Tumor neoantigen prediction method and tumor neoantigen prediction system

Abstract

The invention relates to a tumor neoantigen prediction method and a tumor neoantigen prediction system. The method comprises the following steps: acquiring a plurality of amino acid sequences in a gene of a person to be tested as a plurality of test peptides to be compared with a plurality of human protein sequences in a protein sequence database so as to find out a plurality of similar peptides matched with the human protein sequences; filtering similar peptides from the test peptides, and respectively inputting the filtered test peptides into a plurality of trained human leukocyte antigen models to obtain a plurality of sequencing results of the test peptides; calculating the weighted sum of the sequencing of each test peptide in the sequencing result as the score of the test peptide; and selecting at least one of the test peptides as a neoantigen suitable for the subject according to the fraction.

Description

Technical Field

[0001] The present disclosure relates to a prediction method and system, and in particular to a tumor neoantigen prediction method and a tumor neoantigen prediction system. Background Art

[0002] Neoepitopes and pathogen peptides are fundamental players in the complex interplay between the immune system and disease. Neoepitopes generated by amino acid changes have the ability to be recognized as abnormal by the immune system, thereby triggering a targeted immune response against cancer. These neoepitopes are derived from tumor-specific antigens (TSAs), which are different from tumor-associated antigens (TAAs) produced by abnormal expression within cancer cells. In contrast, pathogen peptides play a key role in the immune response against infectious agents. When invading pathogens replicate in the host, they produce a variety of proteins, some of which are processed into peptides and presented to T cells by major histocompatibility complex (MHC) molecules, which are important effectors of the adaptive immune system.

[0003] When the T-cell receptor (TCR) binds to the antigen-MHC complex, it activates the T cell and triggers its proliferation and differentiation, which leads to the formation of effector T cells that can directly eliminate antigen-presenting cells and memory T cells that can remember the antigen so that they can respond faster and stronger when exposed again. The diversity and specificity of the TCR are attributed to the unique combination of V, D, and J gene segments in the TCR gene. This V, D, J recombination during T cell development produces a variety of TCR alpha and beta chains, allowing the immune system to effectively recognize and respond to a variety of pathogens and abnormal cells.

[0004] Therefore, how to find new antigens suitable for fighting cancer cell gene mutations from the amino acid sequences in human genes is a major issue for those skilled in the art. Summary of the invention

[0005] In view of this, the present disclosure proposes a tumor neoantigen prediction method and a tumor neoantigen prediction system, which can find out neoantigens suitable for combating cancer cell gene mutations.

[0006] The disclosed embodiment provides a method for predicting tumor neoantigens, which includes the following steps: obtaining multiple amino acid sequences in the genes of a subject as multiple test peptides and comparing them with multiple human protein sequences in a protein sequence database to find multiple similar peptides matching the human protein sequences; filtering out similar peptides from the test peptides, and inputting the filtered test peptides into multiple trained human leukocyte antigen (HLA) models respectively to obtain multiple ranking results of the test peptides; calculating the weighted sum of the rankings of each test peptide in the ranking results as the score of the test peptide; and selecting at least one of the test peptides as a neoantigen suitable for the subject according to the score.

[0007] The disclosed embodiment provides a tumor neoantigen prediction system, which includes a data acquisition device, a storage device and a processing device. The data acquisition device is used to obtain multiple amino acid sequences in the genes of the subject to be tested. The storage device is used to store a protein sequence database that records multiple human protein sequences. The processing device is coupled to the data acquisition device and the storage device, and is configured to use the amino acid sequences obtained by the data acquisition device as multiple test peptides to compare with human protein sequences in the protein sequence database to find multiple similar peptides that match the human protein sequence, filter out similar peptides from the test peptides, and input the filtered test peptides into multiple trained human leukocyte antigen models respectively to obtain multiple ranking results of the test peptides, calculate the weighted sum of the rankings of each test peptide in the ranking results as the score of the test peptide, and select at least one test peptide as a neoantigen suitable for the subject to be tested based on the score.

[0008] Based on the above, the tumor neoantigen prediction method and tumor neoantigen prediction system of the embodiments of the present disclosure, by performing self-similarity and pathogen similarity analysis on the amino acid sequences in the genes of the subject, and combining HLA model and driver mutation analysis, can find out the neoantigens suitable for combating the gene mutations of the cancer cells of the subject, which can promote the progress of T cell immunotherapy and personalized cancer treatment. BRIEF DESCRIPTION OF THE DRAWINGS

[0009] Figure 1 is a schematic diagram of building a human biological database according to an embodiment of the present disclosure;

[0010] Figure 2 is a schematic diagram of tumor neoantigen prediction according to an embodiment of the present disclosure;

[0011] Figure 3 is a block diagram of a tumor neoantigen prediction system according to an embodiment of the present disclosure;

[0012] Figure 4is a schematic diagram of a method for predicting tumor neoantigens according to an embodiment of the present disclosure;

[0013] Figure 5 It is a schematic diagram of an amino acid sequence comparison / alignment method according to an embodiment of the present disclosure.

[0014] Explanation of symbols

[0015] 12: Literature

[0016] 14: Pathogen or new epitope database

[0017] 16: Human Biological Database

[0018] 161: Antigen

[0019] 162: MHC

[0020] 163: Detection

[0021] 164:TCR

[0022] 165: PDB

[0023] 166: Mutation

[0024] 167: Metadata

[0025] 20, 52: Test peptide

[0026] 22, 54: Human protein sequences

[0027] 24: Epitope prediction

[0028] 241: HLA-I Model

[0029] 242: HLA-II Model

[0030] 243: Self-similarity analysis

[0031] 244: Pathogen Similarity Analysis

[0032] 26: Driver mutation analysis

[0033] 28: Peptide Screening

[0034] 32: Data acquisition device

[0035] 34: Storage device

[0036] 36: Processing device

[0037] 56: Matched test peptide

[0038] 58: Unmatched test peptide

[0039] 500: Sequence alignment

[0040] S402~S408: Steps DETAILED DESCRIPTION

[0041] Reference will now be made in detail to exemplary embodiments of the present disclosure, examples of which are illustrated in the accompanying drawings. Whenever possible, the same reference numerals are used in the drawings and the description to refer to the same or like parts.

[0042] In order to advance T cell immunotherapy research and enhance the understanding of antigen-pathology relationships, the disclosed embodiments integrate pathogen peptides and neoepitope data with T cell detection information in a centralized human biological database to provide a full range of neoepitope data and corresponding T cell detection information, which is helpful for the research of T cell immunotherapy and the understanding of antigen-pathology relationships, and provides tumor neoantigen prediction, visualization and other functions to support pathogen research. Users can use sequence alignment to search for sequences targeting human amino acids (self-similarity) or known pathogen protein sequences (pathogen similarity) using information verified by T cell detection.

[0043] The disclosed embodiments organize data from published literature and existing databases, and integrate information such as antigens, major histocompatibility complex (MHC) types, T cell detection, T cell receptor (T-cell receptor, TCR)-peptide-MHC (pMHC) structure data and mutation types related to cancer-related entries in the database, thereby providing a unified and easy-to-use platform for researchers in the field of T cell immunotherapy.

[0044] In detail, the database uses, for example, a semi-automated method to ensure data reliability and convert outdated MHC naming conventions to modern equivalent naming conventions, thereby improving data accuracy and consistency. The database provides powerful search and visualization capabilities, allowing researchers to explore and analyze human TCRβ chain sequences associated with specific tissues, clinical conditions or cell types.

[0045] In addition to basic data integration and visualization capabilities, the database provides a variety of analytical tools, including string search methods (pathogen similarity and self-similarity analysis), classification to simplify analysis and species categories by providing high-level / fuzzy groupings, and assigning scores to each data entry based on the completeness of the fields.

[0046] In addition, the database provides motif / peptide analysis and integrates prediction tools to evaluate peptide-MHC (pMHC) binding affinity. The database aims to contribute to all aspects of pathogen peptide and new epitope research, and by providing a reliable and well-curated resource, it can support the new epitope research community and promote the advancement of T cell immunotherapy and personalized cancer treatment.

[0047] Figure 1 is a schematic diagram of constructing a human biological database according to an embodiment of the present disclosure. Figure 1 The disclosed embodiment uses a semi-automatic method to collect entries from published literature 12 and existing pathogen or new epitope databases 14 to construct a human biological database 16, ensuring that the human biological database 16 only contains data from published sources.

[0048] In detail, this embodiment uses various keywords to perform PubMed queries, including complementarity-determining region (CDR), MHC, new epitopes, and pathogens or cancer types, to search for literature 12, and filters the literature 12 obtained by the search to retain information containing experimentally verified pathogens or new antigen-related data. The initial filtering criteria involved the presence of peptides, MHC type, disease type, and detection information, and other criteria considered CDR3 sequence, CDR type, Protein Data Bank (PDB) information, mutation type, wild-type / mutant sequence, and mutant amino acid availability. Other information (such as classification) is filled in using a semi-automatic script, for example.

[0049] On the other hand, this embodiment obtains entries related to human species, infection, cancer, HLA, T cell detection, new epitopes, and structural information by querying the pathogen or new epitope database 14, which includes McPAS, IEDB, VDJdb, TBAdb, NEPdb, dbpepNeo, NeoDB, TSNAdb, etc., which are described as follows:

[0050] [Pathogen Database]

[0051] (1) IEDB: The IEDB is a comprehensive database focused on immune epitopes and their related assays (e.g., T-cell assays). In particular, a query was run on the IEDB homepage to retrieve linear peptide epitopes that have been tested using T-cell assays and showed positive results. The query excluded B cells and no MHC class restrictions were applied. The search was focused on human host organisms for infectious diseases.

[0052] All peptides collected through this query have T cell assay information and show a positive T cell response. These peptides have been experimentally evaluated to show the ability to elicit a T cell response in the context of infectious diseases. The T cell assays in the database can be divided into three major categories: 3D assays, bioactivity assays, and binding assays.

[0053] In the category of binding assays, for some entries, numerical values ​​are provided to indicate T cell recognition and reactivity to neoantigenic determinants. However, data misrecording may occur and result in inaccurate values.

[0054] To solve this problem, this embodiment implements a filtering procedure based on KD value, binding rate and dissociation rate distribution, wherein abnormal entries, such as large KD values, are filtered out by setting specific thresholds, and multiple rounds of screening are performed to select entries that meet the criteria to establish a database to ensure that it contains high-quality and reliable data.

[0055] (2) VDJdb: VDJdb is a carefully curated database containing T cell receptor (TCR) sequences and their corresponding antigen specificities. The goal of this example is to focus on the peptide-MHC (pMHC) information accompanying the TCR data. During the selection process, this example mainly filters the database based on the "species" field and selects entries related to the human species, but excludes entries for monkeys and mice. However, it is worth noting that VDJdb also contains entries related to cancer, which are marked as the "parent species", namely Homo sapiens. This example retains these entries, but marks them accordingly to provide the corresponding PMID for users to judge at their own discretion. These entries are considered to be more closely related to new epitopes rather than pathogens.

[0056] (3) McPAS: McPAS-TCR is a manually curated database focusing on TCR sequence, pathology, and antigen information. Similar to the method using VDJdb, this example filtered the data in McPAS-TCR based on the parental species field, only including entries related to human infection and excluding entries from other species.

[0057] (4) TBAdb: TCR and BCR against known antigens (TBAdb) is a manually organized database in the Pan Immune Repertoire Database (PIRD). This example obtains the database from the repository of the R language tool immunarch. Compared with VDJdb and McPAS-TCR, this database has more detailed information about T cell detection (sequencing method and number of experiments, etc.) and a scoring method to determine the reliability of the data. This example also only obtains entries when the host is human and the source of the disease is infection.

[0058] [New Epitope Database (NEPdb)]

[0059] NEPdb is a comprehensive database containing more than 17,000 validated human immunogenic and non-immunogenic neoepitope entries. These entries provide detailed information about HLA and their corresponding T cell responses. Only positive results are included in the database, ensuring that it focuses only on relevant and impactful data. For dbPepNeo, NeoDB, and TSNAdb, no additional filters were applied, and the entries in these databases were only simply adjusted to ensure compatibility and consistency. TSNAdb was used to identify and evaluate the duplication of data in different databases mentioned above, and it can be used as a reference to detect instances of data duplication in different databases.

[0060] [Structure Database]

[0061] To supplement the pathogen and new epitope databases mentioned above, this embodiment also includes collecting structural information from the TCR3d and StcrDab databases, and aligning the peptide sequences with the pathogen sequences obtained from the structural database. If the sequences have 100% sequence identity, repeated PubMed identifiers (PubMed IDs) and trigger the same MHC, this embodiment will mark the entry with a "match" symbol, and assign the corresponding structural data in the structural database (e.g., PDB ID, chain, and TCR information) to these entries.

[0062] However, if the entry happens to have duplicate sequence and MHC information but has a different PubMed identifier, the present embodiment still assigns the structural information but marks it as "fuzzy" for the user to judge. For these "fuzzy" entries, if the pathogen database and the structural database are different, the present embodiment provides the sequence alignment results between them. For entries without structural information, the present embodiment still uses a simple graphical representation to represent their pMHC and TCR status, which ensures that even without structural data, the relevant pMHC and TCR information can be conveyed to the user.

[0063] Each entry in the human biological database 16 is classified into specific groups based on fields such as antigen 161, MHC 162, detection 163, TCR 164, PDB 165, mutation 166, metadata 167, etc., which are described as follows:

[0064] [antigen]

[0065] This field provides information about the pathological conditions and corresponding diseases associated with the antigen, including details such as the antigen sequence, gene, and sequence length, as well as the parental species, source molecule, and source organism of the corresponding entry. To simplify the representation of various pathological sources, this example divides the parental species into the following four main groups:

[0066] (1) Viruses: Viruses are divided into DNA viruses and RNA viruses based on their amino acid composition.

[0067] (2) Bacteria: including pathological bacteria, environmental bacteria and probiotics.

[0068] (3) Cancer: Entries related to cancer are collected from databases such as dbPepNeo, Neodb, NEPdb, and TSNAdb. It is worth noting that VDJdb also contains entries related to cancer and marks them as "Homo sapiens". Although the specific cancer type may not be specified, this embodiment still retains these entries and classifies them under "Homo sapiens" in the cancer category.

[0069] (4) Others: This category mainly describes synthetic or plant sequences, as well as specific therapeutics for different disease conditions.

[0070] The parent species mentioned in the above entries can be further classified into subcategories, which are then divided into the four main categories mentioned above. For example, under the Virus category, there may be subcategories such as Hepacivirus, which includes Hepacivirus A, Hepacivirus B, and Heron Hepacivirus B.

[0071] [Major histocompatibility complex (MHC)]

[0072] This field displays the MHC type associated with the disease sequence. However, it is worth noting that some sources may contain outdated MHC type naming schemes. To address this issue, this example builds a conversion table to convert these old naming schemes to newer naming schemes. In this field, this example provides four key pieces of information:

[0073] (1) The name provided by the source database or literature;

[0074] (2) conversion status: indicates whether conversion has been performed for a given MHC type;

[0075] (3) Conversion name: represents the MHC type after conversion;

[0076] (4) Higher-resolution conversion nomenclature: After conversion, all possible MHC types are provided in a more detailed format.

[0077] [Assay]

[0078] This example divides the types of T cell content assays into four categories based on the reference data of IEDB:

[0079] (1) Biological Activity (BA): This category includes various T cell tests, which are mainly used to evaluate biologically active T cells. This category covers a variety of sorting methods for cultured T cells, such as sorting and culture, and also includes some specific assays such as 51 chromium, 3H-thymidine, etc.

[0080] (2) Binding assays: This category focuses on the ability of T cells to bind to antigens and includes a variety of detection methods, such as surface plasmon resonance (SPR), limiting-dilution-cloning, tetramer sorting, pentamer sorting, and multi-element assays;

[0081] (3) Cytokines: This category focuses on cytokines released by T cells and is used to evaluate the immune response of T cells, including cytometric bead array, radioimmunoassay (RIA), IFNg capture assay, bioassay, enzyme-linked immunosorbent assay (ELISA), enzyme-linked immunospot (ELISPOT) assay, and other assays;

[0082] (4) Three-dimensional structure (3D): This category mainly involves the determination of the three-dimensional structure of the T cell receptor and antigen complex, mainly using methods such as X-ray crystallography.

[0083] Each data entry is classified into the four main categories listed above. The original method and response are indicated by the "Method" and "Response" fields. For the Binding and 3D categories, the data may include qualitative and quantitative measurements to indicate the degree of binding.

[0084] [Protein Data Bank (PDB) and T cell receptor (TCR) information]

[0085] This embodiment establishes a process to compare sequences in the antigen database with sequences in the structure database. If a sequence is found to have a matching sequence, MHC type or PubMed identifier in the structure database, this embodiment will fill in the corresponding information from the structure database.

[0086] In this area, this example provides the following details in the structural database: PDB identifier (whether complex or single chain structure), PDB organism, PDB PubMed identifier and publication date, and the antigen, TCR and MHC chains present in the PDB, including additional information such as CDR3 and VDJ amino acids.

[0087] [Mutation]

[0088] For cancer entries, this example provides information about mutation types, which are categorized as follows: Single Nucleotide Variants (SNV), Small Insertions and Deletions (InDel), Copy Number Variants (CNV), and Gene Fusion.

[0089] If this information is available in the source database, this embodiment can also include details such as sequence alignment of wild-type peptide, mutant peptide, and mutant amino acid. These additional fields are helpful in describing specific mutations and their impact on the peptide sequence in the context of cancer-related entries.

[0090] [metadata]

[0091] Since all data are from published literature, this field mainly provides PubMed identifiers, titles, authors, journal information, and publication dates. In addition, this embodiment also includes information about sequencing methods and antigen sequencing organizations from literature or specific databases. The "Source" field also indicates the database from which the sequence was obtained or whether the sequence was collected through its own data collection work. If the sequence appears in multiple databases, this embodiment adds a "Common Source" field to record the databases where the sequence can be found, and adds a "Completeness Score" based on the level of available information.

[0092] Users can search the above human biological databases to identify and retrieve TCR sequences similar to their target sequences, and filter the search results by selecting various options (such as MHC type and specific genes) to obtain TCR sequences that meet specific criteria.

[0093] For example, for self-similarity, the queried antigen sequence is aligned to a protein sequence database such as Uniprot to identify similarities or matches with wild-type sequences. This comparison helps determine whether the queried antigen sequence has any known counterparts or related sequences.

[0094] In detail, in view of the diversity of TCR sequences and the tendency of similar sequences to exhibit similar characteristics, the disclosed embodiments adopt a similarity measurement method similar to the Levenshtein distance. This method quantifies the minimum number of insertions, deletions, and substitutions required to convert one sequence into another. The user can specify the maximum number of allowed gaps and character substitutions, so that sequences identical or similar to a given test sequence can be identified. In some embodiments, the BLAST algorithm is used to perform string matching on human protein sequences in a protein sequence database, and the strings are filtered according to specific criteria. The criteria include sequence search, sequence similarity, sequence length, gaps, and character substitutions. For example, the user can specify the number of allowable gaps and character substitutions.

[0095] For self-similarity, the disclosed embodiment searches in a protein sequence database to identify wild-type sequences and variant peptides, and then conducts a comprehensive search in the above-mentioned human biological database. This step is based on the assumption that the human immune system has the nature of distinguishing between self and non-self, and the proportion of peptide fragments that bind to cancer cell surface antigens derived from autologous cells can significantly affect the ability of effective immune control. Then, the mutant peptide obtained from the protein sequence database is used to search the entire human biological database to determine whether it can bind to MHC and trigger a TCR response.

[0096] Figure 2 is a schematic diagram of tumor neoantigen prediction according to an embodiment of the present disclosure. Figure 2 In the present embodiment, the prediction of tumor neoantigens is, for example, to perform next-generation sequencing (NGS), i.e., gene sequencing, on the subject to obtain multiple amino acid sequences in the gene of the subject, which are input as multiple test peptides 20 into the above-mentioned human biological database to perform epitope prediction 24.

[0097] Epitope prediction 24 includes modules for providing HLA-I model 241, HLA-II model 242, self-similarity analysis 243, and pathogen similarity analysis 244. The HLA-I model 241 is, for example, a machine learning model trained using a plurality of human proteins including 8-11 amino acids and a plurality of corresponding peptides, and is suitable for evaluating sequences of 8-11 amino acids, and the HLA-II model 242 is, for example, a machine learning model trained using a plurality of human proteins including 12-30 amino acids and a plurality of corresponding peptides, and is suitable for evaluating sequences of 12-30 amino acids, but the present embodiment is not limited thereto.

[0098] After receiving the test peptide 20 input by the user, the self-similarity analysis 243 obtains multiple human protein sequences 22 from a protein sequence database such as Uniprot, and compares them with the test peptide 20 to find multiple similar peptides that match the human protein sequence 22. The human protein sequence 22 includes amino acid sequences that mutate during the carcinogenesis of human cells. Therefore, by comparing with the human protein sequence 22, it can be determined whether the antigen sequence of the subject to be tested has any known counterparts or related sequences.

[0099] In some embodiments, the comparison is, for example, comparing the amino acid sequence as a string, and cutting out multiple fragments containing the mutant amino acid in the amino acid sequence containing the mutant amino acid using a sliding window. By string comparison, the amino acid sequences of these fragments are compared with the amino acid sequence of the test peptide 20 to confirm whether there are matching fragments. By comparing with multiple amino acid sequences containing mutant amino acids, a sequence that can be related to a known counterpart can be found from the test peptide 20.

[0100] By filtering out amino acid sequences similar to autologous cells from the test peptides, the amino acid sequences that can affect effective immune control are retained for sorting and scoring. In detail, the present embodiment inputs a test peptide 20 including 8-11 amino acids into multiple trained HLA-I models 241 to sort the test peptide 20 through the HLA-I model 241, and outputs multiple sorting results. In addition, the present embodiment also inputs a test peptide 20 including 12-30 amino acids into multiple trained HLA-II models 242 to sort the test peptide 20 through the HLA-II model 242, and outputs multiple sorting results. Among them, for each test peptide 20, the present embodiment calculates the weighted sum of its ranking in multiple sorting results as the score of the test peptide 20.

[0101] The pathogen similarity analysis 244 is to compare each of the filtered test peptides 20 with a plurality of known pathogen protein sequences to search for test peptides that match the pathogen protein sequence. Among them, the pathogen similarity analysis 244 is, for example, to compare the amino acid sequence of the test peptide 20 with the pathogen protein sequence by string comparison. This comparison method is the same or similar to the method described in the aforementioned embodiment, so it is not repeated here. In response to finding the test peptide 20 that matches the pathogen protein sequence, the pathogen similarity analysis 244 will adjust the ranking of the test peptide 20 in the aforementioned ranking result (i.e., the ranking result output by the HLA-I model 241 or the HLA-II model 242). In some embodiments, the pathogen similarity analysis 244 moves the ranking of the test peptide 20 that matches the pathogen protein sequence forward or to the first place, but is not limited to this.

[0102] The ranking results and scores of the test peptides 20 obtained through epitope prediction 24 are, for example, further evaluated through driver mutation analysis 26. Among them, the driver mutation analysis 26, for example, compares the mutated amino acids corresponding to each test peptide 20 after filtering with multiple driver mutations of known cancers to search for amino acids that match the driver mutations. In response to finding amino acids that match the driver mutations, the driver mutation analysis 26 adjusts the ranking of the test peptides containing amino acids that match the driver mutations in the ranking results. In some embodiments, the driver mutation analysis 26 moves the ranking of the test peptides 20 containing amino acids that match the driver mutations forward, and recalculates the scores of each test peptide 20 based on the changed ranking.

[0103] Finally, peptide screening 28 selects at least one of the test peptides as a new antigen suitable for the subject according to the score.

[0104] The above-mentioned tumor neoantigen prediction process can be applied to the existing immunotherapy system to predict tumor neoantigens.

[0105] Figure 3 is a block diagram of a tumor neoantigen prediction system according to an embodiment of the present disclosure. Figure 3 The tumor neoantigen prediction system 30 of the embodiment of the present disclosure includes a data acquisition device 32, a storage device 34 and a processing device 36, and its functions are described as follows:

[0106] The data acquisition device 32 is, for example, a wired or wireless transmission interface such as Universal Serial Bus (USB), RS232, Bluetooth (BT), Wireless fidelity (Wi-Fi), or a network card that supports Ethernet or wireless network standards such as 802.11g, 802.11n, 802.11ac, etc., and is used to obtain multiple amino acid sequences in the genes of the subject.

[0107] The storage device 34 is, for example, any type of fixed or removable random access memory (RAM), read-only memory (ROM), flash memory, hard disk or other similar devices, integrated circuits and combinations thereof. In some embodiments, the storage device 34 stores a protein sequence database having a plurality of human protein sequences, such as Uniprot.

[0108] The processing device 36 is an electronic device with computing capabilities such as a computer, workstation, server, etc., including a central processing unit (CPU), a graphic processing unit (GPU), or other programmable general-purpose or special-purpose microprocessors (microprocessors), digital signal processors (DSPs), image signal processors (ISPs), programmable controllers, application specific integrated circuits (ASICs), programmable logic devices (PLDs), or other similar devices or combinations of these devices. In this embodiment, the processing device 36 can execute a computer program to implement the tumor neoantigen prediction method of the disclosed embodiment.

[0109] Figure 4 is a schematic diagram of a method for predicting tumor neoantigens according to an embodiment of the present disclosure. Figure 3 and Figure 4 The method of this embodiment is applicable to Figure 3 The tumor neoantigen prediction system 30 is described below with reference to each device in the tumor neoantigen prediction system 30 to explain the detailed process of the method of this embodiment.

[0110] In step S402, the data acquisition device 32 acquires multiple amino acid sequences in the gene of the subject, and the processing device 36 uses the amino acid sequences acquired by the data acquisition device 32 as multiple test peptides to compare with human protein sequences in the protein sequence database stored in the storage device 34 to find multiple similar peptides that match the human protein sequences.

[0111] The data acquisition device 32, for example, is connected to a database recording the genetic test results of the subject to be tested, thereby obtaining multiple amino acid sequences in the genes of the subject to be tested. The processing device 36, for example, compares the test peptide and the human protein sequence by string comparison or using the BLAST tool, and the present embodiment does not limit the comparison method. Through the comparison, the processing device 36 can obtain the test peptide that matches the human protein sequence and the test peptide that does not match the human protein sequence, including the sequence length of the test peptide, the protein ID of the matched human protein sequence, and the alignment position, etc.

[0112] In detail, Figure 5 is a schematic diagram of an amino acid sequence comparison / alignment method according to an embodiment of the present disclosure. Figure 5 In this embodiment, multiple amino acid sequences in the gene of the subject are used as test peptides 52 and multiple human protein sequences 54 provided by the protein sequence database are input into the tumor neoantigen prediction system for sequence comparison / alignment 500. Among them, by string comparison or alignment using the BLAST tool, a test peptide 56 that can match the human protein sequence 54 can be found from the test peptide 52, so that the matched test peptide 56 is output, including the amino acid sequence, sequence length, matched protein identifier and alignment position, etc. On the other hand, the unmatched test peptide 56 is output, including the amino acid sequence, sequence length, etc.

[0113] In step S406 , the processing device 36 filters out similar peptides from the test peptides, and inputs the filtered test peptides into a plurality of trained human leukocyte antigen (HLA) models respectively to obtain a plurality of ranking results of the test peptides.

[0114] In some embodiments, the processing device 36 may compare each filtered test peptide with multiple known pathogen protein sequences to search for test peptides that match the pathogen protein sequence, and in response to finding a test peptide that matches the pathogen protein sequence, adjust the ranking of the test peptide in the ranking results.

[0115] In some embodiments, the processing device 36 may compare the mutated amino acids corresponding to each test peptide after filtering with multiple driver mutations of known cancers to search for amino acids that match the driver mutations, and in response to finding amino acids that match the driver mutations, adjust the ranking of the test peptides containing amino acids that match the driver mutations in the ranking results.

[0116] In step S408, the processing device 36 calculates the weighted sum of the rankings of each test peptide in the ranking results as the score of the test peptide. That is, the ranking of each test peptide in the ranking results output by different HLA models may be different, and by weighting these rankings and calculating the sum of the weighted rankings, the score of the test peptide can be obtained to evaluate whether it is suitable as a new antigen for the subject.

[0117] In step S408, the processing device 36 selects at least one test peptide as a neoantigen suitable for the subject according to the score. The selected neoantigen can be used to design a tumor neoantigen vaccine suitable for responding to gene mutations in cancer cells of the subject.

[0118] In summary, the tumor neoantigen prediction method and tumor neoantigen prediction system of the embodiments of the present disclosure perform self-similarity and pathogen similarity analysis on the amino acid sequences in the genes of the subject, and combine the HLA model and driver mutation analysis to find out the neoantigens suitable for fighting the gene mutations of the cancer cells of the subject, which can promote the advancement of T cell immunotherapy and personalized cancer treatment.

[0119] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present disclosure, rather than to limit them. Although the present disclosure has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or replace some or all of the technical features therein by equivalents. However, these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present disclosure.

Claims

1. A method for predicting tumor neoantigens, the method comprising the following steps: Acquire multiple amino acid sequences in the gene of the subject as multiple test peptides and compare them with multiple human protein sequences in a protein sequence database to find multiple similar peptides matching the human protein sequences; Filtering out the similar peptides from the test peptides, and inputting the filtered test peptides into a plurality of trained human leukocyte antigen models respectively to obtain a plurality of ranking results of the test peptides; Calculating the weighted sum of the rankings of each of the test peptides in the ranking results as the score of the test peptide; as well as At least one of the test peptides is selected as a new antigen suitable for the subject according to the score.

2. The method according to claim 1, wherein after the step of obtaining the ranking result of the test peptide, the method further comprises: Comparing each of the filtered test peptides with a plurality of known pathogen protein sequences to search for the test peptide that matches the pathogen protein sequence; as well as In response to finding the test peptide matching the pathogen protein sequence, adjusting the ranking of the test peptide in the ranking results.

3. The method according to claim 1, wherein the human protein sequence comprises an amino acid sequence that has mutated during the carcinogenesis of human cells.

4. The method according to claim 3, wherein after the step of obtaining the ranking result of the test peptide, the method further comprises: Comparing the mutated amino acid corresponding to each of the test peptides after filtering with a plurality of driver mutations of known cancers to search for the amino acid matching the driver mutation; and In response to finding the amino acid matching the driver mutation, adjusting the ranking of the test peptide comprising the amino acid matching the driver mutation in the ranking results.

5. The method according to claim 1, wherein each of the human leukocyte antigen models comprises a machine learning model trained using a plurality of human proteins and a plurality of peptides corresponding thereto.

6. A tumor neoantigen prediction system, comprising: A data acquisition device for acquiring multiple amino acid sequences in the gene of the subject to be tested; A storage device storing a protein sequence database recording a plurality of human protein sequences; as well as A processing device, coupled to the data acquisition device and the storage device, is configured to: Comparing the amino acid sequences acquired by the data acquisition device as a plurality of test peptides with the human protein sequences in the protein sequence database to find a plurality of similar peptides matching the human protein sequences; Filtering out the similar peptides from the test peptides, and inputting the filtered test peptides into a plurality of trained human leukocyte antigen models respectively to obtain a plurality of ranking results of the test peptides; Calculating the weighted sum of the rankings of each of the test peptides in the ranking results as the score of the test peptide; as well as At least one of the test peptides is selected as a new antigen suitable for the subject according to the score.

7. The tumor neoantigen prediction system according to claim 6, wherein the processing device further compares each of the filtered test peptides with multiple known pathogen protein sequences to search for the test peptides that match the pathogen protein sequence, and in response to finding the test peptides that match the pathogen protein sequence, adjusts the ranking of the test peptides in the ranking results.

8. The tumor neoantigen prediction system according to claim 6, wherein the human protein sequence comprises an amino acid sequence that has mutated during the carcinogenesis of human cells.

9. The tumor neoantigen prediction system according to claim 8, wherein the processing device further compares the mutated amino acids corresponding to each of the filtered test peptides with multiple driver mutations of known cancers to search for the amino acids matching the driver mutations, and in response to finding the amino acids matching the driver mutations, adjusts the ranking of the test peptides containing the amino acids matching the driver mutations in the ranking results.

10. The tumor neoantigen prediction system according to claim 6, wherein each of the human leukocyte antigen models comprises a machine learning model trained using a plurality of human proteins and a corresponding plurality of peptides.

Images