Method of selecting neoantigen for development of personalized cancer vaccine

Abstract

Provided are a method of selecting a tumor-specific neoantigen (immunogenic peptide), and a use of the selected tumor-specific neoantigen for preparing a personalized cancer vaccine.

Description

A method for selecting novel antigens for developing personalized anticancer vaccines Cross-citation with related applications This application claims the benefit of priority from Republic of Korea Patent Application No. 10-2023-0110887, dated August 23, 2023, the entire contents of which are incorporated herein by reference. The present application provides a method for selecting a neoantigen (tumor-specific immunogenic peptide) and a use of the selected tumor-associated neoantigen for producing a personalized anticancer vaccine. As used herein, the term “neoantigen” refers to an antigen that arises due to a cancer-specific mutation, more specifically, a peptide (surface protein) that is specifically expressed on the cell surface of a cancer cell and contains an antigenic determinant and can be recognized as an antigen. Since it is expressed only in cancer cells, it can be applied to the development of targeted anticancer drugs. As used herein, a neoantigen is a type of immunogenic peptide, and can also be expressed as a “tumor-specific neoantigen,” “tumor-specific antigen,” or “tumor-specific immunogenic peptide.” “Personalized Cancer Vaccine” refers to a treatment that activates the patient’s immune system to remove cancer cells, or a treatment that uses the above treatment, by selecting a neoantigen with immune-boosting ability among cancer-specific antigens expressed in each patient, producing it in the form of protein or mRNA, and administering it to the patient using a delivery material such as lipid-based nanoparticles including classifications such as Lipid Nanoparticles (LNP and Solid Lipid Nanoparticle (SLN), Nanostructured Lipid Carrier (NLC), etc.) and liposome-nucleic acid complexes (e.g., lipoplex, liposomes, etc.). To maximize the efficacy of personalized anticancer vaccines, identifying individual patient neoantigens and prioritizing and selecting immunogenic neoantigens is a key factor. Immunogenic neoantigens must bind to human leukocyte antigens (HLA), which play a role in the patient's immune response to antigen stimulation and in regulating cell-mediated and humoral immunity. Immune activation against the neoantigen occurs when the T cell receptor (T cell receptor) recognizes the specific neoantigen sequence bound to HLA. However, because cancer-specific mutations and HLA types expressed by each patient vary, it is difficult to discover immunogenic neoantigens that can be applied to the development of personalized anticancer vaccines. Furthermore, there is a lack of established methods for collecting and selecting data for training AI prediction models that can predict immunogenic neoantigens, and for predicting and selecting neoantigens applicable to patients when developing personalized anticancer vaccines, making it difficult to efficiently use neoantigen prediction models. Previously, AI prediction models were developed that measured the binding affinity between immunogenic peptides containing mutations in patient cancer cells and HLA to select neoantigens expected to have anticancer efficacy when applied to anticancer vaccines. However, high binding affinity between patient epitopes and HLA did not necessarily result in a high T cell immune response, making it difficult to use the derived immunogenic peptides as vaccines. In addition, there was a technology to compare the amino acid physicochemical properties between peptides containing mutations in cancer cells and peptides containing non-mutations in non-cancerous cells to predict T cell immunogenicity, but the amino acid sequence information alone did not reflect the three-dimensional protein structure, so the HLA-peptide complex combined with the neoantigen peptide containing mutations in cancer cells was recognized comprehensively by the T cell receptor and caused an immune response, so the binding strength of HLA and neoantigens alone did not cover various immune responses, and thus the success rate of predicting immune responses was not high. An example of the present application provides a method (prediction method, selection method, identification method, or method providing information for selection) of a tumor-specific immunogenic peptide (neoantigen, tumor-specific antigen). The above tumor-specific immunogenic peptide selection method may include (I) a step of constructing an immunogenic peptide prediction model and (II) a step of predicting and / or selecting a tumor-specific immunogenic peptide. The above (I) immunogenic peptide prediction model construction step may include the following steps (a) to (d): (a) A step of obtaining peptide information from a peptide database, wherein the peptide information may include one or more pieces of information selected from sequence information of a peptide derived from a tumor cell or tumor tissue, amino acid sequence information of a peptide derived from a normal cell or normal tissue, class I HLA type information to which the peptide binds, binding affinity information between the peptide and the class I HLA, and / or immunogenicity information of the peptide; (b) a step of obtaining protein sequence information of the class I HLA from a biological sequence database; (c) a step of building a binding affinity prediction model by training a model capable of predicting a binding affinity score (BA) for HLA using the information obtained in steps (a) and (b) above, and obtaining a binding score; and (d) A step of constructing an immunogenicity prediction model by training a model capable of predicting an immunogenicity score (IMM) using the information obtained in steps (a) and (b) and obtaining an immunogenicity score. The peptide database usable in (a) of the above step (I) may be one or more peptide databases selected from the group consisting of IEDB (Immune Epitope Database and Analysis Resource), MHCBN (Comprehensive Database of MHC Binding and Non-binding Peptides), etc., but is not limited thereto, and any database including peptide information, for example, amino acid sequence information of peptides of human origin (derived from normal cells / normal tissues and derived from tumor cells / tumor tissues), binding information for HLA (e.g., IC50, KD value, etc.) and / or immunogenicity-related information may be used without any particular limitation. The biological sequence database available in (b) of the above step (I) may be one or more protein databases selected from the group consisting of the National Center for Biotechnology Information (NCBI), Protein database, MHC Motif Atlas, etc., but is not limited thereto, and any database from which the amino acid sequence of the HLA protein can be obtained may be used for learning without any special restrictions. Information obtained from the above peptide database and / or biological sequence database can be used to learn the prediction model in steps (c) and (d). The above (II) tumor-specific immunogenic peptide prediction and / or selection step may include the following steps (1) to (7): (1) A step of obtaining NGS sequence information from patient-derived tumor cells or tumor tissues and normal cells or normal tissues; (2) A step of obtaining sequence information of a tumor-specific peptide having a length of 7 amino acids or more and having at least one amino acid mutation in the patient-derived tumor cell or tumor tissue NGS sequence by comparing it with a normal cell or normal tissue NGS sequence; (3) A step of measuring the expression level (TPM; Transcripts per Million) of the tumor-specific peptide or the coding gene of the tumor-specific peptide; (4) A step of applying the tumor-specific peptide sequence information to the binding affinity prediction model constructed in step (c) to measure (confirm, predict) the binding affinity score (BA) for the HLA molecule; (5) A step of measuring (confirming, predicting) an immunogenicity score (IMM) by applying the tumor-specific peptide sequence information to the immunogenicity prediction model constructed in step (d); (6) Priority determination step to determine the ranking of tumor-specific peptides based on the values ​​calculated by the following formula: Neoantigen prioritization score = min(TPM,1) * (W1*BA + W2*IMM) / (W1+W2) [0 <W1<1(또는 0.1≤W1≤0.9), 0<W2<1(또는 0.1≤W2≤0.9) 범위에서 각각 독립적으로 설정; “min(TPM,1)”은 0<TPM≤1인 경우 해당 값을 그대로 사용하고, 1≤TPM인 경우 TPM=1로 하여 계산함(TPM이 1 이상(예컨대, 100, 1000)인 경우 Score에 Bias가 커질 수 있어서 이로 인한 결과 해석의 오류를 줄이기 위함)을 의미함]; 및 (7) A step of selecting two or more tumor-specific immunogenic peptides based on the priority of the tumor-specific peptides determined above. The NGS (next-generation sequencing) sequence information of steps (1) and (2) above may be a nucleic acid (gene) sequence and / or sequence information converted into an amino acid (protein) sequence. In one example, step (1) may further include, after the step of obtaining the NGS sequence information, a step of converting the obtained NGS sequence into an amino acid sequence. The tumor cells or tumor tissues and normal cells or normal tissues of steps (1) and / or (2) above may be derived from the same patient, but are not limited thereto. The TPM measurement step of the above step (3) may be performed on tumor cells and / or tumor tissues, and the tumor cells and / or tumor tissues may be obtained from patient-derived samples or available tumor (cancer) cell lines. When the TPM measurement of the above step (3) is performed on patient-derived tumor cells or tumor tissues, the tumor cells or tumor tissues of the above steps (1) and / or (2) may be derived from the same patient as the normal cells or normal tissues, but is not limited thereto. There is no particular limitation on the order in which the TPM measurement step of step (3), the combined score measurement step of step (4), and the immunogenicity score measurement step of step (5) are performed. For example, they may be performed sequentially regardless of the order, or two or more may be performed simultaneously, but this is not limited thereto. The above-mentioned two or more tumor-specific immunogenic peptides can be used as candidates for the final tumor-specific immunogenic peptide (neoantigen). In one example, the method for selecting a tumor-specific immunogenic peptide provided herein may include the step of predicting and / or selecting a tumor-specific immunogenic peptide (i.e., steps (1) to (7)) described above (II). In another example, the method for selecting a tumor-specific immunogenic peptide may further include, in addition to the step of predicting and / or selecting a tumor-specific immunogenic peptide (II), a step of constructing an immunogenic peptide prediction model (i.e., steps (a) to (d)). In this case, the additionally included step of constructing an immunogenic peptide prediction model (I) may be included before the step of predicting and / or selecting a tumor-specific immunogenic peptide (II). For the peptide database used in building the prediction model in the above step (I) of the above selection method, data obtained by going through one or more of the following selection processes selected from i) to iv) may be used: i) Deletion of data not corresponding to HLA-A, HLA-B, and HLA-C; ii) Deletion of data with mutations in HLA molecules; iii) Deletion of data with a peptide length of 7 or less or 15 or more, and iv) Removal of amino acid residues other than the 20 human-derived amino acids in the amino acid sequence. The above step (I) (one or more of steps (a) to (d)) and step (II) (one or more of steps (1) to (7)) may be performed in a computer processing device such as a computer, or a computer processing apparatus. The peptide database that can be used for learning the binding affinity prediction model of the above step (I) may be one or more peptide databases selected from the group consisting of IEDB (Immune Epitope Database and Analysis Resource), MHCBN (Comprehensive Database of MHC Binding and Non-binding Peptides), etc., but is not limited thereto, and any database that includes peptide information, for example, amino acid sequence information of human-derived peptides, binding information for HLA (e.g., IC50, KD value, etc.) and / or immunogenicity-related information, may be used for learning without any particular limitation. In one specific example, the step of determining (measuring, determining) an immunogenicity score (IMM) includes two or more immunogenicity model training data associated with an immune response when a peptide-MHC (HLA) complex binds to a T cell receptor (TCR). The above immunogenicity model learning data may be two or more selected from the group consisting of all immunogenicity-related information for peptide candidate sequences obtained from a peptide database or related literature, such as T cell-derived cytokine secretion amounts including INF-g, TNF-a, IL-2, IL-1b, etc., T cell proliferation due to immunogenic activation, Chemokine levels including CCL4, CXCL9, values ​​measuring cytotoxicity of activated T cells such as Granzyme B, etc. Specifically, the peptide database may be one or more peptide databases selected from the group consisting of IEDB (Immune Epitope Database and Analysis Resource) and MHCBN (Comprehensive Database of MHC Binding and Non-binding Peptides), but is not limited thereto, and any database including peptide information, for example, amino acid sequence information of human-derived peptides and immunogenicity measurement values, may be used for learning without any particular limitation. The above immunogenicity measurement values ​​are values ​​measured by methods including ELISPOT, ELISA, Flow cytometry intracellular staining, binding assay, Proliferation assay (e.g., T cell Proliferation assay, etc.), cytotoxicity assay (e.g., in vitro / ex vivo / in vivo Cytotoxicity assay, etc.). The above immunogenicity model learning data includes both qualitatively measured values ​​and quantitatively measured values, and among them, the qualitatively measured values ​​are statistically converted into numerical values, such as beta distribution transformation, compared to the negative control group and included as data. The above step (II) may be performed using NGS sequence information obtained from samples (tumor cells / tissues and normal cells / tissues) derived from a personalized target patient (mammal; e.g., human) and / or information converted into peptide information thereof. The tumor-specific immunogenic peptide may be a peptide derived from a cancer cell mutation. In one example, the tumor-specific immunogenic peptide may be derived from a sequence obtained by genome sequence analysis (e.g., Next Generation Sequencing (NGS) etc.) of a patient-derived sample (e.g., cancer cells) (e.g., a nucleic acid sequence obtained by NGS or an amino acid sequence converted therefrom). More specifically, the tumor-specific immunogenic peptide may be derived from a sequence (a nucleic acid sequence or an amino acid sequence converted therefrom) obtained by genome sequence analysis (e.g., Next Generation Sequencing (NGS) etc.) of a patient-derived sample (e.g., cancer cells) that is different from a normal cell (e.g., a sequence in which a mutation occurs compared to a normal cell). The HLA molecule may be one or more selected from the group consisting of human-derived major histocompatibility complex (MHC) molecules. The HLA molecule may comprise different HLA molecule types and / or different HLA alleles. In one example, the HLA molecule may be an HLA type I molecule. In the above immunogenic peptide prediction model construction step (step (I)), the combined score of step (c) may be obtained by combining (ensembling) prediction values ​​obtained by learning in two or more ways using the amino acid sequence information of the immunogenic peptide and the HLA molecule. The amino acid sequence information of the HLA molecule may be obtained from the database of candidate peptides described above, but is not limited thereto. The immunogenicity score of the above step (d) may be obtained by the amount of cytokine secretion when the immunogenic peptide is treated to a sample containing immune cells. The sample containing the immune cells is selected from the group consisting of blood, leukocytes, and peripheral blood mononuclear cells (PBMCs), and the cytokine may be at least one selected from the group consisting of IFN-g, IL-2, TNF-a, and granzyme B, but is not limited thereto. In one example, the immunogenicity score of the above step (2) may be determined by information on the amount of cytokine secretion when the candidate peptide is treated to immune cells (e.g., PBMCs, etc.) or a sample containing the same. The information on the amount of cytokine secretion may be obtained from a database selected from the group consisting of, for example, IEDB, MHCBN, etc., but is not limited thereto. The immunogenicity score can be indirectly measured as the ratio of positive experiments to the total number of experiments. In one example, the IEDB database provides qualitative measurements, the number of experiments (tested), the number of reactions (responded), and the ratio (frequency) as data instead of quantitative measurements of cytokine secretion, which can be used as training data for quantifying the immunogenicity score. The actual experimental values ​​can be converted into numbers by reflecting the ratio of the number of reactions to the number of experiments, and used as training data when retraining the immunogenicity prediction model. The quantification can be performed by one or more methods selected from the group consisting of mean, median, and beta distribution transformation. The immunogenicity score of the above step (d) may be obtained by combining (ensembling) predicted values ​​obtained by learning in two or more ways using the peptide amino acid sequence information of the immunogenicity learning data, the amino acid sequence corresponding to the HLA molecule of the immune cell, and the information on the amount of cytokine secretion when the peptide is treated by the immune cell. The ensemble of steps (c) and (d) above may be performed by selecting two or more prediction models trained using two or more different learning methods, and combining the prediction values ​​by selecting two or more models with the smallest prediction errors for a validation dataset different from the training data. Specifically, the validation dataset may be a portion of the model training data separated in advance. In the above step (c) (combined prediction modeling and score measurement) and / or step (d) (immunogenicity prediction modeling and score measurement), - The above learning can be performed with one or more models selected from commonly used AI (artificial intelligence) learning models such as GRU (Gated Recurrent Unit), Transformer, FFN (Feed Forward Neural-network), GNN (Graph Neural Network), etc. - The learning of the above steps (c) (combined score measurement) and (d) (immunogenicity score measurement) can be performed by combining the AI ​​model with one learning method selected from among the application and non-application of learning methods commonly used in the relevant field. - The above combined score and / or immunogenicity score may be a final predicted value obtained by combining (ensembling) the predicted values ​​obtained from each of the prediction models obtained by learning with one or more AI models and one or more learning methods by a typical ensemble model, and in one example, the final predicted value may be obtained as [(average of predicted values) - (variance of predicted values)]. In the above tumor-specific immunogenic peptide prediction and / or selection step (step (II)), step (1) may be performed through a bioinformatics analysis process between the NGS sequence of patient-derived tumor cells or tumor tissues and the NGS sequence of patient-derived normal cells or normal tissues. The tumor cells or tumor tissues and the normal cells or normal tissues may be derived from the same patient. The bioinformatics analysis process may include the following steps i) to vii), but is not limited thereto, and detailed steps may be added or removed. i) Verifying the integrity of NGS data; this may be done using software such as, but not limited to, FastQC. ii) Removal of low-quality data; this may be done using software such as, but not limited to, Cutadapt. iii) Sequence alignment to a reference genome; for example, but not limited to, the hg19 reference genome and bwa software. iv) Quality verification and correction of sequence alignment results; for example, using software such as GATK, Indel realignment, BQSR (Base Quality Score Recalibration), etc. can be applied, but are not limited thereto. v) Tumor-specific mutation discovery; examples include, but are not limited to, software such as GATK, Strelka, and Mutect2; vi) Classification of tumor-specific mutations; for example, software such as SnpEff can be used to classify mutations, and only missense, insertion, and deletion mutations that affect peptide sequences can be selected, but this is not limited thereto. vii) Extraction of peptide sequences with a length of 7 or more that can be generated, including tumor-specific mutations. The TPM (transcriptome per million) of the above step (3) is an expression value that measures the degree of expression in the process of expression in which a gene composed of a base sequence is transcribed to form a transcriptome. Methods for measuring the expression value include, but are not limited to, measuring the mRNA concentration and measuring the concentration of the generated transcript, and the expression value can be measured in various ways. According to one embodiment of the present invention, the expression value is the number of reads obtained as a result of NGS mapped to the transcriptome, and can be calculated in TPM units, but is not limited thereto and can be calculated in various other units to represent the expression value. A step of filtering an RNA base sequence having a TPM (transcriptome per million) of 1 or more among a mutant base sequence or a total RNA base sequence may be further included. The expression level of the tumor-specific peptide or its coding gene in the above step (3) (e.g., TPM, Transcripts Per Million; Li et al. (2010); normalized based on the proportion of each transcript (number of mapped NGS reads) in the library, the sum of all TPM values ​​in one library is 1,000,000, and comparison between different samples is possible) may indicate the possibility that the tumor-specific peptide exists in an actual patient sample (e.g., cancer cells, etc.) (e.g., the possibility, frequency, or degree that a genetic mutation in a cancer cell is produced into a protein through transcription and translation). The expression level of the tumor-specific peptide or its coding gene in the above step (3) may be obtained through sequence analysis of patient sample data that is the target of personalized treatment, and more specifically, may be calculated by processing microarray or RNA-seq data. In one example, patient sample RNA-seq data can be analyzed to calculate the TPM of the tumor-specific peptide or its encoding gene, and tumor-specific peptides having a TPM value of 1 or greater can be prioritized. For example, since TPM represents the expression level of the gene per million RNA-seq data normalized by gene length and data production, by using TPM ≥ 1 for prioritization, it is possible to prioritize candidate peptides that occupy at least one millionth of the total RNA-seq data and are likely to exist in actual patient samples among the candidate peptides. The priority determination step of the above step (6) is performed using the following formula: Neoantigen prioritization score = min(TPM,1) * (W1*BA + W2*IMM) / (W1+W2). In the above formula, W1 and W2 represent the weights of the binding score (BA) and immunogenicity score (IMM), respectively, and are 0 <W1<1 (또는 0.1≤W1≤0.9), 0<W2<1 (또는 0.1≤W2≤0.9) 범위서 각각 독립적으로 설정할 수 있으며, 일 예에서, W1+W2=1일 수 있으나, 이에 제한되는 것은 아니다. 일 예에서, 상기 수식은 TPM > 1, 0.6 > It may be performed on candidate peptides that satisfy the conditions of avidity score (BA) ≥ 0.1, and immunogenicity score (IMM) ≥ 0.25, but is not limited thereto. The selection step of the tumor-specific immunogenic peptides (neoantigens) of the above step (7) is to sort the results of the above step (6) in descending order and select one or more tumor-specific immunogenic peptides from among the top 2 to 100, 2 to 75, 2 to 50, 2 to 40, 2 to 30, 2 to 20, 2 to 10, 5 to 100, 5 to 75, 5 to 50, 5 to 40, 5 to 30, 5 to 20, 5 to 10, 10 to 100, 10 to 75, 10 to 50, 10 to 40, 10 to 30, 10 to 20, 20 to 100, 20 to 75, 20 to 50, 20 to 40, or 20 to 30 peptides. An additional step of selecting a peptide may be included. When the neoantigen (tumor-specific immunogenic peptide) prediction and personalized anticancer vaccine provided herein are produced as mRNA, there may be no need to specifically consider hydrophobicity. However, when producing an anticancer vaccine as a peptide or conducting an immunogenicity test, it may be necessary to measure the hydrophobicity of the peptide, because production is impossible if the peptide has high hydrophobicity (e.g., high hydrophobic amino acid content). In one specific example, the hydrophobicity may be measured by calculating the Gravy (Grand average of hydropathy) score, but is not limited thereto. The tumor-associated immunogenic peptide selected by the method for selecting a tumor-specific immunogenic peptide provided herein may be a peptide (e.g., a mutant peptide) that is specifically present in a cancer (tumor) sample and is not present in a normal sample that does not have cancer (tumor). Therefore, the tumor-specific immunogenic peptide can be usefully used in the preparation of a pharmaceutical composition for the prevention and / or treatment of cancer, such as an anticancer vaccine, and particularly, a personalized anticancer vaccine for a patient used in the selection method. Another example provides an anticancer vaccine composition comprising a tumor-specific immunogenic peptide (neoantigen) selected using the above-described tumor-associated immunogenic peptide selection method and / or a gene (mRNA, DNA, etc.) encoding the same. The anticancer vaccine composition may additionally comprise a pharmaceutically acceptable carrier and / or excipient and / or one or more adjuvants, stabilizers, etc. Another example provides a method for producing an anticancer vaccine, comprising a step of mixing a tumor-specific immunogenic peptide selected by the method for selecting the tumor-specific immunogenic peptide and / or a gene (mRNA, DNA, etc.) encoding the same with a pharmaceutically acceptable carrier and / or and / or one or more adjuvants, stabilizers, etc. The carrier and / or excipient and / or one or more adjuvants, stabilizers, etc. may be selected from among all pharmaceutically acceptable substances commonly used in vaccine formulation. The above vaccine may be in the form of a therapeutic vaccine or a prophylactic vaccine. Another example provides a method for preventing and / or treating cancer, comprising administering to a patient in need of prevention and / or treatment of cancer a tumor-specific immunogenic peptide selected by the method for selecting the tumor-specific immunogenic peptide or the vaccine. The treatment of the above cancer may be combined with other anticancer agents, and / or surgical resection, radiotherapy, and / or conventional chemotherapy. The other anticancer agents may be selected from any known anticancer agent (e.g., chemotherapeutic agents, antibodies, etc.), including, but not limited to, immunotherapy agents (e.g., immune checkpoint inhibitors (e.g., PD-1 inhibitors and / or PD-L1 inhibitors, etc.)). To identify highly immunogenic neoantigens among cancer-specific sequence mutations, we developed a neoantigen prediction model using AI learning technology that combines a Binding Affinity model, which predicts the binding affinity of HLA and neoantigen sequences, with an Immunogenicity model, which predicts the immunogenicity of neoantigen sequences. The neoantigen prediction model was developed based on the database in the Immune Epitope Database and Analysis Resource (IEDB), with algorithm learning for each model performed and in silico improvements achieved through data updates. By training an AI neoantigen prediction model, we were able to predict immunogenic neoantigen candidates, and by applying the prediction model to cancer patient NGS data, we derived a group of neoantigen candidates. When applying the prediction model, we set thresholds for the neoantigen candidates predicted by the binding affinity and immunogenicity models. We then produced peptides for the neoantigens and performed binding affinity and immunogenicity tests (e.g., IFNγ ELISPOT) to verify the model's neoantigen prediction rate and select neoantigens applicable to personalized anticancer vaccines. The composition of the present application in one aspect is exemplified as follows: Neoantigen prediction model (Transformer based Binding Affinity Model + Immunogencity model); Selection of input data for the new antigen prediction model; Learning method of new antigen prediction model; Method of processing WES (whole exosome sequencing; mutation selection) and RNA-seq (expression level) data of patient data sets; Neoantigen Prioritization (Confirm Neoantigen Ranking Score using TPM, BA Score and IMM Score; Binding affinity and immunogenicity tests were conducted to verify the applicability of the predicted new antigen to personalized anticancer vaccines. Definition of terms In the present invention, the term "peptide" means a substance comprising 2 or more, 3 or more, 4 or more, 6 or more, 8 or more, 10 or more, 13 or more, 16 or more, 21 or more, and at most 8, 10, 20, 30, 40, 50 or 100 amino acids covalently linked by peptide bonds. The term "polypeptide" or "protein" may mean a substance in which more amino acid residues than a peptide are covalently linked by peptide bonds, but in the present specification, "peptide", "polypeptide" and "protein" may be used interchangeably. In the present invention, the term "modification, mutation" in relation to a peptide, polypeptide or protein may mean that a mutation is introduced into the original sequence (parental sequence) of the wild-type peptide, polypeptide or protein, and the mutation may be selected from the group consisting of an amino acid insertion mutation, an amino acid addition mutation, an amino acid deletion mutation and an amino acid substitution mutation, and may be, for example, an amino acid substitution mutation. In the present specification, all sequence changes can potentially create new epitopes. In this specification, the term “immunogenic peptide” means a peptide capable of inducing an immune response, including an antigenic determinant (epitope). As used herein, the term "neoantigen" may be a peptide or protein that contains one or more amino acid modifications compared to a parental peptide or protein, wherein the amino acid modifications may be amino acid modifications resulting from tumor-specific mutations. As used herein, a neoantigen is a type of immunogenic peptide that contains tumor-specific mutations and may not be present or detectable in normal cells. The neoantigen may also be expressed as a tumor-specific neoantigen (tumor-specific antigen) or a tumor-specific immunogenic peptide. The term "immune response" refers to the body's integrated response to a target, such as an antigen, and may refer to, for example, a cellular immune response or a cellular and humoral immune response. The immune response may be protective / preventive / preventative and / or therapeutic. "Inducing an immune response" can refer to the absence of an immune response prior to induction, but it can also mean that a certain level of immune response existed prior to induction and that this immune response was enhanced after induction. Therefore, "inducing an immune response" also encompasses the meaning of "enhancing an immune response." For example, after inducing an immune response in a given subject (patient), the subject may be protected from the development of a disease, such as cancer, or the condition may be improved, alleviated, and / or mitigated by the induction of an immune response. The terms "cellular immune response" and "cellular response" or similar terms refer to an immune response directed against cells characterized by the presentation of antigens by class I or class II MHC associated with T cells or T-lymphocytes that act as "helper" or "killer" cells. Helper T cells (also called CD4+ T cells) play a central role in regulating the immune response, and killer cells (also called CD8+ T cells, cytotoxic T cells, cytolytic T cells, CD8+ T cells, or CTLs) kill diseased cells, such as cancer cells, thereby preventing the generation of diseased cells. In a preferred embodiment, the present invention may comprise stimulating an anti-tumor CTL response against a tumor cell expressing one or more tumor-expressed antigens, and / or presenting such tumor-expressed antigens by HLA class I or class I MHC. "Antigen" encompasses any substance, such as a peptide or protein, that is the target of an immune response, such as a specific reaction with an antibody or T-lymphocyte (T cell), and / or that induces such an immune response. For example, an antigen comprises at least one epitope, such as a T cell epitope. As used herein, an antigen may be a molecule (including a cell expressing the antigen) that, optionally after processing, induces an immune response specific for that antigen. An antigen or its T cell epitope may be presented in the context of an MHC molecule by an antigen presenting cell, such as a cell, such as a diseased cell, particularly a cancer cell, to induce an immune response against the antigen (including a cell expressing the antigen). In this specification, the terms “tumor-specific antigen”, “tumor-specific expressed antigen”, “cancer-specific antigen” and “cancer-specific expressed antigen” have equivalent meanings and can be used interchangeably with neoantigen. The term "immunogenicity" may refer to the ability to induce an immune response associated with a treatment, such as a treatment for cancer (e.g., antibody formation, etc.). The term "major histocompatibility complex (MHC)" refers to the genetic complex that occurs in all vertebrates, including MHC class I and MHC class II molecules. MHC proteins or molecules are important in signaling between lymphocytes and antigen-presenting cells or diseased cells in the immune response, where MHC proteins or molecules bind peptides and present them for recognition by T cell receptors. Proteins encoded by MHC are expressed on the cell surface, presenting both self-antigens (peptide fragments from the cell itself) and non-self-antigens (such as fragments from invading microorganisms) to T cells. The MHC domain is divided into three subgroups: class I, class II, and class III. MHC class I proteins contain an α-chain and a β2-microglobulin (not part of the MHC encoded by chromosome 15). They present antigen fragments to cytotoxic T cells. On most immune cells, particularly antigen-presenting cells, MHC class II proteins contain α and β-chains and present antigen fragments to T helper cells. The MHC class III domain encodes other immune components, such as those encoding complement components and cytokines. The MHC is both polygenic (there are several MHC class I and MHC class II genes) and polymorphic (there are multiple alleles for each gene). In the present invention, the term "haplotype" refers to an HLA allele found on a single chromosome and the protein encoded by it. A haplotype may also refer to an allele present at any location within the MHC. Each class of MHC can be represented by several locations: for example, - For class I, HLA (Human Leukocyte Antigen)-A, HLA-B, HLA-C, HLA-E, HLA-F, HLA-G, HLA-H, HLA-J, HLA-K, HLA-L, HLA-P, HLA-V, etc.; and - For class II, HLA-DR (HLA-DRA, HLA-DRB1-9, etc.), HLA-DQ (HLA-DQA1, HLA-DQB1, etc.), HLA-DP (HLA-DPA1, HLA-DPB1, etc.), HLA-DM (HLA-DMA, HLA-DMB, etc.), HLA-DO (HLA-DOA, HLA-DOB, etc.), etc. In this specification, the terms “HLA or HLA allele” and “MHC or MHC allele” may be used interchangeably. MHCs exhibit a high degree of polymorphism. For each locus in the human population, a large number of haplotypes, each containing distinct alleles, exist. Different polymorphic MHC alleles, both class I and class II, possess distinct peptide specificities: each allele encodes a protein that binds a peptide with a specific sequence pattern. In one specific example, the MHC molecule may be an HLA molecule. When designating an HLA molecule as a serological type based on HLA DNA type, reference may be made to “HLA dictionary 2008” (John Wiley & Sons A / S; Tissue Antigens 73, 95-170; incorporated herein by reference). The term "MHC binding peptide" may include an MHC class I and / or class II binding peptide or peptides that can be processed to produce an MHC class I and / or class II binding peptide. For a class I MHC / peptide complex, the peptide (binding peptide) may typically be, but is not limited to, 7-12, 7-10, 8-12, or 8-10 amino acids in length. For a class II MHC / peptide complex, the peptide (binding peptide) may typically be, but is not limited to, 9-30 or 10-25 amino acids in length, such as, but is not limited to, 13-18 amino acids in length. In one example, when the peptide is presented directly, i.e. without processing, in particular without cleavage, the length may be any length suitable for binding to an MHC molecule, in particular a class I MHC molecule, such as 7-30, 7-20, 7-12, 8-11, 9, or 10 amino acids in length. In another example, when the peptide is part of a larger entity comprising additional sequence, such as a vaccine sequence or polypeptide, and is presented after processing, in particular cleavage, the length of the peptide produced by processing may be any length suitable for binding to an MHC molecule, in particular a class I MHC molecule, such as 7-30, 7-20, 7-12, 8-11, 9, or 10 amino acids in length. The tumor-specific immunogenic peptide selected by the method provided herein may bind to a class I MHC (HLA) molecule. The term "epitope" refers to an antigenic determinant within a molecule, such as an antigen, i.e., a portion or fragment of a molecule recognized by the immune system. An epitope of a protein, such as a tumor antigen, comprises a continuous or discontinuous portion of the protein and may be 5 to 100, 5 to 50, 8 to 30, or 10 to 25 amino acids in length. As used herein, an epitope may be a T cell epitope. In this specification, an epitope may be an “MHC binding peptide” capable of binding to an MHC molecule on a cell surface. As used herein, the term "neo-epitope" refers to an epitope found in cancer cells but not in a reference cell, such as a normal non-cancerous cell or germ cell. This particularly includes situations where a corresponding epitope is found in a normal non-cancerous cell or germ cell, but one or more mutations in the cancer cell alter the sequence of the epitope, resulting in a neo-epitope. In the present invention, the term "T cell epitope" refers to a peptide that binds to an MHC molecule in a configuration recognized by a T cell receptor. Typically, a T cell epitope is present on the surface of an antigen-presenting cell. A T cell epitope may comprise an amino acid sequence substantially corresponding to the amino acid sequence of an antigen fragment. For example, the antigen fragment may be an MHC class I and / or class II presented peptide. A T cell epitope according to the present invention may be associated with a part or fragment of an antigen capable of stimulating an immune response, or a part or fragment of an antigen capable of stimulating a cellular response against an antigen or cell characterized by the expression of the antigen and, preferably, by presentation of the antigen, such as a diseased cell, particularly a cancer cell. For example, the T cell epitope may be capable of stimulating a cellular response against a cell characterized by presentation of the antigen by class I MHC, and may be capable of stimulating antigen-reactive cytotoxic T-lymphocytes (CTLs). "Antigen processing" or "processing" means that a peptide, polypeptide or protein is broken down (e.g., a polypeptide is broken down into peptides) and one or more of these fragments are then associated with an MHC molecule (e.g., via binding) so that they are presented to specific T cells by a cell, such as an antigen presenting cell. The terms “T cell” and “T lymphocyte” are used interchangeably herein and include cytotoxic T cells (CTLs, CD8+ T cells) and T helper cells (CD4+ T cells), including cytolytic T cells. T cells belong to a group of white blood cells known as lymphocytes and play a central role in cell-mediated immunity. They are distinguished from other lymphocytes, such as B cells and natural killer cells, by the presence of a specialized receptor on their cell surface, called the T cell receptor (TCR). The thymus is the primary organ responsible for T cell maturation. Several different T cell subsets have been discovered, each with a unique function. T helper cells play a role in assisting other white blood cells in immunological processes, particularly in the maturation of B cells into plasma cells and the activation of cytotoxic T cells and macrophages. These cells are also known as CD4+ T cells because they express the CD4 protein on their surface. Helper T cells are activated when presented with peptide antigens by MHC class II molecules expressed on the surface of antigen-presenting cells (APCs). Once activated, they rapidly divide and secrete small proteins called cytokines that regulate or assist the active immune response. Cytotoxic T cells destroy virus-infected and tumor cells and are also involved in transplant rejection. These cells are also known as CD8+ T cells because they express the CD8 glycoprotein on their surface. These cells recognize their targets by binding to antigens associated with MHC class I, which are present on the surface of virtually every cell in the body. Most T cells possess a T-cell receptor (TCR), which exists as a complex of several proteins. The actual T-cell receptor is composed of two distinct peptide chains, referred to as the α- and β-TCR chains, which are produced from independent T-cell receptor α and β (TCRα and TCRβ) genes. γδ T cells (γδ T cells) represent a small subset of T cells that possess a unique T-cell receptor (TCR) on their surface. However, in γδ T cells, the TCR consists of one γ-chain and one δ-chain. This group of T cells is much rarer (2% of all T cells) than αβ T cells. The first signal for T cell activation is provided by the binding of the T cell receptor to a short peptide presented by the MHC on another cell. This ensures that only T cells with a TCR specific for that peptide are activated. The partner cells are typically antigen-presenting cells, such as professional antigen-presenting cells (APCs). In naive responses, these are typically dendritic cells, but B cells and macrophages can also be important APCs. In this specification, if a molecule has significant affinity for a given target and binds to said target in a standard assay, then the molecule is said to be capable of binding to the given target. "Affinity" or "binding affinity" can be measured by the equilibrium dissociation constant (KD). If a molecule does not have significant affinity for the target and does not significantly bind to the target in a standard assay, then the molecule is said to (substantially) not bind to the target. Specific activation of CD4+ or CD8+ T cells can be confirmed in a variety of ways. Methods for confirming specific T cell activation include assessing T cell proliferation, cytokine production (e.g., lymphokines), or cytolytic activity. For CD4+ T cells, specific T cell activation can be confirmed through T cell proliferation. For CD8+ T cells, specific T cell activation can be confirmed through cytolytic activity. In this specification, the term "score" is the result of any test or examination, usually expressed numerically. In the present invention, determining the binding score of a given peptide to an MHC molecule includes detecting the binding potential of the peptide to the MHC molecule. The binding score of a peptide to the above MHC molecule can be determined using a peptide:MHC binding prediction tool. For example, the Immune Epitope Database Analysis Resource (IEDB-AR: http: / tools.iedb.org) can be used. According to the present invention, amino acid modifications for which immunogenicity is to be detected may result from mutations in cellular nucleic acids. Such mutations may be identified using known sequencing techniques. In one embodiment, cancer-specific somatic mutations or sequence differences can be detected in the exome of a tumor sample, such as the entire exome. Accordingly, the present invention may comprise identifying cancer mutation signatures in the exome, preferably the entire exome, of one or more cancer cells. In one embodiment, the step of identifying cancer-specific somatic mutations in a tumor sample from a cancer patient comprises identifying an exome-wide cancer mutation profile. In one embodiment, cancer-specific somatic mutations or sequence differences can be detected in the transcriptome of a tumor sample, e.g., the entire transcriptome. Accordingly, the present invention may comprise identifying a cancer mutation signature in the transcriptome of one or more cancer cells, e.g., the entire transcriptome. In one embodiment, the step of identifying cancer-specific somatic mutations in a tumor sample of a cancer patient comprises identifying a transcriptome-wide cancer mutation profile. In one embodiment, the step of identifying a cancer-specific somatic mutation or the step of identifying a sequence difference comprises single cell sequencing of one or more, for example, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, or more, cancer cells. Accordingly, the invention may comprise identifying a cancer mutation signature of said one or more cancer cells. In one embodiment, the cancer cells are circulating tumor cells. Cancer cells, such as circulating tumor cells, may be isolated prior to single cell sequencing. In one embodiment, the step of identifying a cancer-specific somatic mutation or the step of identifying a sequence difference can be performed using next generation sequencing (NGS). In one embodiment, the step of identifying a cancer-specific somatic mutation or the step of identifying a sequence difference may comprise sequencing genomic DNA and / or RNA of the tumor sample. To identify cancer-specific somatic mutations or sequence differences, it is desirable to compare sequence information obtained from a tumor sample with reference sequence information, such as sequence information obtained by sequencing nucleic acids, such as DNA or RNA, from normal non-cancerous cells, such as germ cells, obtained from the patient or another individual. In one embodiment, the normal genomic germ cell DNA may be obtained from peripheral blood mononuclear cells (PBMCs). The term "genome" refers to the total amount of genetic information within the chromosomes of an organism or cell. The term "exome" refers to the portion of an organism's genome formed by exons, the coding portions of expressed genes. The exome provides the genetic blueprints used to synthesize proteins and other functional gene products. As the most functionally relevant portion of the genome, it is likely to contribute most significantly to an organism's phenotype. The human exome is estimated to account for 1.5% of the total genome. The term "transcriptome" refers to the set of all RNA molecules, including mRNA, rRNA, tRNA, and other non-coding RNA, produced by a cell or a population of cells. As used herein, the transcriptome may refer to the set of all RNA molecules produced by a cell, a population of cells, e.g., a cancer cell population, or all cells of a given individual at a particular point in time. The term "nucleic acid" in the present invention may refer to deoxyribonucleic acid (DNA) or ribonucleic acid (RNA). Nucleic acids include genomic DNA, cDNA, mRNA, recombinantly produced molecules, and chemically synthesized molecules according to the present invention. In the present invention, nucleic acids may exist as single-stranded or double-stranded molecules and as linear or covalently converted molecules. The term "mutation" refers to a change or difference (nucleotide substitution, addition, or deletion) in the nucleic acid sequence compared to a reference and / or a change in the amino acid sequence of the peptide (protein) encoded by such a change in the nucleic acid sequence. A "somatic mutation" can occur in any somatic cell other than the germ cells (sperm and egg) and is therefore not passed on to children. Such changes can (but do not always) cause cancer or other diseases. Preferably, the mutation is a nonsynonymous mutation. The term "nonsynonymous mutation" refers to a mutation that results in an amino acid change, such as an amino acid substitution, or preferably a nucleotide substitution, in the translation product. In the present invention, the term "mutation" may include point mutations, indels, fusions, chromothripsis, base corrections, and / or RNA edits. The term "cancer mutation signature" refers to the set of mutations present in cancer cells compared to non-cancerous reference cells. The above "reference" can be used to correlate and compare results obtained from the method of the present invention with tumor samples. Typically, the "reference" can be obtained based on one or more normal samples, such as samples obtained from individuals of the same species that are not afflicted with cancer. Any suitable sequencing method can be used for mutation detection, for example, Next Generation Sequencing (NGS). In the future, third-generation sequencing methods may replace NGS technologies to speed up the sequencing step. The term "next-generation sequencing" or "NGS" refers to all new high-throughput sequencing technologies, in contrast to the "conventional" sequencing methodology known as Sanger chemistry, which reads nucleic acid templates in a parallel, random manner along the entire genome by breaking the entire genome into small fragments. These NGS technologies (also known as massively parallel sequencing technologies) can deliver nucleic acid sequence information of the whole genome, exome, transcriptome (all transcribed sequences of the genome), or methylome (all methylated sequences of the genome) in a very short period of time, for example, within 1-2 weeks, 1-7 days, or even within 24 hours, essentially enabling a single-cell sequencing approach. Multiple NGS platforms available commercially or mentioned in the literature can be used. The term "RNA" relates to a molecule comprising at least one ribonucleotide residue and preferably consisting or consisting essentially of ribonucleotide residues. A "ribonucleotide" relates to a nucleotide having a hydroxyl group at the 2'-position of the β-D ribofuranosyl group. The term "RNA" includes double-stranded RNA, single-stranded RNA, isolated RNA such as partially or completely purified RNA, essentially pure RNA, synthetic RNA, and recombinantly produced RNA such as modified RNA which differs from naturally occurring RNA by the addition, deletion, substitution, and / or alteration of one or more nucleotides. Such alterations may include the addition of non-nucleotide material to the terminus(s) of the RNA or within the RNA, such as to one or more nucleotides of the RNA. The nucleotides within an RNA molecule may also include non-standard nucleotides, such as non-naturally occurring nucleotides or chemically synthesized nucleotides or deoxynucleotides. These altered RNAs may also be referred to as analogs or analogs of naturally occurring RNAs. The term "RNA" may include "mRNA." The term "mRNA" refers to a "messenger RNA" that is produced using a DNA template and encodes a peptide or polypeptide, and may be referred to as a "transcript." Typically, mRNA comprises a 5'-UTR, a protein-coding region, and a 3'-UTR. mRNA has only a limited half-life in cells and in vitro. mRNA can be produced by in vitro transcription from a DNA template. In vitro transcription methodologies are known to those skilled in the art. Terms such as "reduce" or "inhibit" refer to the ability to cause an overall reduction, for example, an overall reduction of at least 5%, at least 10%, at least 20%, at least 50%, or at least 75%. The term "inhibit" or similar expressions include complete or essentially complete inhibition, i.e., a reduction to zero or essentially zero. The terms "increase," "enhance," "promote," "or prolong" relate to an increase, augmentation, promotion or prolongation of, for example, at least about 10%, at least 20%, at least 30%, at least 40%, at least 50%, at least 80%, at least 100%, at least 200%, or at least 300%. These terms also relate to an increase, augmentation, promotion or prolongation from zero or an immeasurable or undetectable level to a level greater than zero or a measurable or detectable level. The term "vaccine" refers to a pharmaceutical preparation (pharmaceutical composition) or product that, when administered, induces an immune response, particularly a cellular immune response, that recognizes and attacks diseased cells or pathogens, such as cancer cells. Vaccines can be used to prevent or treat diseases. The term "personalized cancer vaccines" refers to a cancer vaccine tailored to the specific circumstances of each cancer patient, as needed. In one embodiment, a vaccine provided according to the present invention may comprise one or more peptides selected as immunogenic peptides according to the method of the present invention or a nucleic acid (e.g., RNA) encoding said peptides. The anticancer vaccine provided according to the present invention, when administered to a patient, provides one or more T cell epitopes suitable for stimulating, priming, and / or expanding T cells specific for the patient's tumor. The T cells may be directed against cells expressing the antigen from which the T cell epitope is derived. Accordingly, the vaccines described herein may be capable of inducing or promoting a cellular response, such as cytotoxic T cell activity, against a cancer disease characterized by the presentation of one or more tumor-associated neoantigens by class I MHC. Since the vaccine provided according to the present invention targets cancer-specific mutations, it may be specific to the tumor of the patient. The vaccine provided in the present invention relates to a vaccine which, when administered to a patient, provides one or more, for example, 2 or more, 5 or more, 10 or more, 15 or more, 20 or more, 25 or more, 30 or more and up to 60, 55 or less, 50 or less, 45 or less, 40 or less, 35 or less or 30 or less, T cell epitopes associated with a peptide selected as an immunogenic peptide by the method of the present invention. Such T cell epitopes are also referred to herein as "neo-epitopes." Presentation of these epitopes by cells of the patient, particularly antigen presenting cells, preferably results in T cells targeting the epitopes upon binding to MHC, and thus targeting tumors of the patient, preferably the primary tumor as well as tumor metastases, expressing the antigen from which the T cell epitopes are derived and presenting the same epitopes on the surface of cells of said tumor. The term "tumor" or "neoplastic disease" preferably refers to an abnormal growth of cells (also called neoplastic cells, neoplastic cells, or tumor cells) that forms a swelling or lesion. A "tumor cell" refers to an abnormal cell that grows rapidly and uncontrolled and continues to grow after the stimulus that initiated the new growth has ceased. A tumor forms a distinct mass of tissue that partially or completely lacks functional connections and structural organization with normal tissue and may be benign, premalignant, or malignant. Cancer is a group of diseases in which groups of cells exhibit uncontrolled growth (dividing beyond normal limits), invasion (invasion and destruction of adjacent tissues), and sometimes metastasis (spread to other parts of the body through the lymph or bloodstream). These three malignant characteristics of cancer distinguish them from benign tumors, which are self-limiting and do not invade or metastasize. Most cancers form tumors, but some, such as leukemia, do not. Malignancy, malignant neoplasm, and malignant tumor are basically synonymous with cancer. For the purposes of the present invention, the terms “cancer” and “cancer disease” may be used interchangeably with the terms “tumor” and “tumor disease”. The above "cancer" may be carcinoma, adenocarcinoma, blastoma, leukemia, seminoma, melanoma, teratocarcinoma, lymphoma, neuroblastoma, glioma, rectal cancer, endometrial cancer, kidney cancer, adrenal cancer, thyroid cancer, blood cancer, skin cancer, brain cancer, cervical cancer, bowel cancer, liver cancer, colon cancer, stomach cancer, intestinal cancer, head and neck cancer, gastrointestinal cancer, lymph node cancer, esophageal cancer, colon cancer, pancreatic cancer, ear, nose and throat (ENT) cancer, breast cancer, prostate cancer, uterine cancer, ovarian cancer, lung cancer, etc. The above cancer may be primary cancer or metastatic cancer. “Treatment” may mean reducing the size or number of tumors in a subject; stopping or slowing the progression of a disease in a subject; inhibiting or slowing the development of a new disease in a subject; reducing the frequency or severity of symptoms and / or recurrences in a subject who is currently or has previously had a disease; and / or prolonging the life of a subject. "Immunotherapy" relates to a therapeutic method involving the activation of a specific immune response. In the context of the present invention, the terms "protection," "prevention," "prevention," "preventing," "protective," and the like refer to the prevention or treatment of the onset and / or spread of a disease in a subject, or both, and in particular, to minimizing the chance of the disease developing or delaying the progression of the disease in a subject. For example, an individual at risk for developing cancer is a candidate for cancer prophylaxis. Prophylactic administration of immunotherapy, such as the vaccine of the present invention, preferably protects the recipient from the development of a disease. Therapeutic administration of immunotherapy, such as the vaccine of the present invention, may result in inhibition of the progression / growth of the disease. This includes slowing the progression / growth of the disease, particularly abrogation of the progression of the disease, for example, leading to elimination of the disease. Immunotherapy can be performed using a variety of techniques, wherein the formulations provided herein function to remove diseased cells from a patient. This removal may occur as a result of inducing or enhancing an immune response in the patient specific for an antigen or cells expressing the antigen. The formulations and compositions provided in the present invention may be used alone or in combination with treatments such as surgery, radiotherapy, chemotherapy, and / or bone marrow transplantation (autologous, syngeneic, allogeneic, or unrelated). The terms "immunization" or "vaccination" describe the process of treating a patient with the aim of inducing an immune response for therapeutic or prophylactic reasons. The term "in vivo" relates to the situation inside the subject's body. The terms "subject," "individual," or "patient" are used interchangeably and relate to a vertebrate, such as a mammal. For example, in the context of the present invention, mammals encompass humans, non-human primates, domesticated animals such as dogs, cats, sheep, cows, goats, pigs, horses, etc., laboratory animals such as mice, rats, rabbits, guinea pigs, etc., as well as zoo-captured animals. The term "animal" as used herein may also include humans. The term "subject" includes a patient, i.e., an animal, preferably a human suffering from a disease, preferably a disease described herein, or a biological sample (e.g., cancer cells, blood) isolated therefrom. The vaccine composition provided herein may further comprise an adjuvant. The term "adjuvant" relates to a compound that prolongs, enhances, or accelerates an immune response. The composition of the present invention preferably exerts its effect without the addition of an adjuvant. However, the composition of the present invention may also contain known adjuvants. Adjuvants include a heterogeneous group of compounds, such as oil emulsions (e.g., Freund's adjuvant), mineral compounds (e.g., alum), bacterial products (e.g., Bordetella pertussis toxin), liposomes, and immunostimulatory complexes. An example of an adjuvant is monophosphoryl-lipid-A (MPL SmithKline Beecham). Examples include saponins such as QS21 (SmithKline Beecham), DQS21 (SmithKline Beecham; WO 96 / 33739), QS7, QS17, QS18, and QS-L1, incomplete Freund's adjuvant, complete Freund's adjuvant, vitamin E, montanide, alum, CpG oligonucleotides, and various water-in-oil emulsions prepared from biologically degradable oils such as squalene and / or tocopherol. The above vaccine composition may additionally include a pharmaceutically acceptable substance that stimulates the patient's immune response. For example, the substance may be a cytokine. Examples of such cytokines include interleukin-12 (IL-12), GM-CSF, and IL-18, which enhance the protective activity of the vaccine. The above vaccine composition may further comprise a compound that enhances the immune response. The compound may comprise a co-stimulatory molecule provided in the form of a nucleic acid or protein, such as B7-1 and B7-2 (CD80 and CD86, respectively). The vaccine compositions provided herein can be administered by conventional routes, including injection or infusion. Administration can be performed, for example, orally, intravenously, intraperitoneally, intramuscularly, subcutaneously, or transdermally. In one specific embodiment, administration can be performed intranodally, such as by injection into a lymph node. The above vaccine composition may be administered in a pharmaceutically effective amount. A "pharmaceutically effective amount" refers to an amount capable of producing the desired response or effect, either alone or in combination with additional doses. The pharmaceutically effective amount may vary depending on the condition being treated, the severity of the disease, individual patient variables such as age, physiological condition, size, and weight, the duration of treatment, the type of concomitant treatment (if any), the specific route of administration, and similar factors. The above vaccine composition may be sterilized and may contain an effective amount of a therapeutically active substance to produce the desired response or desired effect. The term "pharmaceutically acceptable" means a non-toxic substance that does not interact with the action of the active ingredient of the pharmaceutical composition, and may contain, for example, salts, buffering substances, preservatives, carriers, adjuvants, auxiliary immune-enhancing substances such as CpG oligonucleotides, cytokines, chemokines, saponins, GM-CSF and / or RNA, and, where appropriate, other therapeutically active compounds. "Carrier" refers to a natural or synthetic organic or inorganic ingredient that is combined with an active ingredient to facilitate, enhance, or enable its application. The term "pharmaceutically acceptable carrier" in the present invention includes one or more suitable solid or liquid fillers, diluents, or encapsulating materials suitable for administration to a patient. The above vaccine composition may contain suitable buffers such as acetate salts, citrate salts, borate phosphate salts. The above vaccine composition may contain suitable preservatives such as benzalkonium chloride, chlorobutanol, parabens and thimerosal. The above vaccine composition is usually provided in unit dosage form and can be prepared by methods known per se. The vaccine composition of the present invention may be in the form of an injection, capsule, tablet, lozenge, solution, suspension, syrup, elixir, or, for example, an emulsion. Compositions suitable for parenteral administration typically include sterile aqueous or non-aqueous formulations of the active compound that are isotonic with the recipient's blood. Suitable carriers and solvents include Ringer's solution and isotonic sodium chloride solution. Additionally, sterile, fixed oils can be used as a solution or suspension medium. When the above-mentioned features of the invention are applied, efficient processing of HLA-type I and mutation variants is possible through NGS data processing for patients. In the present invention, a method is proposed for selecting neoantigen candidates that can be applied to personalized anticancer vaccines by conducting database learning using AI technology, selecting a group of neoantigen candidates through a binding affinity model and an immunogenicity model, and applying and verifying a new neoantigen prioritization process. Figure 1 shows the distribution of binding score (BA) and immunogenicity score (IMM). Figure 2a schematically shows the immunogenicity score measurement process according to Example 4. Figures 2b to 2d show the results of immunogenicity confirmation through INFg ELISPOT by HLA-I type in Example 4 (2b: A*02:01, 2c: A*24:02, 2d: A*11:01). Figure 3 schematically shows the test process of Example 5.1. Figure 4 is a graph showing the splenocyte IFNg ELISPOT results obtained in Example 5.1. Figure 5 schematically shows the test process of Example 5.2. Figure 6 is a graph showing the splenocyte IFNg ELISPOT results obtained in Example 5.2. Figure 7 is a graph showing the anticancer effect (in vivo) of the new antigen confirmed in Example 6. The present invention will be described in more detail below with reference to the following examples. However, these examples are provided solely to illustrate the present invention, and the scope of the present invention is not limited by these examples. Example 1. Building a Neoantigen Prediction Model (Model Learning) (Neoantigen Prediction Model I) An artificial intelligence model was prepared to predict the binding affinity and immunogenicity of candidate peptides of new antigens through learning using the amino acid sequence information of HLA IA, B, C among HLA (Human Leukocyte Antigen) type I and peptides in the IEDB (Immune Epitope Database and Analysis Resource) (https: / www.iedb.org). Specifically, the binding affinity and immunogenicity prediction model is an ensemble model consisting of 150 different models trained by combining different model structures, learning data divisions, whether or not transfer learning is included, learning optimization methods, and learning stop points. One of the Gated Recurrent Unit (GRU) and Transformer was selected as the model structure. 5-Fold Cross Validation was applied to the learning data division. For the avidity prediction model, transfer learning used a classifier trained on qualitative data. For the immunogenicity prediction model, the avidity prediction model with the lowest prediction error on the validation data was used. The learning optimization method was either AdamW or Stochastic Weights Averaging (SWA). The five learning stop points were selected from the five points with the lowest prediction errors on the validation data throughout the entire learning process. Below, the process of preparing a new antigen prediction model is described in more detail: 1.1. Selecting Input Data for the New Antigen Prediction Model Among the IEDB data, data necessary for learning HLA type I were selected and converted so that they can be used for learning a neoantigen prediction model, and then applied to the Binding Affinity model and the Immunogenicity model, respectively. For example, the data selection may apply one or more tasks, such as deleting data that does not correspond to HLA-A, HLA-B, or HLA-C, deleting data with mutations in HLA molecules, deleting peptides whose length is 7 or less or 15 or more, and removing amino acid sequences that include sequences other than 20 human-derived amino acids. For example, the data transformation may be the application of one or more tasks, such as normalizing a binding affinity measurement value (e.g., IC50) with a large range of data values ​​to the range of 0 to 1, or converting immunogenicity score data provided as [number of experiments, number of reactions] instead of quantitative measurements to the mean value of a beta distribution (L. Guangyuan, et al. "DeepImmuno: deep learning-empowered prediction and generation of immunogenic peptides for T-cell immunity." Briefings in bioinformatics 22.6 (2021)). When learning the immunogenicity prediction model, among the various experimental results of the IEDB data, the test method related to immunogenicity, such as the amount of cytokines secreted, such as IFNg, IL-2, TNFa, Granzyme A, etc. (data corresponding to T cell-derived cytokines including INF-g, TNF-a, IL-2, IL-1b, etc. when PBMCs are treated with each peptide from the IEDB, T cell proliferation due to immunogenic activation, Chemokine including CCL4, CXCL9, Granzyme B, cytotoxicity of activated T cells, etc.) was selected and learning was performed. 1.2. Learning Method of the New Antigen Prediction Model Using the data selected in Example 1.1 above, the following was performed: For the binding affinity prediction model, one of two model structures (GRU (J. Chung et al,https: / arxiv.org / abs / 1412.3555, 2014), transformer (Vaswani, Ashish, et al. "Attention is all you need." Advances in neural information processing systems 30 (2017))) was selected, one of three learning methods was selected, one of the data splitting methods of 5-fold Cross Validation was selected, and the model was saved at the five learning points with the lowest prediction error for the validation data during learning, resulting in a total of 150 model training (2 X 3 X 5 X 5 = 150). The final binding affinity was predicted by calculating the mean and variance of the prediction values ​​of the 150 trained models (prediction value 1). The immunogenicity prediction model was trained by selecting one of two model structures (GRU, transformer), one of three learning methods (including training the model with the lowest prediction error on the validation data among the binding affinity prediction models), one of the data splitting methods of 5-fold Cross Validation, and saving the model at the five learning points with the lowest prediction error on the validation data during training, resulting in a total of 150 models (2 X 3 X 5 X 5 = 150). The final immunogenicity was predicted by calculating the mean and variance of the prediction values ​​of the 150 trained models (prediction value 2). In calculating the above prediction values ​​1 and 2, [(mean of 150 model predictions) - (variance of 150 model predictions)] was used as the final prediction value. This is because the smaller the deviation between the prediction values ​​of 150 models using different data and learning methods, the more stable the prediction can be considered. Example 2. New antigen prediction and selection steps A new antigen was selected from the new antigen prediction model constructed in Example 1 above, and the selection of the new antigen was carried out in the following steps: 1) Collection of NGS data of cancer patients from public databases; 2) Using the NGS data of the above cancer patient, determination of HLA type and cancer cell-derived peptides and calculation of TPM; 3) Calculation of prediction score using the new antigen prediction model of Example 1 above; 4) Calculation of priority of new antigens and selection of peptides for experiment using the formula of Example 3 below; 5) Binding strength and immunogenicity test using PBMCs with the same HLA type as determined in 2) above. The publicly available data used to validate the neoantigen prediction model were from four melanoma patients (HLA type: A*02:01; see “John Wiley & Sons A / S, Tissue Antigens 73, 95-170”) who achieved a complete response to pembrolizumab treatment (hereinafter referred to as “patient sample data” or “patient sample dataset”). [Hugo, Willy, et al. “Genomic and transcriptomic features of response to anti-PD-1 therapy in metastatic melanoma.” Cell 165.1 (2016): 35-44.] 2.1. Data preparation for neoantigen selection (whole exome sequencing of patient-derived samples, RNA-seq data processing) For the above patient sample dataset, data preprocessing, variant discovery, and callset refinement were performed by combining each code to derive information such as mutation sequences and HLA types. More specifically, 1) DNA mutation information was converted into peptide mutation information, and RNA abundance data and HLA type data were concatenated. Furthermore, a process was performed to confirm whether the peptides obtained by conversion were found in known protein sequences of Homo sapience in Uniprot's SwissProt. 2) Results were obtained from Mutect2 from Broad Institute and Strelka developed by Illumina, which are included in the GATK package, a program that searches for cancer-specific genetic mutations. Then, by processing the DNA mutation information, the protein-level missense mutation and in-frame deletion data were organized by each tool, and the peptides obtained in duplicate from Mutect2 and Strelka were unified (Koboldt, DC et al., VarScan 2: Somatic mutation and copy number alteration discovery in cancer by exome sequencing. Genome Research 22, 568-576 (2012); Cai, L., Yuan, W., Zhang, Z., He, L. & Chou, KC, In-depth comparison of somatic point mutation callers based on different tumor next-generation sequencing depth data. Scientific Reports 6, 9, https: / doi.org / 10.1038 / srep36540 (2016); Kim, S. et al., Strelka2: fast and accurate calling of germline and somatic variants. Nature Methods 15, 591-594, https: / / doi.org / 10.1038 / s41592-018-0051-x (2018); McKenna, A. et al., The Genome Analysis Toolkit: a MapReduce framework for analyzing next-generation DNA sequencing data. Genome Res. 20, 1297-1303, https: / doi.org / 10.1101 / gr.107524.110 (2010); Cibulskis, K., Lawrence, M.S., Carter, S.L., Sivachenko, A., Jaffe, D., Sougnez, C., & Getz, G. (2013). Sensitive detection of somatic point mutations in impure and heterogeneous cancer samples. Nature Biotechnology, 31(3), 213-219. doi:10.1038 / nbt.2514). 3) By calculating the hydrophobicity of the peptide, data for future neoantigen production was prepared. When neoantigen prediction and personalized anticancer vaccines are produced using mRNA, special consideration of hydrophobicity may not be necessary. However, when producing an anticancer vaccine using peptides or conducting immunogenicity tests, production may be impossible if the peptide has high hydrophobicity (e.g., high hydrophobic amino acid content). Therefore, calculating the hydrophobicity of the peptide may be necessary. In this example, the hydrophobicity was determined by calculating the Gravy (Grand average of hydropathy) score. 2.2. Selection of neoantigens using a neoantigen prediction model (binding affinity and immunogenicity) (confirmation of applicability to personalized anticancer vaccines) Using the data prepared in Example 2.1 above, the binding affinity and immunogenicity thresholds were set, and the neoantigen prediction model was applied to perform verification on the selected neoantigen: - The binding affinity (BA) test verifies the binding affinity of the predicted neoantigen by comparing it with the binding affinity of NY-ESO-1 (New York esophageal squamous cell carcinoma 1 peptide (157-165, SLLMWITQC; synthesized by Anygene) with a known IC50 value. - Immunogenicity test conducted using IFNγ ELISPOT immunogenicity test method Normal PBMCs of the HLA-A*02:01 type were rested for 24 hours, then pulsed with 50 μM candidate peptides and IL-2 (10 IU / ml) (72 hours, 37°C, 5% CO2). The PBMCs were then washed and transferred to an ELISPOT assay plate for re-stimulation with the same peptides (50 μM, 24 hours, 37°C, 5% CO2). The ELISPOT assay was performed according to Immunospot's ELISPOT protocol. Re-testing was performed with PBMCs from three different donors (to confirm reproducibility and statistical significance). The above HLA-A*02:01 used a sequence of a portion of the alpha chain, and the sequence used was as follows: SHSMRYFFTSVSRPGRGEPRFIAVGYVDDTQFVRFDSDAASQRMEPRAPWIEQEGPEYWDGETRKVKAHSQTHRVDLGTLRGYYNQSEAGSHTVQRMYGCDVGSDWRFLRGYHQYAYDGKDYIALKEDLRSWTAADMAAQTTKHKWEAAHVAEQLRAYLEGTCVEWLRRYLENGKETLQRT (SEQ ID NO: 1) Since this is a test using cells, there is a deviation in the immunogenicity test, so we established a method for selecting new antigens for personalized anticancer vaccine application by applying Min-Max Normalization and Weight Value (1 Lot Detection: x1, 2 Lot Detection: x2, 3 Lot Detection: x3) to rank new antigens. The results obtained by applying the information on 118,057 mutant peptides identified in patient-derived samples (4 melanoma patients (HLA type: A*02:01) who showed a complete response to pembrolizumab treatment) in Example 2.1 to the binding affinity (BA) prediction model and immunogenicity (IMM) prediction model of Example 1 are shown in Figure 1. In addition, among the results of Figure 1, the actual binding affinity and immunogenicity for the BA high group (G1; 0.15 ≤ IMM Score < 0.25 and 0.6 ≤ BA Score) and the IMM high group (G2: IMM Score ≥ 0.25 and 0.6 > BA Score ≥ 0.1) were measured using the method described above and are shown in Table 1: As shown in the results in Table 1 above, When the threshold is set to 0.15 ≤ IMM Score < 0.25 and 0.6 ≤ BA Score (G1), the prediction rate of positive new antigens with binding affinity is 100%. When the threshold was set to IMM Score ≥ 0.25 and 0.6 > BA Score ≥ 0.1 (G2), the positive neoantigen prediction rate was confirmed to be 23.6%. Example 2.3. Measurement of RNA expression level (TPM) as an additional variable As a variable for evaluating the neoantigens predicted above, the TPM (RNA expression level; transcripts per million) variable was additionally measured. TPM refers to the expression rate of the RNA sequence of the neoantigen candidate group, and can indicate the possibility that a mutation in the patient's cancer cells will actually be generated as a peptide. However, since the IEDB does not provide the TPM data, it can be used for evaluation (prioritization) of model prediction results rather than model learning. In order to develop an evaluation criterion utilizing the TPM, public NGS data of cancer patients (TESLA dataset, Wells, Daniel K., et al. "Key parameters of tumor epitope immunogenicity revealed through a consortium approach improve neoantigen prediction." Cell 183.3 (2020): 818-834.) were collected to select a group of cancer-derived neoantigen candidates (peptide candidates), and the prediction model was used to predict the predicted values ​​1 and 2 of the neoantigen candidates (see Example 1.2). The RNA-seq data included in the public NGS data of the above cancer patients were processed to calculate the expression rate (TPM) of the RNA sequences of the above neoantigen candidates. Example 3. Evaluation of Selected Neoantigens (Neoantigen Prioritization) For the prediction values ​​1 and 2 obtained in the previous Example 1.2, optimization of prediction model variables in the binding affinity model and the immunogenicity model was performed through multivariate linear regression analysis, and finally, TPM (RNA expression level), binding affinity prediction score (BA; prediction value 1 of the above Example 3), and immunogenicity prediction score (IMM; prediction value 2 of the above Example 3) were determined as variables, and the ranking of the new antigen was selected. The above TESLA dataset includes HLA types and NGS data of 8 patients, peptide sequences of 845 neoantigen candidates derived from the patients, and immunogenicity test results. Among the 845 candidates, 38 peptides were verified to be immunogenic. To compare the superiority of neoantigen prediction criteria, TPM was applied to the neoantigen candidate group and the weights of the binding affinity prediction score (BA) and immunogenicity prediction score (IMM) were varied, and 20 neoantigen candidates were selected per patient (160 in total). Table 2 below shows how many of the 38 neoantigen candidates with immunogenicity were included in this group (Threshold: TPM (transcripts per million) > 1): WeightDistributionW1: 1.0W2: 0.0W1: 0.8W2: 0.2W1: 0.5W2: 0.5W1: 0.2W2: 0.8W1: 0.0W2: 1.0TESLA1418262323 (In Table 2 above, W1 represents the weight of BA, and W2 represents the weight of IMM, respectively) As confirmed in Table 2 above, when the weights of the binding affinity prediction score (BA) and the immunogenicity prediction score (IMM) were made equal (1:1), the most immunogenic neoantigen could be found. Based on the above results, the formula for ranking new antigens was established as follows: Neoantigen prioritization scoring scheme: [min(TPM,1) * (BA + IMM) / 2] The number of new antigens detected before and after applying the above-mentioned Prioritization is shown in Table 3 below: Before PatientPositivePrioritization After Prioritization PilotRound#4111#8101#9212

[0310] 424MainRound#1937#2432#31329#12422#16412

[0535] 341122 As shown in Table 3, it can be confirmed that the number of selectable new antigens increases by applying Prioritization, thereby improving the prediction rate of new antigens. Example 4. Neoantigen Prediction Model II and Neoantigen Selection Referring to the methods of Examples 1 to 3, patients with melanoma who possessed three HLA types (A*02:01, A*24:02, A*11:01) were selected, and 20 types of neoantigens were predicted using a neoantigen prediction model. The immunogenicity of the predicted neoantigens was confirmed using IFNg ELISPOT. More specifically, the criteria for selecting the patient group data are as follows: - High Mutation Burden Tumor → Melanoma; - Complete WES (whole exosome sequencing; Mutation selection), RNA-seq (Expression level) Dataset (converted to peptide sequence and used); - ICI (Immune Checkpoint Inhibitor) treatment (anti-PD-1 therapy: Pembrolizumab, Nivolumab) → CR (Complete Response), PR (Partial Response), PD (Progressive Disease); - HLA type: A*02:01, A*24:02, A*11:01 ([HLA-A*02:01]: SHSMRYFFTSVSRPGRGEPRFIAVGYVDDTQFVRFDSDAASQRMEPRAPWIEQEGPEYWDGETRKVKAHSQTHRVDLGTLRGYYNQSEAGSHTVQRMYGCDVGSDWRFLRGYHQYAYDGKDYIALKEDLRSWTAADMAAQTTKHKWEAAHVAEQLRAYLEGTCVEWLRRYLENGKETLQRT (SEQ ID NO: 1), [HLA-A*11:01]: SHSMRYFYTSVSRPGRGEPRFIAVGYVDDTQFVRFDSDAASQRMEPRAPWIEQEGPEYWDQETRNVKAQSQTDRVDLGTLRGYYNQSEDGSHTIQIMYGCDVGPDGRFLRGYRQDAYDGKDYIALNEDLRSWTAADMAAQITKRKWEAAHAAEQQRAYLEGRCVEWLRRYLENGKETLQRT (SEQ ID NO: 2), [HLA-A*24:02]: SHSMRYFSTSVSRPGRGEPRFIAVGYVDDTQFVRFDSDAASQRMEPRAPWIEQEGPEYWDEETGKVKAHSQTDRENLRIALRYYNQSEAGSHTLQMMFGCDVGSDGRFLRGYHQYAYDGKDYIALKEDLRSWTAADMAAQITKRKWEAAHVAEQQRAYLEGTCVDGLRRYLENGKETLQRT (Sequence number 3); - Source: Hugo et al., 2016, Cell 165, 35-44, Riaz et al., 2017, Cell 171, 934-949 (incorporated herein by reference). The neoantigen prediction model applied to the above-mentioned selected patient group data is as follows: - Binding Affinity Score (BA) = (Average of Binding Affinity scores of the selected patient group) - (Variance of Binding Affinity Score of the selected patient group); - Immunogenicity Score (IMM) = (Mean Immunogenicity Score of the selected patient group) - (Variance of Immunogenicity Score of the selected patient group) The above BA score is obtained from the binding results of each patient's neoantigen and the patient's unique A*02:01, A*24:02, and A*11:01, and is a value obtained using the binding affinity prediction model that learned the prediction value 1 of Example 1.2 (see Example 2.1). The above IMM score is obtained from the immunogenicity prediction results of each patient's neoantigen and the patient's unique HLA type, and is a value obtained using the immunogenicity prediction model that learned the prediction value 2 (see Example 1.2) (see Example 2.1). Neoantigen Prioritization Score was calculated by applying TPM (RNA expression level; transcripts per million) score (TPM threshold: 1; see Example 3): Prioritization Score = min(TPM,1) * [(BA + IMM) / 2] Min(TPM,1): If TPM > 1, use it by replacing it with TPM = 1, and if TPM < 1, use the TPM Value as is.) The top 20 peptides were predicted as new antigens by ranking them using the obtained Prioritization Score (Tables 4 to 6). RankSequenceSEQID NOHLA typeScore1YLFNFVVNV4A*02:010.5578305192RLAEFIIGM5A*02:010.5515050943YLFNFVVNVG 6A*02:010.5012699734RLAEFIIGMM7A*02:010.4990018655LLTDIIPDGV8A*02:010.494219272 6FLALCADAV9A*02:010.4938086727YIAGILARI10A*02:010.4903580348RLAEFIIGMMG11A*02: 010.4773052989YLMVALASV12A*02:010.47199835710RLAEFIIGMMGT13A*02:010.45954869311 GLVECLPEI14A*02:010.45633326712AYLFNFVVNV15A*02:010.44813959413VLFYLSSAV16A*02 :010.44811717914RLFFCFMNV17A*02:010.44329168215CILAVLPSL18A*02:010.44089690816I MFNKGFLA19A*02:010.43877191617KTDPWFLLAV20A*02:010.43736530218SLIGVSFFALPV21A*0 2:010.43704042319SLWAESLVHV22A*02:010.43661999120FLLAVLGAPV23A*02:010.435148758 RankSequenceSEQ ID NOHLA typeScore1TFYWTFVNAF24A*24:020.4559100572VFYQFSKLIF25A*24:020.4034933673RMYSSWRL LNL26A*24:020.3712024824TWNVHQQFLW27A*24:020.3668472555RMYSSWRLL28A*24:020.36676 0256FYWTFVNAFSIF29A*24:020.3633757657QWILKTRAYFL30A*24:020.3389677038STWNVHQQF31A*24:020.3380139999QWILKTRAYF32A*24:020.33091987810STWNVHQQFLW33A*24:020.33000028 111FYWTFVNAF34A*24:020.32866869612STWNVHQQFLW35A*24:020.32760641713FYWTFVNAF36A*24:020.32446590514STWNVHQQFLW37A*24:020.32080623515FYWTFVNAF38A*24:020.3195821391 6STWNVHQQFLW39A*24:020.31682304317FYWTFVNAF40A*24:020.31602684718STWNVHQQFLW41A*24:020.31438974419FYWTFVNAF42A*24:020.31321298620LYHFSDRMYSSW43A*24:020.312662407 RankSequenceSEQID NOHLA typeScore1STACFHPVTK44A*11:010.4633504792TTFGIQVRYLK45A*11:010.4369374663KIGDFG LATMK46A*11:010.428431794KIGDFGLATEK47A*11:010.4279198685SSSGSINLLK48A*11:010.42 07427046AVAVDFQFSK49A*11:010.4084704247KTAPVEWRK50A*11:010.3929220758STSYLQHREK5 1A*11:010.3902456129VAVDFQFSK52A*11:010.38225777710SAFTIQEYFVK53A*11:010.3802814 711SVSNTTVGSK54A*11:010.37811472112RVQTGTIPK55A*11:010.37740502513SSGSINLLK56A*1 1:010.37332762614GTAAISVTGALK57A*11:010.37312020115STPNFSRLRRY58A*11:010.3727081 8116RSVLSSGYQK59A*11:010.37257794717VVSPVSLPRR60A*11:010.37044439418AAISVTGALK61 A*11:010.3696300319GSQPPSLGRK62A*11:010.36942562420TVYCRYHNY63A*11:010.368830969 To verify the predicted new antigen, PBMCs (Peripheral Blood Mononuclear Cells) from three adult donors with HLA type information were obtained and immunogenicity evaluation (IFNg ELISPOT) was performed using them (the three donors each had different HLA types, and the obtained PBMCs had A*02:01, A*24:02, or A*11:01, respectively). The test process is schematically shown in Fig. 2. For the ELISPOT results obtained above, normalization was performed using the following formula and analyzed: (Select immunogenicity Positive by setting Threshold >10; X: Number of spots (ELISPOT) in the test group; X(media): ELISPOT count of the group treated only with media (untreated group; negative control); X(CPI): ELISPOT count of the group treated with (protein antigens of CMV, Parainfluenza and influenza virus) (positive control); Normalization is performed to maintain consistency of results for each test. For each HLA type-matched patient, 8 (A*02:01), 7 (A*24:02), and 5 (A*11:01), Top 20 and Top 10 neoantigens were predicted using the neoantigen prediction model described in Examples 1 and 2, and immunogenicity of 20 predicted neoantigens was evaluated using INFg ELISPOT as described above using the test method corresponding to Examples 3 and 4, thereby confirming the number of neoantigens judged to be immunogenic among the 20 neoantigens of each patient. The results obtained above are shown in Table 7 and Figures 2b-2d: HLA-I type Number of patients Top 20 (%) Top 10 (%) A*02:01826.822.5 A*24:02720.714.3 A*11:01521.018.0 When immunogenicity was experimentally evaluated for 8 people with A*02:01, 7 people with A*24:02, and 5 people with A*11:01, the prediction rates of immunogenic new antigens were confirmed to be 26.8%, 20.7%, and 21.0%, respectively. Example 5. Ex vivo Immunogenicity 5.1. Human Cell Line (MCF7 cell line) Derived Xenograft Model The MCF7 breast cancer cell line is a human cancer cell line that expresses HLA-A*02:01. Referring to Example 4, WES (whole exosome sequencing; Mutation Selection) and NGS Sequencing of RNA-Seq (Express level) were performed on MCF7 cell line (ATCC, cat# HTB-22), and the top 20 (Top20) were selected using a neoantigen prediction model (BA, IMM, TPM) to select 20 neoantigens (mRNA), which were made into mRNA-lipoplex anticancer vaccines (see Example A(1) of KR 10-2022-0149461 A, KR 10-2022-0149460 A) and administered (60ug x 3 times QW, tail-vein IV injection) to A02:01 Transgenic mice (C57BL / 6 mice made to express HLA-A2.1; Taconic biosciences, cat# 8906-F) to determine whether they were immunogenic or not. It was confirmed through ELISPOT. The immunogenicity assay was confirmed as follows: 96-well ELISPOT plates coated with anti-IFNγ antibodies were prepared. After washing the plates once with PBS, 1.0 × 10^6 splenocytes collected from the immunized mice were loaded per well. For each immunized group, cells were plated in 3 wells and restimulated with 10 μM peptide or medium alone in a total volume of 200 μl. The plates were then incubated at 37°C for 18 hours. After washing three times with PBST, each well was treated with biotin-labeled mouse IFNγ detection antibody, and the plates were incubated for 2 hours at room temperature. The cells were washed again, and AP-conjugated streptavidin was added to each well and incubated for 30 minutes at room temperature. The AP reaction was developed using a substrate reagent set (Immunospot). The number of spot forming units (SFUs) per well was automatically counted using an ELISPOT reader (CTL ELISPOT reader). The test process, including the above vaccine administration process, is schematically shown in Figure 3. The 20 selected new antigens are shown in Table 8: RankNameNeoantigenSEQ ID NO1MCF7-1LLFDMTTHSV642MCF7-2FQSFFSLIV653MCF7-3FLGSGVHGGV664MCF7-4YLFFCGHMFHA675MCF7-5ITFEHILAL686MCF 7-6NLANEITYI697MCF7-7VLCEPVHGL708MCF7-8MLLESLRHAV719MCF7-9SLFGLGFEEHV7210MCF7-10HLLTDELGYV7311MCF7-11 KLSEASMIV7412MCF7-12ALLDMRSLL7513MCF7-13GLWEALTPCNV7614MCF7-14SLARLHLTV7715MCF7-15TLLPLPLLL7816MCF7- 16HLFWFADLNYRL7917MCF7-17STMSFLRSLYL8018MCF7-18TLNHEIYEA8119MCF7-19ILWGSISFNLAV8220MCF7-20MVCGAAFFI83 The splenocyte IFNg ELISPOT results obtained above are shown in Fig. 4. In Fig. 4, the control is the result of the untreated group that was not treated with the neoantigen peptide, and 'MCF7 NEO' is the result of the group administered the mRNA-lipoplex anticancer vaccine loaded with the 20 selected neoantigens. The threshold was set to be three times or more the average of the ELISPOT of the untreated group (control), and an ELISPOT result above the threshold was judged to have anticancer efficacy. As a result, among the 20 neoantigens tested, the neoantigen expected to have anticancer efficacy was evaluated to be 25% (marked with * in Fig. 4; 5 / 20). 5.2.B16F10 (Melanoma) AI Model Predicted Neoantigens Referring to Example 4, NGS of B16F10 (Mouse Melanoma; ATCC, cat# CRL-6475) cell line was performed (converted into amino acid sequence), and Top 50 and Top 20 neoantigens were predicted through neoantigen prediction models (BA, IMM, TPM). The predicted neoantigens were made into mRNA-lipoplex anticancer vaccines and administered to C57BL / 6 mice (60ug x 2 times QW IV injection), and the immunogenicity was confirmed through splenocyte IFNg ELISPOT (see Fig. 5). As a positive control, a mouse model was used that was administered mTRP2 (Tyrosinase-related protein 2) (Melanoma Tumor-associated Antigen; SVYDFFVWL (SEQ ID NO: 134)) in the same manner as described above, and the immunogenicity of the predicted neoantigens was determined by setting the threshold. The Top 50 neoantigen sequences used in this example are listed in Table 9: B16F10 Top50 SequenceRankB16F10 Neo SequenceSEQ ID NO1MSLPFSFL842ANFIFRQL853STYKLTNL864NMYGFRNV875VIYSHLQV886SGFRYNVL897VVMEYENL908IV HYFIAL919VSFAPLVQL9210ATFYFHHPV9311VHYLRLSL9412IGIFFQMRV9513VAHVAAFL9614VFLRVRAL971 5TIFMMLRL9816KGFLYSAL9917SSVLFKGL10018MQFYLRHL10119ARMMFSGL10220KSLYYASCL10321LVFVV GRL10422SIGVYIPL10523TVYRCGHL10624SLFTLHQSV10725FTFVYPTI10826CMLRYQTCL10927VCFDYTFA L11028LNVYYLPL11129VPYAVFPPL11230VSFANAVV11331VVPSFSAEI11432CNFCYQRNQL11533QMYPAFKR V11634ILVSTYRPL11735SFLAGFRPL11836AIYHHASRAI11937IAMGFPVERL12038VLILHAAAL12139SLVAP FGSV12240MYRYFVTL12341VMMDNFDYL12442VTFQAFIDV12543RAFAAVRL12644KNFEFNKQRSIV12745LMY VYEMLGAGL12846CMLRYQTCLV12947AIALHLHL13048TSYSTAAPL13149SMPHYSSPAI13250TIPFYDNPI133 The results obtained above are shown in Fig. 6. As shown in Fig. 6, immunogenicity was confirmed for 45% (9 / 20) of the top 20 predicted neoantigens (see the upper part of Fig. 6). Among the 12 highly immunogenic neoantigens among the top 50 neoantigens, 9 neoantigens were distributed within the top 20, confirming the high possibility of including target neoantigens when searching for neoantigens based on the top 20 (see Table 10). RankPositiveTop 209 (45%)Top21~Top 503 (10%)Total12 (24%) (Immunogenicity Threshold: > 3x Media ELISPOT Avg.) Example 6: In vivo Efficacy: Predicted Neoantigen Anticancer Efficacy 5x10 B16F10 tumor cells 4 The dog was transplanted into C57BL / 6 mice using a subcutaneous transplantation method, and 50 mm 3 Group separation was performed based on tumor size. The test group consisted of a group administered mRNA-lipoplex loaded with the top 10 of the 12 B16F10 predicted neoantigens secured in Example 5.2, a positive control administered mRNA-lipoplex loaded with mTRP2 (Tyrosinase-related protein 2) (Melanoma Tumor-associated Antigen) (SEQ ID NO: 134), and a negative control administered only lipoplex without neoantigen-loaded mRNA. mRNA-lipoplex was administered 60 ug at a time for 3 times every 5 days, and the tumor size was recorded for 3 weeks. The above-obtained results (tumor size) are shown in Figure 7 (Neoantigen Cancer Vaccine: 10 neoantigen administration group, Positive control: mTRP2 administration group, Control: lipoplex only administration (negative control)). As shown in Figure 7, a 63% tumor reduction was confirmed in the group administered mRNA-lipoplex loaded with 10 B16F10 predicted neoantigens. From the above description, those skilled in the art will understand that the present invention can be implemented in other specific forms without altering its technical spirit or essential characteristics. In this regard, it should be understood that the embodiments described above are illustrative in all respects and not restrictive. The scope of the present invention should be interpreted as encompassing all changes or modifications derived from the meaning and scope of the following claims and their equivalent concepts, rather than the detailed description above.

Claims

1. A method for selecting a patient-derived tumor-specific immunogenic peptide, comprising the following steps (1) to (7): (1) A step of obtaining NGS sequence information from patient-derived tumor cells or tumor tissues and normal cells or normal tissues; (2) A step of obtaining tumor-specific peptide sequence information having a length of 7 amino acids or more and having at least one amino acid mutation in the tumor cell or tumor tissue NGS sequence by comparing the tumor cell or tumor tissue NGS sequence with the normal cell or normal tissue NGS sequence; (3) A step of measuring the expression level (TPM; Transcripts per Million) of a tumor-specific peptide or its coding gene in tumor cells or tumor tissue; (4) A step of applying the above tumor-specific peptide sequence information to a binding affinity prediction model to determine a binding affinity score (BA) for an HLA molecule; (5) A step of applying the above tumor-specific peptide sequence information to an immunogenicity prediction model to determine an immunogenicity score (IMM); (6) Priority determination step to determine the ranking of tumor-associated peptides based on the values ​​calculated by the following formula: Neoantigen prioritization score = min(TPM,1) * (W1*BA + W2*IMM) / (W1+W2) [0 <W1<1, 0<W2<1 범위서 설정]; 및 (7) A step of selecting two or more tumor-specific immunogenic peptides based on the priority of the tumor-specific peptides determined above.

2. A method for selecting a patient-derived tumor-specific immunogenic peptide, further comprising the following steps (a) to (d) in paragraph 1: (a) a step of obtaining peptide sequence information derived from tumor cells or tumor tissues and peptide sequence information derived from normal cells or normal tissues from a peptide database; (b) a step of obtaining class I HLA peptide sequence information from the above biological sequence database; (c) a step of training the data obtained in steps (a) and (b) to a model capable of predicting a binding affinity score (BA) for HLA to build a binding affinity prediction model and obtain a binding score; and (d) A step of constructing an immunogenicity prediction model by training a model capable of predicting an immunogenicity score (IMM) using the data obtained in steps (a) and (b) and obtaining an immunogenicity score.

3. In the second paragraph, the peptide database used to build the prediction model of step (a) is characterized in that it undergoes one or more selection processes among the following processes i) to iv): i) Deletion of data that does not correspond to HLA-A, HLA-B, or HLA-C; ii) Deletion of data with mutations in HLA molecules; iii) Deletion of data with a peptide length of 7 or less or 15 or more, and iv) Removal of amino acid residues other than the 20 human-derived amino acids in the amino acid sequence.

4. A method for selecting a patient-derived tumor-specific immunogenic peptide, wherein the HLA comprises different HLA types or different HLA alleles in the first to third clauses.

5. In any one of paragraphs 1 to 3, the learning in steps (c) and (d) is performed by combining one or more artificial intelligence models and one or more learning methods. Method for selecting patient-derived tumor-specific immunogenic peptides.

6. A method for selecting a patient-derived tumor-specific immunogenic peptide, wherein the binding score is determined by an IC50 value or Kd value related to binding to HLA, in any one of paragraphs 1 to 3.

7. A method for selecting a patient-derived tumor-specific immunogenic peptide, wherein the immunogenicity score is determined by one or more pieces of information selected from the group consisting of information related to the amount of T cell-derived cytokine secretion, T cell proliferation due to immunogenic activation, the amount of chemokine secretion, and cytotoxicity of activated T cells when the candidate peptide is treated on an immune cell or a sample containing the same.

8. A method for selecting a patient-derived tumor-specific immunogenic peptide, wherein the sample containing the immune cells in paragraph 7 is selected from the group consisting of blood, leukocytes, and peripheral blood mononuclear cells (PBMC), and the information is two or more selected from the group consisting of IFN-g, IL-2, TNF-a, CCL4, CXCL9, and granzyme B.

9. A method for selecting a patient-derived tumor-specific immunogenic peptide, wherein the expression level of the candidate peptide or its coding gene in the patient-derived sample of step (3) is obtained through RNA sequencing from patient sample data in any one of paragraphs 1 to 3.

10. A method for selecting a patient-derived tumor-specific immunogenic peptide according to any one of claims 1 to 3, wherein the patient-derived tumor-specific immunogenic peptide is for use in the production of a personalized anticancer vaccine for the patient.

11. A tumor-specific immunogenic peptide selected by any one of the methods in paragraphs 1 to 3 or a nucleic acid molecule encoding the peptide; and pharmaceutically acceptable carrier An anticancer vaccine composition comprising:

12. A tumor-specific immunogenic peptide selected by any one of the methods in paragraphs 1 to 3 or a nucleic acid molecule encoding the peptide; and pharmaceutically acceptable carrier A method for producing an anticancer vaccine, comprising the step of mixing.

Images