KR102182091B1 - Prediction method for resistance to immunotherapeutic agent and analysis apparatus - Google Patents

Abstract

The present invention is to provide a technique for rapidly predicting the resistance of an anticancer immuno-therapeutic agent of a specific patient. According to the present invention, a method for predicting resistance to an anticancer immuno-therapeutic agent comprises the steps of: receiving, by an analysis device, genome data of a sample; inputting, by the analysis device, the genome data to a previously learned classifier; and predicting, by the analysis device, resistance to the anticancer immuno-therapeutic agent with respect to the sample by the analysis device based on output information of the classifier. The classifier predicts resistance to the anticancer immuno-therapeutic agent based on characteristics of a functional mutation related sequence caused by the tumor.

Description

Method and analysis device for predicting resistance to anticancer drugs of immunotherapy TECHNICAL FIELD [PREDICTION METHOD FOR RESISTANCE TO IMMUNOTHERAPEUTIC AGENT AND ANALYSIS APPARATUS}

The technology to be described below relates to a technique for predicting resistance to cancer-containing immunological agents.

Unlike existing anticancer drugs that attack cancer itself, cancer immunotherapy is a therapeutic drug that induces immune cells to selectively attack only cancer cells by injecting artificial immune proteins into the body to stimulate the immune system. Immune anticancer drugs include immune checkpoint inhibitors (CTLA4 inhibitor, PD-1 inhibitor, PD-L1 inhibitor), immune cell therapy, and immunoviral therapy.

Cancer cells evade immunity by using the immune checkpoint of immune cells. Immune checkpoint inhibitors suppress immune checkpoints and kill cancer cells by activating immune cells in the body. However, immune checkpoint inhibitors are not responsive to all patients. Therefore, it is important to discover biomarkers that can predict responsiveness to immunotherapy.

Korean Patent Publication No. 10-2019-0094710

Tumor mutation burden (TMB) is a representative biomarker that predicts the responsiveness to immunotherapy. When TMB is high, epitopes of neoantigens are well recognized by T cells, and it is known that the responsiveness to immune chemotherapy is good. However, a large proportion of cancer cell mutations are not immunogenic, and increased functional mutations may lead to resistance to treatment.

The technique described below is intended to provide a technique for predicting the resistance of an immune anticancer agent (immune checkpoint inhibitor). In addition, the technique described below is intended to provide a technique for discovering a marker that predicts the resistance of an anticancer drug.

The method of predicting resistance to anticancer drugs includes the steps of: receiving, by an analysis device, genome data of a sample, by the analysis device, inputting the genome data to a pre-learned classifier, and by the analysis device And predicting the resistance of the anticancer agent to the sample based on the output information. The classifier predicts the resistance of an immune anticancer drug based on the characteristics of a sequence associated with a functional mutation caused by a cancer.

The analysis device for predicting resistance to anticancer drugs is an input device that receives genome data of a sample, and a classifier that predicts the resistance of anticancer drugs based on the characteristics of the sequence associated with functional mutations caused by tumors. And a storage device that stores and a computing device that inputs the genome data to the classifier to predict resistance of the immuno-anticancer agent to the sample.

The technique described below can quickly predict the resistance of a specific patient to an anticancer drug by using a learning model. Therefore, the technology described below can contribute to customized treatment for each patient. Further, the technology described below may contribute to customized treatment by discovering a marker that predicts resistance to an immuno-anticancer agent targeting a specific disease or a specific cohort.

1 is an example of a system for predicting resistance to anticancer drugs.
2 is an example of a process of training a model for predicting immune anticancer drug resistance.
3 is an example of a process for predicting resistance to anticancer drugs.
4 is an example of a process of discovering a marker for determining resistance to an anticancer drug.
5 is an example of the structure of an analysis device for predicting resistance to anticancer drugs.
6 is a result of evaluating a model for predicting immune anticancer drug resistance.

The technology to be described below may be modified in various ways and may have various embodiments, and specific embodiments will be illustrated in the drawings and described in detail. However, this is not intended to limit the technology to be described below with respect to a specific embodiment, and it should be understood to include all changes, equivalents, and substitutes included in the spirit and scope of the technology described below.

Terms such as 1st, 2nd, A, B, etc. may be used to describe various components, but the components are not limited by the above terms, only for the purpose of distinguishing one component from other components. Is only used. For example, a first component may be referred to as a second component, and similarly, a second component may be referred to as a first component without departing from the scope of the rights of the technology described below. The term and/or includes a combination of a plurality of related listed items or any of a plurality of related listed items.

In terms of the terms used in the present specification, expressions in the singular should be understood as including plural expressions unless clearly interpreted differently in context, and terms such as "includes" are specified features, numbers, steps, actions, and components. It is to be understood that the presence or addition of one or more other features or numbers, step-acting components, parts or combinations thereof is not meant to imply the presence of, parts, or combinations thereof.

In addition, in performing the method or operation method, each of the processes constituting the method may occur differently from the specified order unless a specific order is clearly stated in the context. That is, each process may occur in the same order as the specified order, may be performed substantially simultaneously, or may be performed in the reverse order.

Hereinafter, terms used in the description will be described.

Antigens are substances that induce an immune response.

Neoantigens are antigens with alterations that occur through mutations in tumor cells or post-translational modifications specific to tumor cells. The neoantigen may comprise a polypeptide sequence or a nucleotide sequence. Mutations can be frame shifted or non-lattice shifted indels, missense or nonsense substitutions, splice site alterations, genomic rearrangements or gene fusions, or any genomic or expression alterations that result in a newborn ORF. It may include. Mutations can also include splice variants. Post-translational modifications specific to tumor cells can include abnormal phosphorylation. Post-translational modifications specific to tumor cells can also include proteasome-generated spliced antigens.

Exomes are a subset of the genome that encodes proteins. An exome may refer to a cell, a group of cells, or a collection of exons present in an individual.

An epitope may refer to a specific portion of an antigen to which an antibody or T-cell receptor usually binds.

Immunogenicity is the ability to elicit an immune response through T cells, B cells, or both.

Tolerance, immune tolerance, or resistance is a state of immune non-reactivity to one or more antigens.

A sample or sample means a single cell or multiple cells, cell fragments, body fluids, etc. collected from an individual to be analyzed.

Subjects include cells, tissues or organisms. The entity is primarily intended for humans, but is not limited thereto.

Genomic data or genome information refers to genetic information that is calculated by analyzing a sample. For example, genomic data is a base sequence obtained from deoxyribonucleic acid (DNA), ribonucleic acid (RNA), or protein from cells, tissues, etc., gene expression data, genetic mutation with standard genomic data, DNA methylation ), etc. In general, genomic data includes sequence information obtained by analyzing a specific sample. Genomic data can be obtained in a variety of ways. For example, genome data can be generated through NGS analysis. Genomic data can be expressed as digital data understood by computers.

Machine learning, or learning, is a field of artificial intelligence, which refers to the field in which algorithms are developed so that computers can learn. A machine learning model or learning model means a model developed so that a computer can learn. Learning models include various models such as artificial neural networks and decision trees, depending on the approach.

Ensemble is a generic term for a technique that uses a plurality of learning algorithms in machine learning. Representatively, the ensemble technique includes a bagging technique including a random forest or a boosting technique.

Random forest is a kind of bagging algorithm composed of a combination of CART decision trees. The random forest consists of a plurality of decision trees. Each of the plurality of decision trees is trained in advance by randomly selecting some of the training data and feature variables. In the random forest, each tree individually determines the target variable and then aggregates the decisions of all trees to make a final decision.

1 is an example of a system 100 for predicting resistance to anti-cancer drugs. The analysis device predicts resistance to anticancer drugs. In FIG. 1, the analysis device is shown in the form of a server 130 and a computer terminal 140. The server 130 may provide a service for predicting resistance to anti-cancer drugs on a network. The computer terminal 140 may be connected to a network or analyzed genomic data with an individual device to predict resistance to anti-cancer drugs. The analysis devices 130 and 140 may be implemented in various forms.

The analysis devices 130 and 140 analyze resistance to an anticancer drug by using the genome data. Here, the genome data includes information on the genome sequence. The genome analysis device 110 analyzes a sample and generates genome data. For example, the genome analysis device 110 may be an NGS analysis device. The genome analysis device 110 may store the generated genome data in a separate DB 120.

The genome analysis device 110 generates genome data using a genome library. The genome analysis device 110 may perform whole exome sequencing on the genome library. The genome library can be prepared using a commercial kit. For example, using the AllPrep DNA / RNA Mini Kit (Qiagen, 80204), AllPrep DNA / RNA Micro Kit (Qiagen, 80284), or QIAamp DNA FFPE Tissue Kit (Qiagen, 56404) to generate a genomic library for sequence analysis. I can.

Users (10, 20) can check the result of resistance to anticancer drugs for a specific patient. The user 10 may access the server 130 through a user terminal (PC, smartphone, etc.) and check the analysis result performed by the server 130. The user 20 may check the result of resistance to anticancer drugs through the computer terminal 140 used by the user 20.

Users (10, 20) may be researchers who perform resistance evaluation against anticancer drugs. Alternatively, the users 10 and 20 may be medical staff who consider prescribing an anticancer drug for a specific patient.

The analysis devices 130 and 140 predict resistance to anticancer drugs through genomic data analysis. The analysis devices 130 and 140 predict immune anticancer drug resistance using a learning model prepared in advance. The analysis devices 130 and 140 may use various learning models. For example, the analysis apparatuses 130 and 140 may analyze the resistance to an anticancer drug using an ensemble technique. For convenience of description below, it is assumed that the analysis devices 130 and 140 analyze the resistance of the immune anticancer drug using a random forest model.

The learning model used by the analysis devices 130 and 140 must be prepared in advance. 2 is an example of a process 200 of training a model for predicting immune anticancer drug resistance. The learning model is trained using training data prepared in advance.

The cancer patient cohort contains genomic data from multiple patients. The first cohort of cancer patients is the population. Now you can select your training data by several criteria. Training data can be selected by repeating the process of selecting a group with a certain criterion from the initial population. The process of filtering a population based on a certain criterion is irrelevant.

Several criteria for selecting training data are described. (i) The amount of TMB may be the standard. That is, individuals with more TMBs than the reference value can be selected from the population. (ii) The number of new antigens produced by cancer cells can be a standard. That is, individuals with more new antigens than the reference value in the population can be selected. (iii) Functional mutation can be the criterion. That is, individuals with a degree of functional mutation in the population that are greater than or equal to the reference value may be selected. Meanwhile, functional mutations can be evaluated by various algorithms.

For example, functional mutations can be assessed to the extent that the sequence in which the mutation occurs affects protein function. The degree to which the mutation affects the function of the associated protein may be measured using several solutions or algorithms. Here are some examples.

(i) SIFT (Sorting Intolerant From Tolerant, https://sift.bii.a-star.edu.sg/) quantifies the degree to which amino acid substitution affects protein function. The SIFT score is a quantification of the extent to which mutations affect protein function. (ii) PROVEAN (Protein Variation Effect Analyzer, http://provean.jcvi.org) quantifies the extent to which amino acid substitution or deletion (indel) affects the function of a protein. The PROVEAN score is a quantification of the extent to which mutations affect protein function.

When the SIFT score is equal to or higher than the reference value and the PROVEAN score is equal to or higher than the reference value for the group to be analyzed, the individual may be determined to have a functional mutation equal to or higher than the threshold value.

Meanwhile, clinical data may be used to select training data. In this case, clinical data on individuals constituting the population are premised. Clinical data include information on whether or not you have resistance to actual anticancer drugs. For example, it is possible to filter the training data by selecting an individual with resistance to an anticancer drug based on clinical data from the population.

Referring to FIG. 2, a process of selecting training data will be described. A cohort of cancer patients to be analyzed is obtained (210). For a cohort of cancer patients, a data set of individuals whose number of new antigens is greater than or equal to a reference value is selected (220). For example, a data set having more than 70 new antigens may be selected. Training data in which the degree of functional mutation is greater than or equal to a reference value in the selected set are selected (230). Finally, the random forest model is trained using the prepared training data (240).

The learning model may be prepared using positive training data and negative training data. The training data includes genomic data for a plurality of individuals. Therefore, the positive training data corresponds to the positive training data group, and the negative training data corresponds to the negative training data group. The process of selecting a positive training data group in a cohort of cancer patients is as described above. 2 is an example of selecting data in which the number of new antigens is greater than or equal to the reference value and the degree of functional mutation is greater than or equal to the reference value in a cohort of cancer patients as a positive training data group. The degree of functional mutation can be determined based on the SIFT score and PROVEAN score. The negative training data group consists of training data excluding the positive training data group from the population.

The random forest model may learn genomic data for a plurality of individuals included in the training data. A plurality of decision trees constituting the random forest are trained by randomly selecting training data and randomly selecting feature variables, respectively.

The characteristic variable from which the random forest is learned may be a sequence in which a mutation occurs among genome sequences. That is, the random forest can be learned using a specific sequence section that is highly related to immuno-anticancer resistance without using the entire sequence. For random forest learning, the genome sequence can be converted into information in a certain vector form in advance.

3 is an example of a process 300 for predicting resistance to anticancer drugs. 3 is an example of predicting immune anticancer drug resistance using a learning model trained in advance by an analysis device. The analysis device predicts the effect of an anticancer drug in advance for a specific patient.

The analysis device receives the genome data of the sample (310). The sample refers to an individual (patient) who wants to determine the resistance to an anticancer drug. The sample genome data refers to the genome data of a patient to be analyzed.

The analysis device inputs sample genome data into the model learned in advance. The learning model analyzes the sample genomic data (320). 3 illustrates a random forest model. Each decision tree constituting a random forest makes a decision starting with input data and outputs a final decision result. Referring to FIG. 3, the decision tree A outputs a result that resistance is high, and the decision tree B outputs a result that resistance is low. The random forest considers all of the output results of each decision tree to make a final decision. For example, a random forest can decide its final conclusion according to the principle of majority vote. The analysis device predicts immune anticancer drug resistance to the sample based on the information output from the random forest (330). For example, the analysis device may output information indicating that the patient has high resistance to anticancer drugs. The learning model can be referred to as a classifier by performing a function of classifying resistance to a sample.

4 is an example of a process 400 of discovering a marker for determining resistance to an anticancer drug. Biomarkers can be determined individually for specific patient cohorts or specific patients.

The analysis device receives sample genome data (410). The analysis device inputs sample genome data into a learning model (classifier) and selects candidate data having high resistance to anticancer drugs (420).

The analysis device may detect a candidate gene based on variable importance for the candidate data (430). The importance of a variable is a quantification of the effect of a specific variable on the analysis result of the learning model.

As described above, the analysis device may take the mutation sequence as a variable in the genomic data. In this case, the importance of a variable may be determined based on a result of changing the composition of a specific sequence and inputting data including the changed specific sequence into a learning model. For example, the analysis apparatus changes a sequence, which is a specific variable, in a random order, and inputs a specific variable or input data including a specific variable into a learning model. The sample genome data including the changed variables is referred to as processed sample genome data.

When the processed sample genome data is input to the classifier, the output result may be different compared to the case where the original sample genome data is input to the learning model. The original sample genome data refers to sample genome data without arbitrarily changing the sequence of the sequence. At this time, the degree to which the prediction accuracy varies is the importance of the variable. Variable importance can be quantified by various criteria. In the case of a random forest, variable importance can be calculated based on a plurality of decision trees.

For example, when a plurality of decision trees receives original sample genome data and processed sample genome data, output results may be different. At this time, the number of decision trees with different output results may be a criterion for determining the importance of a variable. For example, when the number of decision trees with different output results is 3 or more, it may be determined that the variable importance of the corresponding variable is high. In this case, the analysis device may detect the variable (sequence) as a candidate gene (430).

An analysis device or a researcher can determine the candidate gene as a marker that identifies the resistance of an immuno-cancer drug.

Further, by further performing an interactional analysis on the protein associated with the candidate gene, the analysis device or the researcher may determine the candidate gene. The related protein refers to a protein that a candidate gene affects, a protein produced by translation of a candidate gene, and the like. The analysis device can determine which mechanism pathway the related protein affects through the analysis of the interactor of the related protein. The analysis device may determine a corresponding candidate gene as a marker when the associated protein affects the expression of resistance to an anticancer drug. Alternatively, the analysis device may determine a corresponding candidate gene as a marker when the associated protein affects the occurrence of cancer.

5 is an example of the structure of the analysis device 500 for predicting resistance to anti-cancer drugs. The analysis device 500 is a device corresponding to the analysis device 130 or 140 of FIG. 1.

The analysis device 500 may predict the resistance of the immune cancer drug using the above-described learning model. The analysis device 500 may be physically implemented in various forms. For example, the analysis device 500 may have a form such as a computer device such as a PC, a server of a network, and a chipset for image processing. The computer device may include a mobile device such as a smart device.

The analysis device 500 includes a storage device 510, a memory 520, an operation device 530, an interface device 540, a communication device 550, and an output device 560.

The storage device 510 stores a classifier that predicts the resistance of an immuno-cancer agent. The classifier must be learned in advance. Furthermore, the storage device 510 may store programs or source codes required for data processing. The storage device 510 may store input dielectric data and predicted resistance data.

The memory 520 may store data and information generated in the process of analyzing the data received by the analysis device 500.

The interface device 540 is a device that receives certain commands and data from the outside. The interface device 540 may receive dielectric data from an input device physically connected or an external storage device. The interface device 540 may receive a learning model for data analysis. The interface device 540 may receive training data, information, and parameter values for training a learning model.

The communication device 550 refers to a component that receives and transmits certain information through a wired or wireless network. The communication device 550 may receive genome data from an external object. The communication device 550 may also receive data for model training. The communication device 550 may transmit information about resistance to an anticancer drug determined for an input sample to an external object.

The communication device 550 to the interface device 540 are devices that receive certain data or commands from the outside. The communication device 550 to the interface device 540 may be referred to as an input device.

The output device 560 is a device that outputs certain information. The output device 560 may output an interface required for a data processing process and an analysis result.

The computing device 530 may predict resistance to an immuno-anticancer drug for input sample genome data using a classifier stored in the storage device 510. The computing device 530 may predict resistance to an anticancer drug by directly or consistently processing a result output from the classifier. The computing device 530 may train a learning model used for predicting resistance to anticancer drugs by using the given training data. The computing device 530 may be a device such as a processor, an AP, or a chip in which a program is embedded that processes data and processes certain operations.

Hereinafter, the process and effect of the researcher creating the model for predicting the above-described immune anticancer drug resistance will be described.

Cohort name references Tumor type Cohort size target Immune checkpoint SMC Lung cancer 122 PD-1/PD-L1 Rizvi Science 348:124 Lung cancer 34 PD-1 Hellmann Cancer Cell 33:843 Lung cancer 75 PD-1 & CTLA-4 Van Allen Science 350:207 Melanoma 110 CTLA-4 Snyder NEJM 371:2189 Melanoma 64 CTLA-4 Roh Sci. Transl. Med. 9:eaah3560 Melanoma 56 PD-1 & CTLA-4 Riaz Cell 171:934 Melanoma 68 PD-1

Table 1 shows the cohorts used in the experiment and development process. The cohort excluding the SMC cohort is the cohort used in previous studies.

The SMC cohort is data provided by domestic hospitals. Detailed information is as follows. We enrolled in 122 patients with advanced non-small cell lung cancer treated with anti-PD-1/PD-L1 from 2014 to 2017 in hospital. The clinical response was evaluated through follow-up for at least 6 months according to the response evaluation criteria of RECIST (Response Evaluation Criteria in Solid Tumours) version 1.1. Responses to immunotherapy were classified as responsive (durable clinical benefit: DCB) or non-responsive (non-durable benefit: NDB). Patients with partial response (PR) or stable disease (SD) or lasting longer than 6 months were considered DCB/responsive. Progressive disease (PD) or SD lasting less than 6 months was considered NDB/non-reactive. Progression-free survival (PFS) was calculated from the start of treatment to the date of progression or death, whichever is earlier. If the patient was alive, it was evaluated on the date of the last follow-up for PFS.

All samples were examined for mutations. Functional mutations were evaluated by SIFT and PROVEAN. Functional mutations were defined as being classified as damaged by SIFT and at the same time classified as deleted by PROVEAN.

In the training data, a random forest was learned by selecting a mutation of a gene with a mutation frequency higher than 5% as a feature or variable. The random forest consisted of 1000 decision trees. The random forest was learned by repeating 5-fold cross validation 10 times in the random forest R package.

In addition, a negative control learning model was created by learning in the same way as a resistance prediction model using the state of a synonymous mutation on the same gene set. The negative control learning model used the same number of features as the resistance prediction model.

In the case of the same cancer, the model was trained by excluding one cohort and integrating the remaining cohorts. Excluding one cohort was used as test data. That is, the results were examined using a model trained with one cohort as input data.

6 is a result of evaluating a model for predicting immune anticancer drug resistance. The red curve is the receiver operating characteristic (ROC) curve for the deletion/damage mutation. The blue curve is the ROC curve for the negative control, synonymous mutation. AUC (area under the curve) refers to the area under the ROC curve, and it is an ideal model when it is 1.

6(A) is an evaluation result of melanoma cohort. The cohort displayed at the top of each graph indicates the cohort used for testing purposes. 6(B) is an evaluation result for a lung cancer cohort. The cohort displayed at the top of each graph indicates the cohort used for testing purposes. For both melanoma and lung cancer, the effect of predicting resistance was better than that of the negative control group.

In addition, the method for predicting resistance to an anticancer drug or a method for discovering biomarkers as described above may be implemented as a program (or application) including an executable algorithm that can be executed on a computer. The program may be provided by being stored in a non-transitory computer readable medium.

The non-transitory readable medium refers to a medium that stores data semi-permanently and can be read by a device, not a medium that stores data for a short moment, such as a register, cache, or memory. Specifically, the above-described various applications or programs may be provided by being stored in a non-transitory readable medium such as a CD, DVD, hard disk, Blu-ray disk, USB, memory card, and ROM.

The present embodiment and the accompanying drawings are merely illustrative of some of the technical ideas included in the above-described technology, and those skilled in the art will be able to easily within the scope of the technical ideas included in the specification and drawings of the above-described technology. It will be apparent that all of the modified examples and specific embodiments that can be inferred are included in the scope of the rights of the above-described technology.

Claims (18)

Receiving, by the analysis device, genome data of the sample;
Inputting, by the analysis device, the genome data into a pre-learned classifier; And
The analysis device comprises the step of predicting the resistance of the anticancer agent to the sample based on the output information of the classifier,
The classifier predicts the resistance of an immune anticancer agent based on the characteristics of a sequence associated with a functional mutation caused by a cancer,
The classifier is trained using a positive training data group having resistance and a negative training data group not having resistance, based on resistance to anticancer drugs,
The positive training data group is a method for predicting resistance to an anticancer agent selected according to the degree to which a tumor-induced mutation affects protein function. The method of claim 1,
The classifier is an ensemble model, a method of predicting resistance to anticancer drugs. The method of claim 1,
The classifier is a random forest (random forest) model, a method of predicting resistance to anticancer drugs. The method of claim 1,
The classifier is a method of predicting resistance to an anticancer drug, which is learned using genomic data of a patient having a tumor-induced neoantigen above a reference value. delete The method of claim 1,
The positive training data group is a method of predicting resistance to an immuno-anticancer drug, which is training data in which the SIFT score and PROVEAN score for quantitatively evaluating the degree to which the amino acid sequence affects protein function is greater than or equal to a reference value. The method of claim 1,
The positive training data group is training data in which the number of neoantigens is greater than or equal to a reference value. Receiving, by the analysis device, genome data of the sample;
Inputting, by the analysis device, the genome data into a pre-learned classifier; And
Including the step of determining a marker capable of determining resistance to the anti-cancer drug based on the output information of the classifier,
The classifier predicts the resistance of an immune anticancer agent based on the characteristics of a sequence associated with a functional mutation caused by a tumor,
The determining of the marker may include detecting a candidate gene having a variable importance greater than or equal to a reference value in the genomic data; And determining the marker among the candidate genes through an interactional analysis of the protein associated with the candidate gene. The method of claim 8,
The classifier is trained using a positive training data group having resistance and a negative training data group not having resistance, based on resistance to anticancer drugs,
The positive training data group is a method of detecting a marker for predicting resistance to an anticancer agent selected according to the degree to which a tumor-induced mutation affects protein function. The method of claim 8,
The positive training data group is a method of detecting a marker for predicting resistance to an immune anticancer drug having a SIFT score and a PROVEAN score for quantitatively evaluating the degree to which an amino acid sequence affects protein function. The method of claim 9,
The positive training data group is training data in which the number of neoantigens is greater than or equal to a reference value. The method of claim 8,
The variable importance is a method of detecting a marker for predicting resistance to an anticancer drug, indicating the degree to which the predicted result of the classifier becomes inaccurate after random permutation of the mutant sequence. An input device for receiving dielectric data of a sample;
A storage device for storing a classifier that predicts resistance of an immuno-anticancer agent based on a characteristic of a sequence associated with a tumor-induced functional mutation; And
Comprising a computing device for predicting the resistance of the immune anticancer agent to the sample by inputting the genome data into the classifier,
The classifier is learned in advance using a positive training data group having resistance based on resistance to an immuno-anticancer drug, and the positive training data group is selected according to the degree to which a tumor-induced mutation affects protein function. An analysis device that predicts resistance to anticancer drugs. The method of claim 13,
The classifier is a random forest (random forest) analysis device for predicting resistance to anti-cancer drugs that are models. delete The method of claim 13,
The positive training data group includes data in which both the SIFT score and the PROVEAN score are positive, which quantitatively evaluates the degree to which the amino acid sequence affects protein function among training data in which the number of neoantigens is higher than the reference value Analysis device that predicts resistance. The method of claim 13,
The computing device is
An immuno-anticancer agent that detects a candidate gene whose variable importance is greater than or equal to a reference value in the genomic data, and determines a resistance marker of the immuno-anti-cancer agent among the candidate genes through interaction analysis of proteins associated with the candidate gene. Analysis device that predicts resistance to A computer-readable recording medium in which a program for executing a method for predicting resistance to an anti-cancer agent according to any one of claims 1 to 4 and 6 to 7 is recorded on a computer.

Images