Method, system for screening of neoantigens and uses thereof

Abstract

The present application relates to a screening method, system and use thereof for a new antigen, and more particularly discloses a screening method, system and use thereof for a new antigen as a diagnostic and / or therapeutic target, the new antigen being derived from a gene that is uniform in all cells within a cancer tissue, and / or the expression of which is essential for the survival of cancer cells.

Description

Technical Field

[0001] This invention relates to a method, system, and use of neoantigens. More specifically, this invention relates to a method, system, and use of neoantigens for screening as diagnostic and / or therapeutic targets, wherein the neoantigens are derived from a homogeneous gene in all cells of cancer tissue, and / or their expression is essential for cancer cell survival. Background Technology

[0002] As one of the most promising anti-cancer therapies, immunotherapy offers a new paradigm for cancer treatment due to its high cure rate expected in responsive patient groups. More than 3,400 clinical trials of immunotherapy agents have been conducted globally, and further research is underway.

[0003] Cancer vaccines, which are immunotherapeutic agents that activate the immune system using cancer-specific antigens, are becoming a core technology in immunotherapy for cancer, particularly in combination with immune checkpoint inhibitors. However, immune checkpoint inhibitors, such as anti-PD-1 / anti-PD-L1, have a practical success rate of only 15% to 20% and involve a process that leads to a loss of overall T-cell immune regulation, thus posing a high risk of side effects such as autoimmune diseases. There remains a need for a therapeutic agent that can improve response rates in cancer patients while minimizing side effects.

[0004] Patient-customized cancer vaccines, which achieve optimal vaccine design targeting patient-specific neoantigens, are being developed as a treatment method to improve cancer patient outcomes and minimize side effects. By inducing the patient's immune response to focus on cancer cell-specific neoantigens, the effectiveness of immunotherapy can be enhanced; therefore, selecting the optimal neoantigen is crucial.

[0005] Cancer cells evolve into a form capable of evading anti-cancer immune responses through an immune-editing process. This evolved population of cancer cells is considered a cause of resistance to immunotherapy and cancer recurrence. There is a need to develop novel antigen targets that can overcome the immune evasion mechanisms caused by the heterogeneity and plasticity of these tumors. Korean Patent Application Publication No. 2018-0107102 discloses a method for identifying and selecting neoantigens for personalized cancer vaccines, but does not disclose or imply any need to overcome the immune evasion of cancer cells or the need to overcome it.

[0006] The inventors have studied strategies to enable cancer cells to evade immune editing mechanisms, making it difficult to detect effective neoantigens as anticancer vaccines, and have invented a method for screening effective neoantigens as diagnostic and / or therapeutic targets. Summary of the Invention

[0007] Technical issues

[0008] The purpose of this invention is to provide a method for screening new antigens that overcomes the immune evasion of cancer cells.

[0009] Another object of the present invention is to provide a system for screening new antigens that overcomes the immune evasion of cancer cells.

[0010] Another object of the present invention is to provide an anticancer vaccine comprising a neoantigen that overcomes immune evasion of cancer cells.

[0011] Another object of the present invention is to provide a composition for predicting the treatment prognosis of cancer patients, comprising a neoantigen that overcomes immune evasion of cancer cells.

[0012] Technical solution

[0013] One aspect of the present invention provides a method for screening novel antigens, comprising:

[0014] Obtain sequencing data from cancer patients, including exome, transcriptome, single-cell transcriptome, peptidome, or whole genome sequencing.

[0015] Select genes that depend on cancer cell survival; and

[0016] Obtain a new antigen derived from the cancer cell survival-dependent gene.

[0017] In one specific embodiment of the present invention, the method for screening neoantigens further includes: determining the binding affinity between the neoantigen and the human leukocyte antigen (HLA) of the antigen-presenting cell.

[0018] In one specific embodiment of the present invention, when the expression level of cancer cell survival-dependent genes is reduced or removed, it can lead to the death of several types of cancer cells or cancer cells derived from the patient.

[0019] In one specific embodiment of the present invention, the cancer cell survival-dependent gene can be a universally dependent gene that is universally essential for the survival of several types of cancer or cancer cells.

[0020] In one specific embodiment of the present invention, the cancer cell survival-dependent gene may be a cancer patient-specific dependent gene essential for the survival of cancer cells derived from the cancer patient.

[0021] In one specific embodiment of the present invention, cancer cell survival-dependent genes may include universally dependent genes or cancer patient-specific dependent genes, or a combination of universally dependent genes and cancer patient-specific dependent genes.

[0022] In a specific embodiment of the present invention, selecting cancer cell survival-dependent genes includes: using a cell survival-dependent prediction model to select cancer cell survival-dependent genes that are essential for cancer cell survival. The cell survival-dependent prediction model is generated by training the relationship between cell gene expression patterns and cell death, and genes that cause cancer cell death based on the reduction or removal of gene expression levels can be selected as cancer cell survival-dependent genes.

[0023] In one specific embodiment of the present invention, the cell survival dependence prediction model can learn from experimental data on the relationship between cell gene expression and cell death based on deep learning.

[0024] In one specific embodiment of the present invention, the relationship between gene expression and cell death can be based on in vitro screening experimental data or computer data regarding whether cancer cell lines are killed due to reduced or removed expression levels of the targeted gene.

[0025] In a specific embodiment of the present invention, the selection of cancer cell survival-dependent genes may further include: determining whether the cancer cell survival-dependent genes selected from the cell survival dependence prediction model are uniformly expressed in all cancer cells obtained from the cancer patient.

[0026] In one specific embodiment of the present invention, obtaining a neoantigen derived from the cancer cell survival-dependent gene may include: comparing sequences from cancer cells and sequences from normal cells based on sequencing data obtained from cancer patients to obtain neoantigens from cancer patients; and collecting neoantigens derived from the cancer cell survival-dependent gene from the obtained neoantigens.

[0027] In one specific embodiment of the present invention, obtaining a neoantigen derived from the cancer cell survival-dependent gene may include: comparing sequences from sequencing data obtained from cancer patients with sequences from sequencing data obtained from normal control groups to obtain a neoantigen from the cancer patient; and collecting neoantigens derived from the cancer cell survival-dependent gene from the obtained neoantigens.

[0028] In one specific embodiment of the present invention, the collection of neoantigens may further include: selecting nonsynonymous mutations in the cancer cell survival-dependent genes.

[0029] In a specific embodiment of the present invention, determining the binding affinity between the neoantigen and the HLA of the antigen-presenting cell includes: inputting the sequence of the neoantigen into a neoantigen binding affinity prediction model to obtain a binding affinity prediction. The neoantigen binding affinity prediction model is used to predict the binding affinity between the antigen peptide and the HLA of the antigen-presenting cell. The neoantigen binding affinity prediction model can be generated by training data on the interaction between the amino acids of the peptide and the amino acids of HLA.

[0030] In one specific embodiment of the present invention, the antigen-presenting cell may be a dendritic cell, a macrophage, a B cell, or a combination thereof.

[0031] In a specific embodiment of the present invention, the HLA may be Major Histocompatibility Class (MHC) I or II.

[0032] In a specific embodiment of the present invention, when the CNN-MHC level between the neoantigen and the HLA of the antigen-presenting cell is >0.5, binding affinity can be determined.

[0033] In one specific embodiment of the present invention, multiple cancer cell survival-dependent genes can be selected based on their necessity for cancer cell survival.

[0034] In one specific embodiment of the present invention, depending on the necessity for cancer cell survival, the first 5 and the last 5, the first 10 and the last 10, or more of the first and more of the last cancer cell survival-dependent genes can be selected respectively.

[0035] In one specific embodiment of the present invention, the neoantigen may be a cancer patient-specific neoantigen.

[0036] Another aspect of the present invention provides a system for screening novel antigens, comprising:

[0037] At least one processor that executes the at least one instruction stored in the memory.

[0038] The processor executes the at least one instruction.

[0039] The relationship between gene expression levels and cell death is trained to generate a cell survival-dependent prediction model, which predicts the dependence of cell survival on gene expression.

[0040] The gene expression profiles of cancer patients are input into the cell survival-dependent prediction model to select cancer cell survival-dependent genes, and neoantigens derived from the cancer cell survival-dependent genes are obtained from the gene expression profiles of cancer patients.

[0041] Based on the interaction between peptides and amino acids on antigen-presenting cells, a new antigen-binding affinity prediction model is generated to predict binding affinity.

[0042] Using the neoantigen binding affinity prediction model, neoantigens with binding affinity to HLA in antigen-presenting cells are selected.

[0043] In one specific embodiment of the present invention, the system can be used to implement a screening method for neoantigens according to a specific embodiment of the present invention.

[0044] In one specific embodiment of the present invention, the cell survival dependence prediction model can select cell survival dependent genes or predict the cell survival dependence of neoantigen-derived genes.

[0045] In a specific embodiment of the present invention, the cell survival dependence prediction model is generated by training the relationship between the expression level of cell genes and cell death, and genes that cause cancer cell death by reducing or removing gene expression levels can be selected as cancer cell survival dependent genes.

[0046] In one specific embodiment of the present invention, the cancer cell survival-dependent gene can be a universally dependent gene that is universally essential for the survival of several types of cancer or cancer cells.

[0047] In one specific embodiment of the present invention, the cancer cell survival-dependent gene may be a cancer patient-specific dependent gene essential for the survival of cancer cells derived from a specific cancer patient.

[0048] In one specific embodiment of the present invention, cancer cell survival-dependent genes may include universally dependent genes or cancer patient-specific dependent genes, or a combination of universally dependent genes and cancer patient-specific dependent genes.

[0049] In one specific embodiment of the present invention, the processor can select a neoantigen with binding affinity when the CNN-MHC level between the neoantigen and the HLA of the antigen-presenting cell is >0.5 by executing the at least one instruction.

[0050] In one specific embodiment of the present invention, the processor can learn the relationship between gene expression and cell death and the relationship between the binding affinity of neoantigens and HLA of antigen-presenting cells by executing the at least one instruction.

[0051] In one specific embodiment of the present invention, the learning can be performed based on deep learning.

[0052] In one specific embodiment of the present invention, the relationship between gene expression and cell death of the cells can be based on in vitro data or computer data regarding whether cancer cell lines are killed due to a reduction or removal of the expression level of targeted genes.

[0053] In one specific embodiment of the present invention, the neoantigen binding affinity prediction model can be generated using interaction data between the amino acids of the training peptide and the amino acids of HLA.

[0054] In one specific embodiment of the present invention, the gene expression profile of the cancer patient can be sequencing data of the exome, transcriptome, single-cell transcriptome, peptidomome, or whole genome.

[0055] In one specific embodiment of the present invention, the HLA may be MHC class I or class II.

[0056] In a specific embodiment of the present invention, when the CNN-MHC level between the neoantigen and the HLA of the antigen-presenting cell is >0.5, binding affinity can be determined.

[0057] In one specific embodiment of the present invention, multiple cancer cell survival-dependent genes can be selected based on their necessity for cancer cell survival.

[0058] Another aspect of the present invention provides a method for preparing an anticancer vaccine, comprising:

[0059] Obtaining a new antigen through the screening method according to one aspect of the present invention; and

[0060] Prepare an anticancer vaccine comprising the neoantigen.

[0061] In a specific embodiment of the present invention, the method for preparing the anticancer vaccine may further include: obtaining a peptide sequence comprising 9 to 30 amino acids, including the novel antigen site; and

[0062] Select a peptide sequence that is hydrophilic and stable from the peptide sequence.

[0063] In one specific embodiment of the present invention, the selected peptide sequence has Kyte-Doolittle GRAVY < 0 and InstaIndex < 40.

[0064] In one specific embodiment of the present invention, the peptide sequence may consist of 12 to 30, 15 to 30, or 15 to 25 amino acids.

[0065] Another aspect of the present invention provides an anticancer vaccine comprising a neoantigen obtained by the screening method described above according to one aspect of the present invention.

[0066] Cancer vaccines can elicit specific cytotoxic T-cell responses and / or specific helper T-cell responses.

[0067] In one specific embodiment of the present invention, the anticancer vaccine may include a peptide containing a neoantigen.

[0068] In one specific embodiment of the present invention, the peptide has a length of 15 to 30 amino acids and can bind to HLA on antigen-presenting cells and activate neoantigen-specific T cells.

[0069] In one specific embodiment of the present invention, the anticancer vaccine may include peptides containing two or more neoantigens.

[0070] In one specific embodiment of the present invention, the anticancer vaccine may further include additional active ingredients such as anticancer agents, additives, excipients, etc.

[0071] The anticancer vaccine is administered in an amount sufficient to elicit an antigen-specific immune response. Those skilled in the art can determine the amount of peptides included in the anticancer vaccine or the dosage of the anticancer vaccine without extensive experimentation. Anticancer vaccines can be administered via intravenous, subcutaneous, intradermal, intraperitoneal, or intramuscular injection. The concentration of peptides in anticancer vaccines can vary widely, for example, less than about 0.1% by weight, about 2% to about 20%, about 50% or more, depending on the method of administration. The dosage of the anticancer vaccine is determined by the clinician by considering factors such as the composition of the peptides containing the neoantigen, the method of administration, the stage and severity of the target disease, the patient's weight, and overall health. Typically, for a patient weighing 70 kg, an amount of peptide ranging from about 1.0 μg to about 50,000 μg can be administered.

[0072] Another aspect of the present invention provides a method for providing information for predicting the treatment prognosis of cancer patients, comprising: obtaining a neoantigen by the screening method described above according to one aspect of the present invention; and

[0073] The amount of the neoantigen was measured from samples taken from cancer patients.

[0074] In one specific embodiment of the present invention, the method for providing information for predicting the treatment prognosis of cancer patients may further include: comparing the amount of the obtained neoantigen with the amount of the neoantigen obtained from a control group consisting of cancer patients with confirmed treatment prognosis.

[0075] In one specific embodiment of the present invention, the control group may consist of cancer patients confirmed to have a good treatment prognosis or cancer patients confirmed to have a poor treatment prognosis.

[0076] In one specific embodiment of the present invention, the amount of the neoantigen can be the quantity of the neoantigen.

[0077] In one specific embodiment of the present invention, when the amount or number of neoantigens in the sample of a cancer patient is higher than that in the control group of cancer patients with poor treatment prognosis, it can be predicted that the cancer patient will have a good treatment prognosis.

[0078] In one specific embodiment of the present invention, when the amount or number of neoantigens in the sample of a cancer patient is lower than that in the control group of cancer patients with good treatment prognosis, it can be predicted that the cancer patient has a poor treatment prognosis.

[0079] Another aspect of the invention provides a composition for predicting the treatment prognosis of cancer patients, comprising a neoantigen obtained by the screening method described above according to one aspect of the invention.

[0080] The composition for predicting the treatment prognosis of cancer patients according to one aspect of the invention may further include additional ingredients necessary for predicting the treatment prognosis of cancer patients.

[0081] Beneficial effects

[0082] According to a specific embodiment of the present invention, neoantigens derived from universally dependent genes essential for cancer cell survival or cancer patient-specific dependent genes can be screened as diagnostic and / or therapeutic targets. This can be effectively used for the development of neoantigen-based anticancer vaccines and the prediction of cancer patient treatment prognosis. The neoantigens can maximize the effect of immunotherapy on cancer cells without worrying about immune evasion by cancer cells. Attached Figure Description

[0083] Figure 1 This is a schematic diagram of a computer-based method for predicting cancer cell survival-dependent genes according to a specific embodiment of the present invention.

[0084] Figure 2 This is a flowchart of a neoantigen screening method according to a specific embodiment of the present invention. The determination of the cancer cell survival dependence of the neoantigen's source gene and the determination of the binding affinity between the neoantigen and antigen-presenting cells can be adjusted in terms of their execution order and number of times as needed.

[0085] Figure 3 This is a block diagram of a system for screening new antigens according to a specific embodiment of the present invention.

[0086] Figure 4 This is a block diagram of a processor according to a specific embodiment of the present invention.

[0087] Figure 5The significance of neoantigens obtained using a screening method for neoantigens with in vitro dependence data according to a specific embodiment of the present invention as diagnostic and / or therapeutic targets is illustrated. The significance of the screening method for neoantigens based on cancer cell survival dependence according to a specific embodiment of the present invention was verified using in vitro dependence data and expression homogeneity data for published cohorts of lung cancer and melanoma. The gray dashed lines represent the differences in the number of neoantigens calculated for all genes according to currently used standard methods.

[0088] Figure 6 This is a schematic diagram of the immune evasion mechanism of cancer cells. It can be seen that when the source protein of the neoantigen is essential for the survival of cancer cells and is a constitutive neoantigen uniformly expressed in all cells, cancer cells respond effectively to immunotherapy. However, when the neoantigen is not essential for the survival of cancer cells, or is a facultative neoantigen unevenly expressed in all cells (lower end), an immune evasion mechanism against immunotherapy occurs.

[0089] Figure 7 This paper illustrates the response patterns of neoantigens to immunotherapy obtained through a screening method based on in vitro dependence data and single-cell gene expression data, according to a specific embodiment of the present invention. Within the melanoma immune checkpoint inhibitor prescription group (Riaz), the results of pre- and post-treatment clonal and gene expression changes were compared in terms of the essentiality of neoantigen-derived gene function to cancer survival (high dependence vs. low dependence) and the homogeneity of gene expression (homogeneous vs. heterogeneous expression). CR / PR and SD / PD represent positive and negative prognoses, respectively. In patients with a positive prognosis, neoantigens derived from high-essence genes are primarily attacked by the immune system, leading to clonal contraction, and as a result of immune editing, RNA expression is reduced. In patients with a negative prognosis, it is confirmed that neoantigens derived from low-essence genes are primarily attacked by the immune system, leading to clonal expansion and reduced expression.

[0090] Figure 8 This demonstrates the significance of neoantigens obtained through a screening method using in vitro-dependent data according to a specific embodiment of the present invention as diagnostic / therapeutic targets. Similar analyses were performed on results from the MSKCC group (468 genes) for all types of cancer patient samples and on survival analysis for the lung cancer and melanoma immunotherapy cohorts. Figure 5 The analysis shown is as follows. Figure 8The upper end (MSK pan-cancer) represents the results of analyzing mutation-originating genes based on the entire MSKCC genome or the top 50% (high fitness) or bottom 50% (low fitness) genes essential for cancer cell survival. The middle (lung cancer) and lower end (melanoma) represent the results of publicly available immunotherapy cohorts for lung cancer and melanoma based on all genes or the top 500 (high fitness) and bottom 500 (low fitness) genes essential for cancer cell survival. For each genome, patients were divided into those with high mutation burden (high) and low neoantigen load (low), and post-treatment survival rates were compared to display HR (hazard ratio) and p-values, respectively. A lower HR and a lower p-value indicate a higher explanatory power for the prognosis of treatment targeting that genome. The results show that using a small number of genomes with high cancer cell survival dependence better explains treatment prognosis than using all genomes according to standard methods.

[0091] Figure 9 This paper demonstrates the significance of neoantigens obtained from cancer patient-specific computer-dependent data as diagnostic / therapeutic targets according to a specific embodiment of the present invention. The significance was demonstrated based on transcriptomic data of lung and breast cancer patients published by The Cancer Genome Atlas (TCGA) (https: / / portal.gdc.cancer.gov / ). Figure 1 The computer-generated dependency prediction is shown. Since these are not patients who have actually received immunotherapy for cancer, they are divided into samples with high immune cell penetration (high white blood cell ratio) and samples with low immune cell penetration (low white blood cell ratio) to hypothesize that samples with high immune cell penetration will show effects similar to immunotherapy. Figure 5 As shown, the explanatory power of the source gene of the neoantigen on patient survival was measured, and similar results were obtained only in samples with high immune cell penetration (R: Responder, NR: Nonresponder). The gray dashed line represents the difference in the number of neoantigens calculated for all genes according to the standard methods currently used. This confirms that patient-specific dependent data show a higher explanatory power difference than general dependent data.

[0092] Figure 10This is a schematic diagram of a model for predicting the binding affinity between a neoantigen and MHC used in a neoantigen screening method according to a specific embodiment of the present invention. A two-dimensional matrix between HLA and the neoantigen peptide is constructed based on amino acid similarity matrix information, and a convolutional neural network (CNN) is applied here to display the model predicting the binding affinity between HLA and the antigen peptide.

[0093] Figure 11 The results show a performance (AUC, F1 score) comparison of CNN with other methods (NetMHCpan, NetMHCcon, ANN, SMMPMBEC) according to a specific embodiment of the present invention using a test dataset provided by the official Immune Epitope Database (IEDB).

[0094] Figure 12 The results confirming the predicted HLA binding affinity of neoantigens according to a specific embodiment of the present invention are shown. It has been confirmed that 80% of peptides containing neoantigens predicted to bind to HLA-AO2 actually bind to HLA-AO2.

[0095] Figure 13 The results confirming the immune response induced by a neoantigen selected according to a specific embodiment of the present invention are shown. Based on enzyme-linked immunospot (ELISpot) analysis based on INFγ secretion to confirm the CD8+ T cell response to the neoantigen, 10 (66.7%) of the 15 predicted reactive candidate neoantigen peptides elicited INFγ secretion from T cells. Detailed Implementation

[0096] The invention will become apparent from the accompanying drawings and the embodiments described in detail below. However, the invention is not limited to these embodiments and can be implemented in various forms. These embodiments are merely intended to complete the disclosure of the invention, and are intended to fully inform those skilled in the art of the scope of the invention. The invention is defined by the claims.

[0097] As used in this article, the term "exome" refers to the combination of exons present in a cell, cell population, or individual.

[0098] As used in this article, the term “transcriptome” refers to the combination of expressed ribonucleic acid (RNA) present in a cell, cell population, or individual.

[0099] As used herein, the term “gene expression profile” refers to a combination of values ​​representing the expression levels of genes, including one or more genes, based on the analysis of gene expression or transcription from the genomic genome of a cell.

[0100] As used in this article, the term "dependency" refers to the necessity for cell proliferation or survival and is used interchangeably with "necessity".

[0101] As used herein, the term "dependent gene" refers to a gene essential for cell proliferation or survival. More specifically, a dependent gene, as a gene whose reduced expression or removal leads to reduced cell proliferation and / or death, refers to a gene on which a cell depends for survival. This can include universally dependent genes and / or cancer patient-specific dependent genes. Universally dependent genes are generally essential for the survival of cancers or cancer cells of various types and / or origins, while cancer patient-specific dependent genes are essential for the survival of cancer cells derived from an individual cancer patient. A dependent gene can also refer to a gene that is constitutively and essentially expressed in cells and uniformly expressed in all individual cells.

[0102] As used in this article, the term "universal dependent gene" refers to a gene that has been identified as essential for the survival of most different cancer cells by in vitro data such as known cancer cell lines.

[0103] As used in this article, the term "cancer patient-specific dependent gene" refers to a gene identified as essential for the survival of cancer cells derived from an individual cancer patient.

[0104] As used herein, the term "neoantigen" refers to a peptide that elicits an immune response. That is, a neoantigen can be an immunogenic peptide. Neoantigens can be induced by cancer cell-specific mutations and can manifest as epitopes of cancer cells. A neoantigen is an antigen that has at least one modification that distinguishes it from its corresponding wild-type or parental antigen, through mutations in cancer cells or cancer cell-specific post-translational modifications. Neoantigens can include amino acid sequences or nucleotide sequences. Mutations can include frameshift or non-frameshift mutations, indels, missense or nonsense, splice site alterations, genomic rearrangements or gene fusions, or any genomic or expression alteration that induces a new ORF.

[0105] Since neoantigens derived from genes with universal or cancer patient-specific dependence are not lost due to immune evasion by cancer cell immune editing, they can serve as effective therapeutic targets for cancer patient-customized cancer vaccines that can bring high immunotherapy efficacy to cancer patients, and can also serve as effective diagnostic targets for immunotherapy prognosis.

[0106] As used herein, the term "immunotherapy" refers to a therapy that utilizes an immune response. Immunotherapy can be used to treat cancer. For example, immunotherapy can be a treatment using immune checkpoint inhibitors, which can be, but are not limited to, anti-CTLA4 blockers or anti-PD-1 / PD-L1 blockers, and can be a variety of immunotherapies.

[0107] As used herein, the term "binding affinity" refers to the binding strength between a neoantigen peptide and the MHC of an antigen-presenting cell, which can be represented as a CNN-MHC value. The "CNN-MHC value" is a value obtained by building a deep learning model based on experimental values ​​obtained by converting the binding strength between each amino acid of the neoantigen and the MHC into a matrix form. The CNN-MHC value refers to a probability value between 0 and 1 converted to a sigmoid activation function. Specifically, immunogenic peptides that can bind to MHC class I or II proteins can have an MHC CNN-MHC value of 0.5 or higher. Furthermore, the closer the CNN-MHC value is to 1, the stronger the binding affinity between the MHC class I or II protein and the immunogenic peptide.

[0108] As used in this article, the term "antigen-presenting cell" refers to a cell that receives and processes protein antigens, and then presents antigen-derived peptide fragments along with MHC class II cells to T cells for activation. Examples of such cells include macrophages, B cells, and dendritic cells.

[0109] As used herein, the term "cell survival-dependent prediction model" refers to a model that predicts the probability of cell survival or death based on the reduction or removal of an individual's gene expression. Specifically, it is a model that uses machine learning to predict the effect of a specific gene on cell survival based on the learning of the impact of gene knockout / knockdown on cell survival. Machine learning can be performed using in vitro gene knockout / knockdown data such as RNAi or CRISPR / Cas9. By inputting a gene expression profile or sequence into the cell survival-dependent prediction model, the cell survival dependence of the gene from which that sequence originates can be predicted.

[0110] As used herein, the term "neoantigen binding affinity prediction model" refers to a model that predicts the binding affinity between a neoantigen and antigen-presenting cells, particularly the MHC of antigen-presenting cells. When a neoantigen's peptide sequence is input and machine learning is used to learn binding affinity based on amino acid interactions between peptide and HLA sequences, its binding affinity with the HLA of antigen-presenting cells can be predicted, and the neoantigen can be classified according to a pre-defined binding affinity scale.

[0111] Example 1: Selection of cancer cell survival-dependent genes

[0112] To determine whether gene function is essential for cancer or cancer cell survival, high-throughput screening (HTS) can be performed using RNAi or CRISPR libraries capable of knocking out / down all genes. Specifically, cancer cells are transfected with shRNA libraries or CRISPR sgRNA, and deep sequencing is performed after a certain time period to compare the results with the initial sequencing. This allows for the selection of genes essential for cancer cell survival in vitro by quantitatively analyzing which genes are inactivated and which kill the cells. Figure 1 This schematically illustrates a method for predicting genes essential for cancer cell survival.

[0113] Through these methods, in vitro data on genes essential for cancer cell survival are continuously generated, while the number of cell lines increases. For major solid cancers, such as lung cancer, ovarian cancer, colorectal cancer, gastric cancer, and breast cancer, dependency data on a large number of cancer cell lines have been established (https: / / depmap.org / portal / or https: / / depmap.sanger.ac.uk / ). This data is used as data on general-dependent genes according to a specific embodiment of the present invention or for the purpose of predicting cancer patient-specific dependent genes on a computer based on deep learning.

[0114] In vitro data from cancer cell lines can be used to obtain universally dependent genes, while data from cancer patients, such as transcriptome data from cancer patients, are applied to cell survival dependency prediction models learned from in vitro data from cancer cell lines to predict the dependency of each patient sample and to select cancer patient-specific dependent genes.

[0115] Furthermore, by applying cell survival dependency prediction models to single-cell transcriptome data, dependency patterns among different cells can be identified through tumor heterogeneity, and genes exhibiting the same dependency across multiple cells can be selected as universally dependent genes. Genes essential for cancer cell survival and exhibiting uniform expression at the single-cell level are selected as dependent genes that can serve as effective diagnostic and / or therapeutic targets.

[0116] The cell survival-dependent prediction model is a deep learning neural network model consisting of an input layer, multiple hidden layers, and an output layer. The neural network is configured to learn the relationship between cell gene expression patterns and cell death in the multiple hidden layers when the aforementioned in vitro data is input to the input layer, and output predetermined probability values ​​in the output layer. These predetermined probability values ​​include values ​​representing the probability of cell death and values ​​representing the probability of cell growth.

[0117] When single-cell transcriptome data from cancer patients is available, it can be used to select cancer patient-specific dependent genes using cell survival-dependent models; otherwise, data from other publicly available cancer patient samples can be used to select universally dependent genes.

[0118] Example 2: Selection of neoantigens derived from universally dependent genes and their significance as diagnostic / therapeutic targets

[0119] Neoantigens are peptides that bind to MHC proteins, originate from the surface of cancer cells, and are recognized as antigens by immune cells. They are protein fragments not present in normal cells but produced by cancer-specific mutations. Neoantigens are a key element of immunotherapy for cancer, and it is known that the greater the number of neoantigens, the better the responsiveness to immunotherapy. Neoantigen load is also used as a diagnostic biomarker. However, because neoantigens originate from various genes and have different characteristics, their uses as diagnostic biomarkers or therapeutic targets will vary. In this embodiment, as described in Example 1, neoantigens derived from cancer cell survival-dependent genes were selected as useful neoantigens to maximize the efficacy of immunotherapy for cancer, and their immunotherapy responsiveness was confirmed.

[0120] Specifically, immunotherapy responsiveness was compared using the prescription groups of immune checkpoint blockades listed in Table 1.

[0121] [Table 1]

[0122]

[0123] 2-1. Pre-therapy group

[0124] The usefulness of a screening method for neoantigens based on cancer cell survival dependence according to a specific embodiment of the present invention was verified using in vitro dependence and expression homogeneity data for publicly available cohorts of lung cancer and melanoma with only pre-treatment outcomes. The analysis results are as follows: Figure 5 As shown. Specifically, based on the necessity of the source genes of the neoantigens for cancer cell survival, 500 to 2000 neoantigens were selected from the top (low dependence) and the bottom (high dependence) regions, and the formula was used. The difference in neoantigen load for each gene was calculated between patients with good and poor prognoses after immunotherapy for cancer. A larger difference indicates better explanatory power for treatment prognosis. The gray dashed line represents the difference in neoantigen load calculated for all genes using existing standard methods. Results show that using a few highly essential genes or uniformly expressed genes provides better explanatory power than using all genes, while using low-essential genes or heterogeneously expressed genes provides poorer explanatory power. Similarly, the difference in neoantigen load was calculated based on expression homogeneity and gene expression level. This suggests that gene essentiality and expression homogeneity are better describers of treatment prognosis than gene expression level itself. Specifically, after ranking the source genes of neoantigens according to their essentiality for cancer cell survival, 500 to 2000 genes were selected (low dependence) and 2000 genes (high dependence), and the difference in neoantigen load for each gene between patients with good and poor prognoses after immunotherapy for cancer was calculated. It is evident that the greater the difference, the better the explanatory power for treatment prognosis. The gray dashed line represents the difference in the number of neoantigens calculated for all genes according to currently used standard methods. The results show that using a few highly essential genes or uniformly expressed genes demonstrates better explanatory power than using all genes, while using low-essential genes or heterogeneously expressed genes provides poorer explanatory power. Similarly, the difference in the number of neoantigens calculated based on expression homogeneity and gene expression level indicates that gene essentiality and expression homogeneity are better predictors of treatment prognosis than gene expression values ​​themselves.

[0125] These results suggest that immune evasion mechanisms can occur more actively when the function of the genes from which the neoantigens from which the immune response is concentrated is not essential for cancer proliferation or survival. Figure 6 This is a schematic diagram of the immune evasion mechanism of cancer cells.

[0126] 2-2. Pre-therapy and on-therapy groups

[0127] Data from the Riaz cohort, including pre-therapy and on-therapy data with immune checkpoint inhibitors, were used to analyze the dependence on neoantigen-derived genes and immunotherapy responsiveness. The results are as follows: Figure 7 As shown.

[0128] Specifically, within the melanoma immune checkpoint inhibitor prescription group (Riaz), the results of clonal and gene expression changes before and after treatment were compared in terms of the necessity of neoantigen-derived gene function for cancer survival (high dependence vs. low dependence) and the homogeneity of gene expression (homogeneous vs. heterogeneous expression). CR / PR and SD / PD represent positive and negative prognoses, respectively.

[0129] It has been confirmed that in patients with good treatment response (CR (complete remission) / PR (partial remission)), neoantigens derived from homogeneously expressed genes and genes essential for cancer cell survival become the main targets of the anti-cancer immune response, resulting in clonal contraction and a decrease in RNA expression levels. In contrast, in patients with poor treatment response (SD (stable disease) and PD (progressive disease)), neoantigens derived from genes exhibiting heterogeneous expression and genes not essential for cancer cell survival become the main targets of immune attack. Immune editing leads to a reduction in gene expression, and due to successful immune evasion, clonal expansion occurs instead.

[0130] 2-3. Integrated Mutation Profiling of Actionable Cancer Targets (MSK-IMPACT)

[0131] MSK-IMPACT was used to verify the dependence of neoantigen-derived genes on various cancers and their responsiveness to immunotherapy.

[0132] In existing studies, mutation levels (proportional to neoantigen levels) were measured in only 468 genes belonging to the so-called MSKCC (Memorial Sloan Kettering Cancer Center) group in 1,662 individuals with various cancer types who received immune checkpoint inhibitor therapy, and the results showed that this level is an important determinant of immunotherapy responsiveness (Nat. Genet. 51:202-206, 2019).

[0133] This embodiment has demonstrated that when measuring mutation levels by selecting the top 50% of genes (highly dependent) with high cancer cell survival essentiality from the 468 genes present in the MSKCC group, the correlation with responsiveness to immune checkpoint inhibition therapy is higher compared to using all 468 genes. The results are as follows... Figure 8 The top of the chart is shown. This is the result of survival analysis based on the entire MSKCC genome or on either the top 50% (high fitness) or bottom 50% (low fitness) genes essential for cancer cell survival. The hazard ratio (HR) and p-value are shown by comparing post-treatment survival rates. A lower HR and a lower p-value indicate a higher explanatory power for the prognosis of treatment targeting that genome. Compared to the mutation load of all 486 genes, the mutation load of the 226 genes belonging to the top 50% that are highly dependent on cancer cell survival showed a stronger correlation with immunotherapy responsiveness.

[0134] in addition, Figure 8 The middle and lower sections show the results of publicly available immunotherapy cohorts for lung cancer and melanoma, analyzed based on all genes or the top 500 and bottom 500 genes essential for cancer cell survival, respectively. Patients were divided into those with high mutation burden or low neoantigen load for each genome, and post-treatment survival was compared to show HR and p-values, respectively. The results indicate that using a small number of genomes with high cancer cell survival dependence better interprets treatment prognosis than using all genomes according to standard methods.

[0135] These results confirm that the essentiality of the source genes of antigens (including neoantigens and surface antigens) in cancer cells and the uniformity of expression patterns among individual cells in tissues are crucial for the immune response to cancer. Therefore, in therapeutics, inducing neoantigens concentrated on genes essential for cancer survival and those exhibiting uniform expression is essential to minimizing immune evasion by cancer, and the amount of these neoantigens can serve as a prognostic marker for immunotherapy.

[0136] Example 3: Selection of neoantigens derived from cancer patient-specific dependent genes and their significance as diagnostic / therapeutic targets

[0137] In this embodiment, data obtained from samples of specific cancer patients were applied to a cell survival dependency prediction model to obtain neoantigens derived from selected cancer patient-specific dependent genes, and the correlation between these neoantigens and cancer patient survival was confirmed.

[0138] In Example 2, the association between neoantigens derived from universally dependent genes and the survival rate of immunotherapy patients was examined. The dependency data used in Example 2 was derived from the in vitro cancer cell line experiments described in Example 1. This was not patient-specific dependency, but rather identified genes with universal dependency across several cancer cell lines. As described above, a cell survival dependency prediction model was generated by training the relationship between gene expression level patterns and cell death from the in vitro dependency data. By inputting the gene expression profiles of cancer patients into the prediction model, cancer patient-specific dependent genes could be selected. To confirm the significance of neoantigens obtained from cancer patient-specific computational dependency data as diagnostic / therapeutic targets, transcriptome data from lung cancer and breast cancer patients published by The Cancer Genome Atlas (TCGA) (https: / / portal.gdc.cancer.gov / ) were used. Since these were not patients who had actually received immunotherapy, they were divided into samples with high immune cell penetration (high white blood cell ratio) and samples with low immune cell penetration (low white blood cell ratio) to hypothesize that immunotherapy-like effects would be seen in samples with high immune cell penetration.

[0139] The result is as follows Figure 9 As shown. Compared with the total number of neoantigens ( Figure 9 The number of neoantigens derived from cancer patient-specific dependent genes essential for cancer cell proliferation or survival (high dependence) is compared to the number of neoantigens derived from non-dependent genes (low dependence) that do not contribute to survival. Figure 9 The black bars in the graph (in the image) have higher explanatory power. In particular, these results were observed only in samples with high immune cell penetration, and it can be confirmed that the differences in explanatory power of cancer patient-specific dependent data are generally greater than those of general dependent data.

[0140] Example 4: Predictive Model for the Binding Force of Neoantigens to Antigen-Presenting Cells

[0141] In order for a neoantigen to exhibit immunotherapy responsiveness, the neoantigen should be processed by antigen-presenting cells and bind to HLA on the cell surface.

[0142] To construct and validate a predictive model for the binding affinity of neoantigens to antigen-presenting cells, benchmark data from the Immune Epitope Database (IEDB), which is used in existing studies, were used, and the predictive power results were compared with those of existing machine learning algorithms publicly available in IEDB.

[0143] The neoantigen and antigen-presenting cell binding affinity prediction model, as a CNN model, includes multiple convolutional layers, pre-connected layers, and output layers, but excludes pooling layers, so that all output values ​​of the convolutional layers can be used for prediction.

[0144] Multiple convolutional layers extract interaction features from the input data, which refers to map data including parameters representing the binding affinity between amino acids of peptides and HLA amino acids. Multiple convolutional layers perform convolutions using a specific number of kernels or weight matrices. Fully connected layers receive and integrate the output values ​​of the convolutional layers as input, and the output layer uses a sigmoid function to output information about binding affinity. Figure 10 A schematic diagram of a model (CNN-MHC) for predicting the binding affinity of a neoantigen to MHC used in a neoantigen screening method according to a specific embodiment of the present invention.

[0145] The model was trained using in vitro peptide-MHC binding assays from 50,000 or more peptides in IEDB. Performance evaluation using weekly newly released test datasets from IEDB confirmed that for 70% or more of the test datasets, it not only outperformed existing algorithms such as SMMPMBEC, ANN, and NetMHCcon, but also significantly outperformed the most widely used NetMHCpan. The results are as follows: Figure 11 As shown.

[0146] 4-1 Confirmation of binding between the neoantigen and HLA

[0147] To confirm the actual binding between the predicted neoantigen and the HLA molecule using the model (CNN-MHC) of this embodiment, the binding ability between the neoantigen and HLA was tested. Specifically, the binding ability of peptides predicted to bind to HLA-A02 was analyzed using the T2 cell line (ATCC CRL-1992) expressing HLA-A02. The peptides expected to bind to HLA-A02 are shown in Table 2 (SEQ ID NO: 1 to SEQ ID NO: 50). In Table 2, CNN-MHC values ​​are values ​​obtained by constructing a deep learning model based on experimental values, which are converted into probability values ​​between 0 and 1. The experimental values ​​are obtained by converting the binding strength between each amino acid of the neoantigen and MHC into matrix form. NetMHC values ​​refer to the predicted binding values ​​between MHC proteins and peptides calculated using NetMHCPan-4.1, a model trained on mass spectrometry data using the latest version of NetMHC, a tool commonly used for neoantigen binding prediction. Higher values ​​for both indicate a higher probability of binding. Specifically, the cultured T2 cell line (1×10⁻⁶) 6The sample was treated with a peptide at a concentration of 50 μg / ml for 24 hours, followed by staining with an APC-labeled HLA-A2 monoclonal antibody. Flow cytometry was used to confirm binding capacity. DMSO was used as a negative control, and Mart-1 and NY-ESO were used as positive controls.

[0148] [Table 2]

[0149]

[0150] Figure 12 The results of the MHC binding analysis are shown. It was confirmed that 80% of the peptides predicted by the model (CNN-MHC) of this embodiment actually bind to HLA-A02. The model of this embodiment has better predictive power compared to the predictions of the widely used NetMHCpan-4.1 (http: / / www.cbs.dtu.dk / services / NetMHCpan-4.1 / ).

[0151] On the other hand, the aforementioned embodiments can be executed by the processor 120, stored in, such as Figure 3 At least one instruction from memory 110 of the system shown is used for execution. See also Figure 4 The processor 120 may include: a data acquisition unit 121 that acquires sequencing data of the exome, transcriptome, or whole genome; a prediction model generation unit 122 that generates a cell survival-dependent prediction model and a binding affinity prediction model; a gene selection unit 123 that uses the cell survival-dependent prediction model to select cancer cell survival-dependent genes; and a neoantigen selection unit 124 that collects neoantigens from the gene expression profile of cancer cells and uses the binding affinity prediction model to select neoantigens that have binding affinity with HLA of antigen-presenting cells.

[0152] As the prediction model generation unit 122 repeatedly executes instructions, the cell survival dependence prediction model and the binding force prediction model can be updated based on new input data. The prediction models generated and updated by the prediction model generation unit can be stored in memory.

[0153] Example 5: Anticancer vaccine containing neoantigen

[0154] As described in the foregoing embodiments, anticancer vaccines comprising neoantigens derived from cancer cell survival-dependent genes can be prepared.

[0155] Exome sequencing of mouse cell lines transplanted with cancer cells or tissues can identify mutations not present in normal cells. These mutations are then applied to a binding affinity prediction model for antigen-presenting cells to select neoantigen candidates. Using a cell survival-dependent model, the necessity of cancer cell survival against genes derived from the neoantigen is determined based on dependence data for that specific cancer. Neoantigens are then ranked according to necessity, with the top 5 and bottom 5 selected. All possible peptide sequences of 9 to 30 amino acids, including the selected neoantigen site, are generated. Sequences with suitable chemical properties for peptide synthesis, such as high hydrophilicity (Kyte-Doolittle GRAVY < 0) and a low instability index (InstaIndex < 40), are selected and synthesized to prepare an anticancer vaccine against the neoantigen.

[0156] 5-1. Selecting a new antigen

[0157] Using the cell survival dependence prediction model and the neoantigen-antigen binding affinity prediction model as described in Examples 1 to 4, mutations originating from genes with high cell survival dependence and HLA binding affinity were selected by analyzing the exome and transcriptome information of cancer cell lines.

[0158] 5-2. Immune response confirmed by neoantigen

[0159] To confirm whether the immune response was induced by neoantigens, the generation of T cells that recognized neoantigen peptides with predicted or experimentally validated HLA-binding ability, including neoantigen peptides selected from 5-1, was tested. Specifically, to confirm a T cell pool exhibiting specificity, mice transplanted with mouse cancer cell lines were used. The mouse model was established by using 1 × 10⁻⁶ mouse lung cancer cell lines LLC-1 (ATCC CRL-1642). 6 It was generated by subcutaneous injection into the flank of 6-week-old male C57BL / 6 mice weighing 20g.

[0160] Specifically, by analyzing the exome and transcriptome of cancer cell lines, 15 mutants predicted to be highly essential for cell survival and exhibiting high binding affinity to antigen-presenting cells were selected as neoantigens predicting HLA binding and CD8+ T cell responsiveness. Information on the selected peptides is summarized in Table 3 (SEQ ID NO: 51 to 65). The CNN-MHC values ​​listed in Table 3 are obtained by building a deep learning model based on experimental values, which are obtained by converting the binding strength between each amino acid of the neoantigen and the MHC into a matrix form. The CNN-MHC values ​​represent probability values ​​between 0 and 1, with higher values ​​indicating a higher probability of binding. To confirm the CD8+ T cell response to the neoantigen using spleen cells extracted from mouse spleens treated with a 9-mer or 15-mer peptide synthesized from this mutant sequence, an ELISpot analysis based on IFNγ secretion was performed. Ten peptides (66.7%) of the 15 predicted responsive candidate neoantigen peptides were confirmed to be IFNγ secreted by T cells. The results are as follows: Figure 13 As shown. This confirms that the neoantigen selected using the method according to a specific embodiment of the present invention can indeed bind to HLA and effectively induce an immune response.

[0161] [Table 3]

[0162]

[0163] Example 6: Predicting the treatment prognosis of cancer patients using neoantigens

[0164] As described in the foregoing embodiments, neoantigens derived from cancer cell survival-dependent genes can be used to predict the treatment prognosis of cancer patients.

[0165] Exome sequencing of patient samples can identify mutations present only in cancer cells but not in normal cells. These mutations are then applied to a model predicting binding affinity to antigen-presenting cells to select neoantigen candidates. Using a cell survival dependency model, the necessity of cancer cells for survival against genes derived from neoantigens is determined based on dependency data for that cancer. Neoantigens are then ranked according to survival necessity, allowing for the selection of an equal number of genes (e.g., 500) in the order of importance. Patients who have received immunotherapy are divided into responsive and non-responsive groups. By comparing the number of neoantigens derived from the selected essential genes in the order of importance, it can be determined whether the number of neoantigens derived from highly essential genes in the order of importance is highly correlated with patient treatment prognosis. By repeatedly determining the correlation between the number of neoantigens and patient treatment prognosis while varying the number of genes to be selected, neoantigens derived from genes dependent on genes essential for predicting patient treatment prognosis can be identified.

[0166] Figure 8 The results show that the number of mutations in genes highly essential for cancer cell survival or the number of neoantigens have a high explanatory power for treatment prognosis and can be used to predict the treatment prognosis of cancer patients.

[0167] The above description of the present invention is for illustrative purposes only, and those skilled in the art will understand that it can be readily modified into other specific forms without altering the technical spirit or essential characteristics of the invention. Therefore, it should be understood that the specific embodiments described above are exemplary in all respects and not restrictive. For example, each component described as a single type can also be implemented in a distributed manner, and similarly, components described as distributed can also be implemented in a combined manner. <110> Korea Advanced Institute of Science and Technology (KAIST) Petmedix Ltd. <120> A screening method, system and application of a new antigen <130> PN190331 <160> 65 <170> KoPatentIn 3.0 <210> 1 <211> 9 <212> PRT <213> Artificial Sequence <220> <223> Neoantigen HLA-A02 binding <400> 1 Arg Leu Trp His Leu Leu Leu Gln Val 1 5 <210> 2 <211> 9 <212> PRT <213> Artificial Sequence <220> <223> Neoantigen HLA-A02 binding <400> 2 Leu Met Phe Ser Arg Cys Thr Ser Val 1 5 <210> 3 <211> 9 <212> PRT <213> Artificial Sequence <220> <223> Neoantigen HLA-A02 binding <400> 3 Arg Leu Leu Asp Glu Tyr Asn Leu Val 1 5 <210> 4 <211> 9 <212> PRT <213> Artificial Sequence <220> <223> Neoantigen HLA-A02 binding <400> 4 Leu Leu Ala Leu Trp Leu Cys Trp Ala 1 5 <210> 5 <211> 9 <212> PRT <213> Artificial Sequence <220> <223> Neoantigen HLA-A02 binding <400> 5 His Leu Tyr Pro Gly Pro Cys His Leu 1 5 <210> 6 <211> 9 <212> PRT <213> Artificial Sequence <220> <223> Neoantigen HLA-A02 binding <400> 6 Arg Leu Met Thr His Tyr Cys Ala Met 1 5 <210> 7 <211> 9 <212> PRT <213> Artificial Sequence <220> <223> Neoantigen HLA-A02 binding <400> 7 Gly Leu Leu Pro Leu Ala Ser Thr Val 1 5 <210> 8 <211> 9 <212> PRT <213> Artificial Sequence <220> <223> Neoantigen HLA-A02 binding <400> 8 Phe Leu Leu Gly Phe Leu His Ser Gly 1 5 <210> 9 <211> 9 <212> PRT <213> Artificial Sequence <220> <223> Neoantigen HLA-A02 binding <400> 9 Leu Met Asn Ser Leu Val Ser Gln Val 1 5 <210> 10 <211> 9 <212> PRT <213> Artificial Sequence <220> <223> Neoantigen HLA-A02 binding <400> 10 Cys Leu Tyr Ala Phe Glu Cys Lys Ile 1 5 <210> 11 <211> 9 <212> PRT <213> Artificial Sequence <220> <223> Neoantigen HLA-A02 binding <400> 11 Lys Thr Phe Pro Val Gln Leu Trp Val 1 5 <210> 12 <211> 9 <212> PRT <213> Artificial Sequence <220> <223> Neoantigen HLA-A02 binding <400> 12 Gly Leu Leu Ala Glu Ser Thr Trp Ala 1 5 <210> 13 <211> 9 <212> PRT <213> Artificial Sequence <220> <223> Neoantigen HLA-A02 binding <400> 13 Ser Met Gln Asn His Ile Pro Gln Val 1 5 <210> 14 <211> 9 <212> PRT <213> Artificial Sequence <220> <223> Neoantigen HLA-A02 binding <400> 14 Cys Gln Leu Ala Lys Thr Cys Pro Val 1 5 <210> 15 <211> 9 <212> PRT <213> Artificial Sequence <220> <223> Neoantigen HLA-A02 binding <400> 15 Leu Leu Gly Pro Asn Ser Phe Glu Val 1 5 <210> 16 <211> 9 <212> PRT <213> Artificial Sequence <220> <223> Neoantigen HLA-A02 binding <400> 16 Ser Leu Ser Tyr Trp Arg Ala Ser Val 1 5 <210> 17 <211> 9 <212> PRT <213> Artificial Sequence <220> <223> Neoantigen HLA-A02 binding <400> 17 Arg Val Ala Gln Tyr Pro Phe Glu Val 1 5 <210> 18 <211> 9 <212> PRT <213> Artificial Sequence <220> <223> Neoantigen HLA-A02 binding <400> 18 Leu Leu Ala Glu Thr Cys Trp Asn Gly 1 5 <210> 19 <211> 9 <212> PRT <213> Artificial Sequence <220> <223> Neoantigen HLA-A02 binding <400> 19 Val Leu Gln Pro Ala Pro His Gln Val 1 5 <210> 20 <211> 9 <212> PRT <213> Artificial Sequence <220> <223> Neoantigen HLA-A02 binding <400> 20 Tyr Leu His Cys Glu Trp Ala Thr Ile 1 5 <210> twenty one <211> 9 <212> PRT <213> Artificial Sequence <220> <223> Neoantigen HLA-A02 binding <400> twenty one Leu Leu Ala Glu Ser Thr Trp Ala Leu 1 5 <210> twenty two <211> 9 <212> PRT <213> Artificial Sequence <220> <223> Neoantigen HLA-A02 binding <400> twenty two Pro Leu Glu Glu Trp Asn Gln Trp Val 1 5 <210> twenty three <211> 9 <212> PRT <213> Artificial Sequence <220> <223> Neoantigen HLA-A02 binding <400> twenty three Ala Leu Asn Glu Met Phe Cys Gln Leu 1 5 <210> twenty four <211> 9 <212> PRT <213> Artificial Sequence <220> <223> Neoantigen HLA-A02 binding <400> twenty four Leu Leu Arg Arg Asn Ser Phe Glu Val 1 5 <210> 25 <211> 9 <212> PRT <213> Artificial Sequence <220> <223> Neoantigen HLA-A02 binding <400> 25 Ala Leu Pro Gln His Leu Ile Arg Val 1 5 <210> 26 <211> 9 <212> PRT <213> Artificial Sequence <220> <223> Neoantigen HLA-A02 binding <400> 26 Met Ala Gly Gly Trp Val Ala His Leu 1 5 <210> 27 <211> 9 <212> PRT <213> Artificial Sequence <220> <223> Neoantigen HLA-A02 binding <400> 27 Gly Leu Asn Ser Phe Glu Val Arg Val 1 5 <210> 28 <211> 9 <212> PRT <213> Artificial Sequence <220> <223> Neoantigen HLA-A02 binding <400> 28 Cys Gln Leu Ala Lys Thr Phe Pro Val 1 5 <210> 29 <211> 9 <212> PRT <213> Artificial Sequence <220> <223> Neoantigen HLA-A02 binding <400> 29 Asn Met Ala Asn Met Pro Pro Gln Val 1 5 <210> 30 <211> 9 <212> PRT <213> Artificial Sequence <220> <223> Neoantigen HLA-A02 binding <400> 30 Tyr Thr Glu Ala Glu Glu Phe Phe Val 1 5 <210> 31 <211> 9 <212> PRT <213> Artificial Sequence <220> <223> Neoantigen HLA-A02 binding <400> 31 Thr Leu Met Ser Val Pro Arg Tyr Leu 1 5 <210> 32 <211> 9 <212> PRT <213> Artificial Sequence <220> <223> Neoantigen HLA-A02 binding <400> 32 Lys Met Trp Val Asp Arg Tyr Leu Ala 1 5 <210> 33 <211> 9 <212> PRT <213> Artificial Sequence <220> <223> Neoantigen HLA-A02 binding <400> 33 Gly Leu Ala Pro Pro Gln Tyr Leu Ile 1 5 <210> 34 <211> 9 <212> PRT <213> Artificial Sequence <220> <223> Neoantigen HLA-A02 binding <400> 34 Phe Leu Ile Asn Ser Ser Tyr Pro Gly 1 5 <210> 35 <211> 9 <212> PRT <213> Artificial Sequence <220> <223> Neoantigen HLA-A02 binding <400> 35 Phe Met Tyr Phe Glu Phe Pro Gln Leu 1 5 <210> 36 <211> 9 <212> PRT <213> Artificial Sequence <220> <223> Neoantigen HLA-A02 binding <400> 36 Ala Ala Phe Val Leu Leu Phe Phe Val 1 5 <210> 37 <211> 9 <212> PRT <213> Artificial Sequence <220> <223> Neoantigen HLA-A02 binding <400> 37 Ile Leu His Val Ala Val Thr Asn Val 1 5 <210> 38 <211> 9 <212> PRT <213> Artificial Sequence <220> <223> Neoantigen HLA-A02 binding <400> 38 Arg Leu Arg Ser Arg Thr Cys Leu Leu 1 5 <210> 39 <211> 9 <212> PRT <213> Artificial Sequence <220> <223> Neoantigen HLA-A02 binding <400> 39 Ala Leu Asp Thr Ile Asn Ile Leu Leu 1 5 <210> 40 <211> 9 <212> PRT <213> Artificial Sequence <220> <223> Neoantigen HLA-A02 binding <400> 40 Leu Ile Tyr Gln Met Ala Pro Ala Val 1 5 <210> 41 <211> 9 <212> PRT <213> Artificial Sequence <220> <223> Neoantigen HLA-A02 binding <400> 41 Ser Gln Leu Pro Gly Leu Leu Glu Leu 1 5 <210> 42 <211> 9 <212> PRT <213> Artificial Sequence <220> <223> Neoantigen HLA-A02 binding <400> 42 Cys Gln Leu Ala Lys Thr Tyr Pro Val 1 5 <210> 43 <211> 9 <212> PRT <213> Artificial Sequence <220> <223> Neoantigen HLA-A02 binding <400> 43 Gly Met Gly Pro Pro Met Pro Thr Val 1 5 <210> 44 <211> 9 <212> PRT <213> Artificial Sequence <220> <223> Neoantigen HLA-A02 binding <400> 44 Phe Thr Ile Asn Arg Asn Thr Gly Val 1 5 <210> 45 <211> 9 <212> PRT <213> Artificial Sequence <220> <223> Neoantigen HLA-A02 binding <400> 45 Leu Met Asp Gly Arg Asp Val Phe Leu 1 5 <210> 46 <211> 9 <212> PRT <213> Artificial Sequence <220> <223> Neoantigen HLA-A02 binding <400> 46 Pro Leu Ile Glu Glu Ala Cys Glu Leu 1 5 <210> 47 <211> 9 <212> PRT <213> Artificial Sequence <220> <223> Neoantigen HLA-A02 binding <400> 47 Thr Met Tyr Gln Gln Gln Gln Gln Val 1 5 <210> 48 <211> 9 <212> PRT <213> Artificial Sequence <220> <223> Neoantigen HLA-A02 binding <400> 48 Ser Ala Leu His Trp Ala Ala Ala Val 1 5 <210> 49 <211> 9 <212> PRT <213> Artificial Sequence <220> <223> Neoantigen HLA-A02 binding <400> 49 Arg Met His Asp Gly Thr Thr Pro Leu 1 5 <210> 50 <211> 9 <212> PRT <213> Artificial Sequence <220> <223> Neoantigen HLA-A02 binding <400> 50 Leu Leu Met Glu Asn Ser Ile Gly Gly 1 5 <210> 51 <211> 15 <212> PRT <213> Artificial Sequence <220> <223> Twistnb <400> 51 Thr Val Val Glu Glu Val Val Glu Lys Thr Pro Lys Lys Lys Lys 1 5 10 15 <210> 52 <211> 15 <212> PRT <213> Artificial Sequence <220> <223> Las1l <400> 52 Ser Thr Gly Lys Ala Pro Tyr Thr Leu Asp Thr Leu His Glu Asp 1 5 10 15 <210> 53 <211> 15 <212> PRT <213> Artificial Sequence <220> <223> Irx1 <400> 53 Ala Leu Pro Glu Arg Asp Leu Val Thr Arg Pro Asp Trp Pro Pro 1 5 10 15 <210> 54 <211> 15 <212> PRT <213> Artificial Sequence <220> <223> Muc5b <400> 54 Gly Ala Thr Thr Met Ser Val Asn Ile Ser Thr Ile Gly Thr Asn 1 5 10 15 <210> 55 <211> 15 <212> PRT <213> Artificial Sequence <220> <223> Smarca5 <400> 55 Leu Gly Lys Thr Leu Gln Thr Ile Ser Leu Leu Gly Tyr Arg Lys 1 5 10 15 <210> 56 <211> 15 <212> PRT <213> Artificial Sequence <220> <223> Eri1 <400> 56 Tyr Asp Tyr Ile Cys Ile Ile Asp Phe Glu Ala Thr Cys Lys Glu 1 5 10 15 <210> 57 <211> 15 <212> PRT <213> Artificial Sequence <220> <223> Tph2 <400> 57 Phe Val Lys Ser Ile Thr Arg Pro Phe Ser Val Tyr Phe Asn Arg 1 5 10 15 <210> 58 <211> 9 <212> PRT <213> Artificial Sequence <220> <223> Eef2 <400> 58 Arg Ala Pro Ala Gly Ala Ala Ala Gly 1 5 <210> 59 <211> 15 <212> PRT <213> Artificial Sequence <220> <223> Ssrp1 <400> 59 Gly Arg Gly Asp Ser Ser Glu Arg Asp Lys Ser Lys Lys Lys Lys 1 5 10 15 <210> 60 <211> 15 <212> PRT <213> Artificial Sequence <220> <223> Oas1g <400> 60 Gly Lys Ser Asp Ala Asp Leu Val Val Phe Leu Asn Asn Leu Thr 1 5 10 15 <210> 61 <211> 15 <212> PRT <213> Artificial Sequence <220> <223> Gm15557 <400> 61 Ser Arg His Ser Cys Ala Glu Phe Val Lys Phe Arg Lys Ser Phe 1 5 10 15 <210> 62 <211> 15 <212> PRT <213> Artificial Sequence <220> <223> Ppp1cc <400> 62 Ser Phe Thr Ser Gly Ala Glu Val Val Ala Lys Phe Leu His Lys 1 5 10 15 <210> 63 <211> 15 <212> PRT <213> Artificial Sequence <220> <223> Pdcd6ip <400> 63 Thr Val Gly Thr Arg Ser Leu Ile Met Leu Ala Gln Ala Gln Glu 1 5 10 15 <210> 64 <211> 15 <212> PRT <213> Artificial Sequence <220> <223> Dync1h1 <400> 64 Pro Thr Ala Pro Ser Val Pro Ile Ile Asp Tyr Glu Val Ser Ile 1 5 10 15 <210> 65 <211> 15 <212> PRT <213> Artificial Sequence <220> <223> Eef2 <400> 65 Ser Gly Val Cys Val Gln Thr Glu Ile Val Leu Arg Gln Ala Ile 1 5 10 15

Claims

1. A method of screening for neoantigens, comprising: sequencing data of exome, transcriptome, single-cell transcriptome, peptidome or whole genome from a cancer patient; selecting a cancer cell survival dependency gene; and obtaining a neoantigen derived from the cancer cell survival dependency gene, wherein the step of selecting a cancer cell survival dependency gene comprises selecting a cancer cell survival dependency gene required for cancer cell survival using a cell survival dependency prediction model, the cell survival dependency prediction model is generated by training the relationship between cell gene expression and cell death, and genes that cause cancer cell death when their expression levels are reduced or removed are selected as cancer cell survival dependency genes, the step of obtaining a neoantigen derived from the cancer cell survival dependency gene comprises: comparing sequences from cancer cells and sequences from normal cells to obtain a neoantigen of a cancer patient according to sequencing data obtained from a cancer patient; and collecting a neoantigen derived from the cancer cell survival dependency gene among the obtained neoantigen.

2. The method of claim 1, further comprising: judging the binding affinity between the neoantigen and human leukocyte antigen of antigen presenting cells.

3. The method of claim 1, wherein, The selected cancer cell survival dependency gene causes cancer cell death when its expression level is reduced or removed, but does not affect the survival of normal cells.

4. The method of claim 1, wherein, The relationship between the gene expression of the cell and the cell death is based on in vitro data or computer data on whether a cancer cell line is killed due to the reduction or removal of the expression level of the targeted gene.

5. The method of claim 1, wherein, The step of selecting a cancer cell survival dependency gene further comprises judging whether the cancer cell survival dependency gene selected from the cell survival dependency prediction model is uniformly expressed in all cancer cells obtained from a cancer patient.

6. The method of claim 1, wherein, The step of collecting a neoantigen further comprises selecting a non-synonymous mutation of the cancer cell survival dependency gene.

7. The method of claim 1, wherein, The neoantigen is specific to a cancer patient.

8. The method of claim 2, wherein, The step of judging the binding affinity between the neoantigen and human leukocyte antigen of antigen presenting cells comprises inputting the sequence of the neoantigen into a neoantigen binding affinity prediction model to obtain a binding affinity prediction, the neoantigen binding affinity prediction model being used to predict the binding affinity between an antigen peptide and human leukocyte antigen of antigen presenting cells, the neoantigen binding affinity prediction model is generated by training interaction data between amino acids of a peptide and amino acids of human leukocyte antigen, and the human leukocyte antigen is MHC class I or MHC class II.

9. The method of claim 8, wherein, The antigen presenting cells are dendritic cells, macrophages, B cells or a combination thereof.

10. The method of claim 8, wherein, When the convolutional neural network-major histocompatibility complex value between the neoantigen and the human leukocyte antigen of the antigen presenting cells is >0.5, it is determined that there is binding affinity in the neoantigen.

11. A system for screening a neoantigen, comprising: a memory for storing at least one instruction; and at least one processor configured to execute the at least one instruction stored in the memory, the processor, by executing the at least one instruction, training a relationship between expression levels of genes of a cell and cell death to generate a cell survival dependency prediction model that predicts dependency of cell survival on gene expression, wherein the cell survival dependency prediction model is generated by training a relationship between expression levels of genes of a cell and cell death, selecting cancer cell survival dependency genes that cause cancer cell death when their expression levels are reduced or removed, comparing a gene expression profile of a cancer patient with a gene expression profile of normal cells or a normal control group to obtain neoantigens, and collecting neoantigens derived from the cancer cell survival dependency genes; inputting a gene expression profile of a cancer patient into the cell survival dependency prediction model to select cancer cell from or dependency genes; generating a neoantigen binding affinity prediction model for predicting binding affinity according to amino acid interactions of a peptide with an antigen presenting cell, and selecting neoantigens having binding affinity with human leukocyte antigens of an antigen presenting cell using the neoantigen binding affinity prediction model.

12. The system of claim 11, wherein, the processor, by executing the at least one instruction, selects a neoantigen as having binding affinity when a convolutional neural network-major histocompatibility complex value between the neoantigen and the human leukocyte antigens of the antigen presenting cell is >0.

5.

13. The system of claim 11, wherein, the processor, by executing the at least one instruction, learns a relationship between gene expression and cell death and a relationship between neoantigens and binding affinity of the neoantigens with human leukocyte antigens of an antigen presenting cell, respectively.

14. The system of claim 11, wherein, the relationship between gene expression and cell death of the cell is based on in vitro data or computer data on whether a cancer cell line is killed due to a reduction in expression levels of targeted genes or removal.

15. The system of claim 11, wherein, the neoantigen binding affinity prediction model is generated by training interaction data between amino acids of a peptide and amino acids of human leukocyte antigens.

16. The system of claim 11, wherein, the gene expression profile of the cancer patient is sequencing data of an exome, a transcriptome, a single-cell transcriptome, a peptidome, or a whole genome. 17.A method of preparing an anticancer vaccine, comprising: obtaining neoantigens by the method of claim 1; and preparing an anticancer vaccine including the neoantigens, wherein the preparing an anticancer vaccine includes obtaining a peptide sequence including the neoantigens, the peptide sequence consisting of 9 to 30 amino acids; and selecting a peptide sequence having affinity and stability from the peptide sequence.

18. The method of claim 17, wherein, the selected peptide sequence has Kyte-Doolittle GRAVY <0, InstaIndex <40.

Images