF non-class I variants.

[0138] Regarding indirect inhibitors that affect downstream of the gene, the association with the pathogenicity of the variant may be related to (i) the number of genes between the mutated gene and the drug target gene in the pathway, and (ii) the dispersion of the effects of the mutated gene on multiple pathways. Since PTEN is a major negative regulator of the PI3K-AKT pathway, it is reasonable that the pathogenicity of the variant is associated with AKT inhibitors. A weak association was identified between the TVA subgroup of PTEN and AKT inhibitors, excluding one outlier cell line with the well-known driver variant R130G in the "strong driver" group (Figure 14). On the other hand, the TVA subgroup of NRAS variants associated with MEK inhibitors differed only between drivers and non-drivers, and there was no difference among all driver subgroups (Figure 15). NRAS has three major downstream effector pathways, and RAF-MEK-ERK is only one of them. This dispersion and the distance of genes in the pathway may be the reasons for the low association with the pathogenicity of NRAS variants. For indirect inhibitors upstream of the gene, a worse response can be predicted for stronger drivers of the gene. Indeed, a weak association was identified between the TVA subgroup of KRAS and BTK inhibitors (Figure 16). On the other hand, the association of the TVA subgroup of TP53 variants with MDM2 inhibitors was only between any TP53 variant and wild-type TP53, and there was no clear difference among all TP53 variant subgroups (Figure 17).

[0139] (Example 6) Correlation of TVA values of POLE variants with the number of tumor variants

[0140] The POLE gene encodes the catalytic subunit of DNA polymerase ε, which is involved in DNA repair and chromosomal DNA replication. Driver variants in DNA polymerase ε cause hyper-mutant cancers. Different driver variants of POLE induce different mutational signatures. The three most frequent pathogenic variants are P286R, V411L, and S459F, each associated with a different POLE signature, namely SBS10a, SBS10b, and SBS28, respectively. The tumor mutation burden (TMB) for some samples with POLE variants is low and equivalent to tumors without POLE variants, while for other POLE variants the TMB is high. This indicates that some POLE variants may be passengers. Recent studies have investigated all POLE variants in TCGA endometrial cancer samples and mapped the pathogenic variants. Indeed, the catalog contains almost all (10 / 11) of the pathogenic variants predicted in the present disclosure, and the deletion variant just exceeds the adjusted p-value threshold (0.11).

[0141] POLE variants are typically dichotomized as pathogenic or non-pathogenic, and few studies have investigated the effect size of each pathogenic variant on total TMB. In TCGA endometrial cancer, the correlation between TMB and POLE variants was examined (since multiple POLE variants may coexist in a single sample, the POLE variant with the highest TVA value was selected in these cases). This analysis (Figure 18) showed a positive correlation (ρ = 0.5, p = 3.39e-06, Spearman correlation) between the TMB of the samples and the TVA values of the POLE variants. Most samples with POLE variants having high TMB and low TVA have microsatellite instability (MSI) according to the high "MSI sensor score". MSI itself causes numerous variants due to defective DNA mismatch repair, which accounts for the high number of variants in samples with low POLE TVA values. The coexistence of known driver variants of POLE and microsatellite instability is relatively rare, and this also seems to apply to samples with high POLE TVA values in the analysis. In samples with POLE driver variants, the POLE-related signature was also further enriched. The effect of different POLE drivers on only the number of POLE-related variants was examined. For this analysis, samples were grouped according to TVA into non-driver, weak driver, moderate driver, and strong driver of POLE (see Materials and Methods). For each tumor, the number of POLE-related variants according to the POLE signature was tabulated (see Materials and Methods). This analysis confirmed a distinction between different TVA groups with statistical significance (Figure 19). By using only variants from the POLE-related signature, the true effect of each driver could be seen more clearly without hiding other reasons for the high number of variants such as MSI and MMR. The same correlation between variant frequency and mutation rate was reported in a recent yeast assay, but only for variants in the DNA binding cleft of POLE.

[0142] (Example 7) Analysis of variant mutation rate

[0143] In any study, there are variants that deviate from the correlation. This deviation can most often be explained by the experimental methodological limitations of the assay or the statistical limitations of TVA. One group of exceptions is variants with high TVA values and normal functional scores. An example of a methodological limitation is TP53 E294X, a known nonsense driver that has a high TVA value (2.97), but whose Giacomelli's first score predicts this as normal activity (0.4) due to methodological limitations as presented above. An example of a statistical limitation is seen in TP53 D391A, a variant that is predicted as having normal activity in all experimental scores but has a moderate TVA value (1.6). This is caused by a very low mutation rate and, as predicted, the variant is not statistically significant in the adjusted binomial test (raw p-value = 0.024, adjusted p-value = 1.0). The other group of exceptions is variants with low TVA values but with experimental scores of loss of function. An example of a methodological limitation is seen in TP53 I232L, a variant that has a zero attributable TVA value and is predicted to have normal function in the Giacomelli and Kotler scores but is predicted to have a loss-of-function score in Kato's score. This discrepancy may be due to Kato's yeast model compared to human cell tissues in all other assays. Examples of statistical limitations are many variants with very low TVA, some with loss-of-function scores and some with normal function as measured by Giacomelli's second score (Figure 28). This can be caused by variants at positions with low mutation rates. In the low TVA variant group, the mutation rate of the loss-of-function group was found to be significantly lower than that of the normal function group (p-value = 2.9e-9, t-test) (Figure 29). Therefore, variants at positions with low mutation rates simply do not have sufficient power of detection to reach statistical significance for weak drivers, which may be the cause of the discrepancy (Figure 30).

[0144] (Example 8) HRAS and CS

[0145] Costello syndrome (CS) is a rare genetic disorder caused by mutations in the HRAS gene. This disorder is characterized by distinctive facial features, short stature, and an increased risk of certain types of cancer (PMID: 16170316). The TVA distribution of all known germline RASopathies variants labeled by HGMD was analyzed. Most of these variants are CS. These variants were compared with those identified as somatic in CSD, and the variants were divided into drivers with an adjusted p-value of less than 0.1 in a binomial test and variants that were not significant.

[0146] As predicted, the TVA values were correlated with the groups. The driver group not labeled in HGMD had the highest TVA values. The second highest values were in the CS group, and the third highest were in the group with more minor RASopathy syndromes (Figure 32). Variants with TVA values above 2.5 are well-known hot-spot drivers in cancer but are rarely seen in patients with RASopathy. Two of the CS variants with such TVA levels are not classical CS variants. The first variant was found in a fetus that died due to hydrops fetalis (PMID: 33027564); the second variant was found in two cases, a fetus with hydrops that died 15 days later (PMID: 32732226), and a mosaic patient (PMID: 34109654) who was not severely affected by this strong variant. Mosaic RASopathy is known to cause well-defined defects restricted to specific tissues (PMID: 30007125). In contrast, most CS variants have TVA values in the range between 1 and 2, and many were classified as drivers by a binomial test.

[0147] CS is typically associated with amino acid variants at positions 12 / 13, while variants at other amino acid positions do not show obvious symptoms (PMID: 28328122). Subgroup analysis of HGMD variants based on TVA found that variants at positions 12 / 13 have higher TVA values than variants at other positions (Figure 33). Higher TVA reflects a higher selection for cancer, which is related to a stronger effect on the protein. Therefore, CS patients at positions 12 / 13 show more classical symptoms, while weaker variants show milder symptoms.

[0148] Therefore, TVA values can stratify the risk of developing cancer among different mutations associated with CS. This identification will contribute to individualized follow-up of patients.

[0149] Summary / Discussion

[0150] In the examples, a catalog of 10,866 driver variants was created from 535 cancer genes based on the adjusted p-values of the binary tests, and a new measure called TVA, which represents the selective power for each variant, was calculated. These findings indicate that TVA is highly correlated with the strength and magnitude of the biological activity of driver variants in many different laboratories and clinical validations. TVA was highly correlated with the functional scores of five different DMS experiments that measured the effects of different variants in the TP53 and PTEN genes. This was also superior to the 31 calculated predictors in most studies. This high correlation suggests that TVA better represents the pathogenicity of cancer than other calculated scores and can therefore be used as a measure of the pathogenicity and strength of the biological activity of driver variants with respect to cancer variants. In pharmacological data, TVA was correlated with drug sensitivity in several cancer genes that are directly or indirectly affected by these drugs. Therefore, TVA can contribute to predicting drug responses to non-classical driver variants. The positive correlation of TVA was also shown in two clinical cases: (i) in the POLE gene, TVA was positively correlated with the number of tumor variants associated with POLE (following the genomic context signature); (ii) in the TP53 gene, TVA was positively correlated with overall survival both in a subgroup of TVA and as a continuous parameter.

[0151] This disclosure is novel in both the amount of a given driver variant and the quantitative measure of the effect of the cancer variant using TVA. This has been extensively validated by data from many different sources and represents the strength and reliability of TVA. This finding reinforces the paradigm that the pathogenicity of variants is much more complex than a dichotomous classification into drivers and passengers, and that the effect of variants on quantification methods can be useful for clinical purposes. All validations have demonstrated that TVA can be used for comparison of variants in the same gene. TVA can also be used for comparison of variants in different genes to measure the variant selection power in the same way for all cancer genes, in contrast to methods based on many different DMS data for each gene. Thus, because positive selection can arise from many different mechanisms, TVA may be well-suited for pathogenicity prediction regardless of gene-specific mechanisms.

[0152] Several conceptual approaches have been used to quantify the effects of variants. Some attempt to infer properties of specific mechanisms such as protein stability, while others predict the combined effects of multiple mechanisms. Mechanism-specific approaches are useful for differentiating and explaining the pathogenic mechanisms of drivers, but doing so limits the predictive power of other mechanisms. More general approaches have the advantage of capturing many biologically relevant effects of variants and potentially increasing the accuracy of pathogenicity prediction. Several studies have presented different implementations of the general approach: DMS experiments on selected proteins, supervised machine learning on DMS data using biochemical, structural, and sequence-based features, unsupervised machine learning based on context-dependent constraints in biological sequences, and selection intensity based on cancer cell lineages. Every approach has its own limitations: (i) DMS studies are expensive, time-consuming, and limited to specific genes per study; (ii) supervised machine learning approaches such as Envision are trained on a few selected DMS studies that are not comprehensive enough and need to normalize scores from many genes with different variant effect metrics and protein properties. Compared with other predictors, the Envision tool was found to yield moderate overall correlation performance for human DMS data, although it was trained for that purpose; (iii) unsupervised tools based on context-dependent constraints such as EVE and DeepSequence lack information on many protein positions and nonsense mutations for methodological reasons. This can lead to misinterpretations for variants that affect RNA such as splicing variants and does not work well for some genes, as shown in the EVE paper; (iv) disclosures based on the “selection intensity” of somatic variants in cancer cell lineages include only a few cancer samples in their calculations, are usually unnecessary, divide the predictions by cancer type, mainly focus on known strong drivers, and have not validated their findings in any clinical setting.In the present disclosure, relying on a large number of sequenced cancer samples including information from a database of healthy populations, using a binomial test threshold for the statistical significance of pathogenic variants, predicting the effects of variants on 535 cancer genes, and validating the quantitative effects of variants by a number of laboratories and clinical areas. In addition, the TVA scale is independent of a specific institution and provides higher accuracy for variants that affect RNA. For example, TP53 E224D has a TVA value of 2.1 and is known to be harmful to TP53 splicing.

[0153] Typically, tumors carry many variants, and it is important to determine which are drivers and which are more important for tumor survival. As more therapies are developed targeting more cancer genes, it becomes important not only to identify pathogenic variants but also to prioritize which variants are more important for tumor survival. The catalog and TVA can be used both to identify driver variants and to prioritize them according to their selective variant effects. This prioritization can contribute not only to prognosis but also to the selection of appropriate combination therapies for the more important driver variants of the tumor. This method may be particularly suitable for the evaluation of different gene variants as all calculations are based on selection power.

[0154] From the foregoing written description of the invention, those skilled in the art will be able to make and use what is considered at present to be the best mode. However, those skilled in the art will understand and recognize the existence of alternatives, adaptations, variations, combinations, and equivalents to the specific embodiments, methods, and examples herein. Those skilled in the art will recognize that the disclosure is illustrative only and that various changes may be made within the scope of the invention. Additionally, although a particular feature of the present teachings may be disclosed with respect to only one of a plurality of implementations, such a feature may be combined with one or more other features of one or more other implementations as desired and advantageous for any given or particular function. Further, as long as the terms "including," "includes," "having," "has," "with," or variations thereof are used in either the detailed description or the claims, such terms are intended to be inclusive in the same manner as the term "comprising."

[0155] Other embodiments of the present teachings will be apparent to those skilled in the art from consideration of the specification and practice of the teachings disclosed herein. Accordingly, the invention should not be limited by the described embodiments, methods, and examples, but should be limited only by all embodiments and methods within the scope and spirit of the invention. Accordingly, the invention is not limited to the specific embodiments shown herein but is limited only by the following claims.

Claims

1. A system for quantitatively evaluating the biological effect of at least one target gene variant, for use with a user device, Measuring device and; Processor and; A memory accessible by the processor and storing computer program instructions, wherein when the computer program instructions are executed by the processor, A step of measuring the number of occurrences of the at least one gene variant using the measuring device; The processor includes the step of analyzing a genome database to determine the mutation rate for at least one gene variant; The processor performs the step of determining the observed number of occurrences of the at least one gene variant in the database; The processor performs the step of calculating the predicted number of occurrences of the at least one gene variant based on the mutation rate and the observed number of occurrences; The processor performs the step of calculating a predictor associated with the at least one gene variant based on the mutation rate, the observed number of occurrences, and the predicted number of occurrences; The processor comprises the steps of generating a quantitative evaluation of the biological effect of the at least one gene variant using the predictor; and Steps to transmit the predictive factors and the quantitative evaluation to the user device. Execute the method, memory and A system that includes this.

2. The system according to claim 1, wherein the quantitative evaluation includes prognosis, risk of developing cancer, or treatment response.

3. The system according to claim 1, wherein the predictor includes tumor variant amplitude (TVA), and the TVA is equal to the logarithm of the ratio obtained by dividing the observed occurrences of the at least one gene variant in the genome database by the predicted occurrences of the at least one gene variant in the genome database.

4. Before analyzing the genome database, the processor filters the genome database to avoid duplication of samples from the same subject, and also, Genome coordinates for each entry; Nucleotide changes for each entry; The somatic cell state of each entry; or Cancer type for each entry The system according to claim 1, wherein the genome database is filtered using at least one of the following.

5. The aforementioned quantitative evaluation, A step of comparing multiple drug therapies for the tumor with gene variants present in the tumor, Based on the above comparison, the step is to identify a drug therapy selected from among the multiple drug therapies for use on the target tumor; Based on the above comparison, a step of predicting the likely response of the target tumor to the selected drug therapy, and The system according to claim 1, including the following:

6. The system according to claim 5, wherein the processor identifies the selected drug therapy from among the plurality of drug therapies by prioritizing the gene variants based on the classification of the gene variants and based on the TVA.

7. The aforementioned quantitative evaluation, The steps include comparing the target germline DNA with a database of gene variants and cancer risk; Based on the above comparison, the steps involve quantifying the risk of the subject developing cancer and The system according to claim 1, including the following:

8. The aforementioned quantitative evaluation, The steps include: comparing the target tumor DNA with a database of gene variants and tumor mutations; Based on the above comparison, the steps include quantifying the prognosis of the subject and The system according to claim 1, including the following:

9. The system according to claim 1, wherein the processor further determines a diagnosis using the predictors and artificial intelligence model.

Images