Ranking of neoantigens for personalized cancer vaccines

Abstract

Disclosed herein is a method for ranking tumor-specific neoantigens from a subject's tumor that are suitable for a subject-specific immunogenic composition. Suitable tumor-specific neoantigens are those that are likely to be presented on the cell surface of the tumor, likely to be immunogenic, predicted to be expressed in sufficient amounts to elicit an immune response in the subject, optionally exhibit sufficient diversity throughout the tumor, and have relatively high manufacturability. The method takes a group of neoantigens (peptide vaccine candidates) and ranks the neoantigens such that the top ranked group of neoantigens simultaneously promotes cell surface presentation of key neoantigens to class I and class II MHC molecules. The top ranked neoantigens can then be further refined according to manufacturability and / or other criteria.

Description

[Technical Field]

[0001] This application claims priority to U.S. Provisional Patent Application No. 63 / 146,392, filed on 5 February 2021, which is incorporated herein by reference.

[0002] Sequence listing reference This application includes a computer-readable sequence listing. The computer-readable format is incorporated herein by reference. The filename of the ASCII copy created on February 3, 2022, is 146401_091686_SL.txt, and its size is 14,005 bytes. [Background technology]

[0003] Cancer is a leading cause of death worldwide, with approximately one in four people dying from it. (Siegel et al., CA: A Cancer Journal for Clinicians, 68:7-30 (2018)). In 2018, 18.1 million people were newly diagnosed with cancer, and 9.6 million died from cancer-related causes. (Bray et al., CA: A Cancer Journal for Clinicians, 68(6):394-424). Numerous standard cancer treatments exist, including surgical techniques (such as surgical procedures and radiation) and chemical techniques (such as chemotherapy). Unfortunately, such treatments often carry significant risks, toxic side effects, and are extremely expensive, and their effectiveness is uncertain.

[0004] Cancer immunotherapy (such as cancer vaccines) is emerging as a promising cancer treatment. The goal of cancer immunotherapy is to activate the immune system to selectively destroy cancer while protecting healthy tissue intact. Conventional cancer vaccines typically target tumor-associated antigens. Tumor-associated antigens are normally present in healthy tissue but are overexpressed in cancer. However, because these antigens are often present in healthy tissue, immune activation may be hindered by immune tolerance. Some clinical trials targeting tumor-associated antigens have failed to demonstrate sustained beneficial effects compared to standard treatment. Li et al., Ann Oncol., 28 (Suppl 12): xii11-xii17 (2017).

[0005] Neoantigens can be very attractive targets for cancer immunotherapy. Neoantigens are non-self proteins with individual specificity. Neoantigens originate from random somatic mutations in the tumor cell genome and are not expressed on the surface of normal cells. Ibid. Neoantigens are expressed only on tumor cells and do not induce central immune tolerance, so cancer vaccines targeting cancer neoantigens can be expected to offer advantages such as reduced central immune tolerance and improved safety profiles. Ibid.

[0006] The landscape of cancer mutations is complex, and tumor mutations are unique to each subject. Most somatic mutations detected by sequencing do not result in effective neoantigens. Only a small fraction of tumor DNA or tumor cell mutations are transcribed and translated with enough precision to design an effective vaccine and be processed into tumor-specific neoantigens. Furthermore, not all neoantigens are immunogenic. In fact, only about 1% to 2% of T cells spontaneously recognize endogenous neoantigens. See Karpanen et al., Front Immunol., 8:1718 (2017). Moreover, the cost and time involved in manufacturing neoantigen vaccines are enormous.

[0007] Therefore, efficiently and accurately predicting, prioritizing, and selecting neoantigen candidates for immunogenic compositions remains a challenge. Thus, there is a significant unmet need for an integrated method to characterize tumor genomic materials to identify neoantigens, identify which neoantigens are targeted by the immune system, and select which neoantigens are likely to be suitable for effective immunogenic compositions.

[0008] Summary of the Invention This disclosure relates to a novel method for ranking one or more appropriate tumor-specific neoantigens from a subject's tumor for a personalized (i.e., subject-specific) immunogenic composition. The disclosure also relates to a method for treating cancer in a subject in need by administering an immunogenic composition containing selected tumor-specific neoantigens, using a novel approach to ranking tumor-specific neoantigens and formulating an immunogenic composition containing the selected tumor-specific neoantigens based on the ranking technique. Appropriate tumor-specific neoantigens are those that are likely to be present on the tumor cell surface, are likely to be immunogenic, are expected to be expressed in sufficient amounts to induce an immune response in the subject, may exhibit sufficient diversity throughout the tumor, and have relatively high manufacturability. The method involves obtaining a group of neoantigens (peptide vaccine candidates), ranking the neoantigens, and ensuring that the higher-ranked groups of neoantigens simultaneously promote the cell surface presentation of key neoantigens to class I and class II MHC molecules. Subsequently, the higher-ranked groups of neoantigens can be further refined according to manufacturability and / or other criteria.

[0009] This approach begins with obtaining sequence data from tumors. The sequence data is used to obtain data representing the polypeptide sequences of one or more tumor-specific neoantigens. The sequence data may be nucleotide sequence data, polypeptide sequence data, exome sequence data, transcriptome sequence data, or whole-genome nucleotide sequence data. A suitable tumor-specific neoantigen is one that is likely to be present on the tumor cell surface, is likely to be immunogenic, is predicted to be expressed in sufficient amounts to induce an immune response in the subject, may exhibit sufficient diversity throughout the tumor, and has relatively high manufacturability. The method obtains a group of neoantigens (peptide vaccine candidates), ranks them, and ensures that the top-ranked groups of neoantigens simultaneously promote the cell surface presentation of key neoantigens to class I and class II MHC molecules. Subsequently, the top-ranked groups of neoantigens can be further refined according to manufacturability and / or other criteria. As will be described in more detail below, the ranking is primarily based on the calculation of the immunogenicity of the neoantigens. For short-chain neoantigens, the immunogenicity of the short-chain neoantigen is determined at least partially based on the probability that at least one allele among a subject's multiple HLA class I alleles presents the short-chain neoantigen and does not present germline siblings of the short-chain neoantigen. Similarly, for long-chain neoantigens, the immunogenicity of the long-chain neoantigen is determined at least partially based on the probability that at least one allele among a subject's multiple HLA class II alleles presents the long-chain neoantigen and does not present germline siblings of the long-chain neoantigen. These probabilities are determined using the output provided by one or more machine learning platforms / models, which are trained to determine the probability that a given allele presents a particular antigen.

[0010] An immunogenic composition formulated at least in part based on the technology of the present invention may contain at least about 10 tumor-specific neoantigens, or at least about 20 tumor-specific neoantigens. The tumor-specific neoantigens may be encoded by short-chain polypeptides or long-chain polypeptides. The immunogenic composition may contain nucleotide sequences, polypeptide sequences, RNA, DNA, cells, plasmids, vectors, dendritic cells, or long synthetic peptides. The immunogenic composition may further contain adjuvants.

[0011] This disclosure also relates to a method for treating cancer in subjects requiring it, comprising administering a personalized immunogenic composition comprising one or more tumor-specific neoantigens selected using the method described herein. The method disclosed herein may be suitable for the treatment of a variety of cancers. Tumors include melanoma, breast cancer, ovarian cancer, prostate cancer, kidney cancer, gastric cancer, colon cancer, testicular cancer, head and neck cancer, pancreatic cancer, brain cancer, B-cell lymphoma, acute myeloid leukemia, chronic myeloid neoplasm, leukemia, chronic lymphocytic leukemia, T-cell lymphocytic leukemia, bladder cancer, or lung cancer. Preferably, the cancers are melanoma, breast cancer, lung cancer, and bladder cancer.

[0012] Various embodiments relating to this disclosure will be described with reference to the drawings. [Brief explanation of the drawing]

[0013] [Figure 1] This document illustrates an exemplary provider network (or “service provider system”) environment in some embodiments. [Figure 2] This is a block diagram of an exemplary provider network that provides storage services and hardware virtualization services to customers, according to one embodiment. [Figure 3] This document shows systems that implement some or all of the technologies described herein, according to certain embodiments. [Figure 4]A method for ranking tumor-specific neoantigens from a subject's tumor with respect to a target-specific immunogenic composition is described by exemplary embodiments. [Modes for carrying out the invention]

[0014] This disclosure relates to a novel approach for ranking tumor-specific neoantigens for inclusion in potent personalized oncoimmunogenic compositions (e.g., target-specific immunogenic compositions). This disclosure also relates to a method for treating cancer in subjects in need by administering an immunogenic composition containing tumor-specific neoantigens, which has been formed using a novel approach for ranking tumor-specific neoantigens and formulating an immunogenic composition containing the selected tumor-specific neoantigens.

[0015] All patents and publications cited herein are incorporated herein by reference in their entirety. In the event of any conflict between the incorporated material and this specification, this specification shall prevail. Furthermore, the reference of any reference herein shall not imply that such reference is prior art to this disclosure. Where a range of values ​​is expressed, embodiments using specific values ​​within that range shall be included. Furthermore, referring to a value within a range shall include all values ​​within that range. All ranges include endpoints and are combinable. Where a value is expressed as an approximation, the use of the antecedent “about” shall be understood to mean that a particular value constitutes another embodiment. References to a particular numerical value shall include at least that particular value unless otherwise clearly indicated in the context. The use of “or” shall mean “and / or” unless otherwise indicated in the specific context.

[0016] This specification and claims use various terms relating to aspects of the present invention. Unless otherwise indicated, such terms shall be given their common meanings in the art. Other terms specifically defined shall be construed in accordance with the definitions set forth herein. The techniques and procedures described or referenced herein are generally well understood by those skilled in the art and employ well-known conventional methodologies, such as Sambrook et al., Molecular Cloning: A Laboratory Manual 4th ed. (2012), Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY. Procedures involving the use of commercially available kits and reagents, where necessary, are generally performed according to the protocols and conditions specified by the manufacturers, unless otherwise indicated.

[0017] In this specification, the singular forms "a," "an," and "the" include the plural form unless the context clearly indicates otherwise. Terms such as "include" and "such as" mean inclusion without restriction unless otherwise specified.

[0018] Unless otherwise indicated, the terms “at least,” “less than,” “about,” or similar terms used as antecedents for a set of elements or scope should be understood to refer to all elements within that set or scope. Those skilled in the art should be able to recognize or confirm many equivalents to the specific embodiments of the invention described herein by ordinary experimentation alone. Such equivalents are intended to be covered by the following claims.

[0019] "Cancer" refers to a physiological condition in a subject characterized by uncontrolled proliferation, immortality, metastatic potential, rapid growth, proliferation rate, and / or specific morphological properties of a cell population. In many cases, cancer takes the form of a tumor or mass, but it may exist alone in the subject's body, or circulate in the bloodstream as independent cells such as leukemia cells or lymphoma cells. The term "cancer" encompasses all types of cancers and metastases, including hematological malignancies, solid tumors, sarcomas, carcinomas, and other solid and non-solid tumors. Examples of cancer include, but are not limited to, carcinoma, lymphoma, blastoma, sarcoma, and leukemia. More specific examples of such cancers include squamous cell carcinoma, small cell lung cancer, non-small cell lung cancer, lung adenocarcinoma, lung squamous cell carcinoma, peritoneal cancer, hepatocellular carcinoma, gastrointestinal cancer, pancreatic cancer, glioblastoma, cervical cancer, ovarian cancer, liver cancer, bladder cancer, hepatic cancer, breast cancer (e.g., triple-negative breast cancer, hormone receptor-positive breast cancer), osteosarcoma, melanoma, colon cancer, colorectal cancer, endometrial (e.g., serous) or uterine cancer, salivary gland cancer, renal cancer, liver cancer, prostate cancer, vulvar cancer, thyroid cancer, liver cancer, and various types of head and neck cancers. In some embodiments, triple-negative breast cancer (TNBC) refers to breast cancer that does not express the estrogen receptor (ER), progesterone receptor (PR), or Her2 / neu gene. Hormone receptor-positive breast cancer refers to breast cancer in which at least one of ER or PR is positive and Her2 / neu (HER2) is negative.

[0020] (As used herein, "neoantigen" refers to an antigen that has at least one alteration that distinguishes it from the corresponding parent antigen, for example via mutations in tumor cells or post-translational modifications specific to tumor cells. Mutations or variations include frameshifts, indels, missense or nonsense substitutions, splice site changes, genomic rearrangements or gene fusions, or changes in genomic expression that give rise to neoantigens. Mutations may include splice mutations. Post-translational modifications specific to tumor cells may include aberrant phosphorylation. Post-translational modifications specific to tumor cells may also include splice antigens generated by the proteasome. See Lipe et al., Science, 354(6310):354:358 (2016). Typically, point mutations account for about 95% of tumor mutations, with indels and frameshift mutations accounting for the remainder. See Snyder et al., N Engl J Med., 371:2189-2199 (2014).

[0021] As used herein, "tumor-specific neoantigen" refers to a neoantigen that is present in the tumor cells or tissue of a subject, but not present in the normal cells or tissue of the subject.

[0022] As used herein, "germline sibling" refers to a germline antigen that represents the non-mutated peptide equivalent of the corresponding neoantigen.

[0023] As used herein, "next-generation sequencing" or "NGS" refers to a sequencing technology with improved throughput compared to conventional approaches (e.g., the Sanger method) and the capability to generate hundreds of thousands of sequence reads at a time.

[0024] As used herein, "neural network" refers to a machine learning model for classification or regression, which is typically composed of multiple layers of linear transformations followed by element-wise non-linearities, trained by stochastic gradient descent and backpropagation.

[0025] In this specification, "subject" refers to any animal, including but not limited to humans, non-human primates, rodents, and other mammals. In some embodiments, the mammal is a mouse. In some embodiments, the mammal is a human.

[0026] In this specification, “cancer cell” means any cell that is a cancer cell or derived from a cancer cell. “Tumor cell” also means a cell that exhibits cancer-like characteristics, such as uncontrolled replication, resistance to antiproliferative signals, metastatic ability, and loss of the ability to undergo programmed cell death.

[0027] This specification contains supplementary information about the method and guidelines for implementing the method.

[0028] I. Methods for ranking tumor-specific neoantigens This specification discloses a method for ranking tumor-specific neoantigens derived from a subject's tumors that are suitable for a target-specific immunogenic composition. A suitable tumor-specific neoantigen is one that is likely to be presented on the tumor cell surface, is likely to be immunogenic, is expected to be expressed in sufficient amounts to induce an immune response in the subject, exhibits sufficient diversity throughout the tumor as needed, and has relatively high manufacturability. The method involves obtaining a group of neoantigens (peptide vaccine candidates), ranking the neoantigens, and ensuring that the higher-ranked group of neoantigens simultaneously promotes the cell surface presentation of key neoantigens to class I and class II MHC molecules. Subsequently, the higher-ranked group of neoantigens can be further refined according to manufacturability and / or other criteria.

[0029] Tumor and subject sequence data are used to rank tumor-specific neoantigens derived from the subject's tumor. Tumor sequence data is used to obtain data representing the polypeptide sequences of one or more tumor-specific neoantigens. Typically, sequence data representing the polypeptide sequences of one or more tumor-specific neoantigens can be obtained by performing sequence analysis on a tumor sample. In some embodiments, obtaining sequence data involves receiving or accessing data stored from previously performed sequencing. Sequence data may be exome sequence data, transcriptome sequence data, whole genome nucleotide sequence data, nucleotide sequence data, or polypeptide sequence data. Various methods for obtaining tumor and subject sequence data may be used in the methods described herein. Some exemplary sequencing methods are described in more detail below.

[0030] Once sequence data representing polypeptide sequences of one or more tumor-specific neoantigens is obtained, this sequence data can be analyzed together with the subject's MHC molecules to identify and rank candidate neoantigens for inclusion in the subject's immunogenic composition.

[0031] In one embodiment, given a subject's HLA I and HLAI I alleles and a list of tumor somatic mutations, a top-ranked group of approximately 30 long-chain peptide candidates and approximately 15 short-chain peptide candidates are identified and subjected to feasibility analysis. The peptide starter group is identified using a sliding window group for each somatic mutation. These are scored using the MHC class I and class II machine learning models described below. The 15 short-chain peptides and 30 long-chain peptides contain at least one MHC class I epitope, and the long-chain peptides may also contain one or more MHC class II epitopes. Then, nine of the 30 long-chain peptide candidates and ten of the 15 short-chain peptide candidates are selected for inclusion in an immunogenic composition based on feasibility. In another embodiment, the top-ranked long-chain and / or short-chain peptide candidates are provided in varying numbers for feasibility analysis. In some embodiments, 20 to 100 (for example, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54, 55, 56, 57, 58) Candidates are provided as follows: 59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 73, 74, 75, 76, 77, 78, 79, 80, 81, 82, 83, 84, 85, 86, 87, 88, 89, 90, 91, 92, 93, 94, 95, 96, 97, 98, 99, 100). In other embodiments, more or fewer top-ranked candidates may be provided for manufacturability analysis.

[0032] Similarly, a variety of long-chain and / or short-chain peptide candidates may ultimately be selected for addition to immunogenic compositions. Typically, more long-chain neoantigens are required because they have more manufacturing constraints than short-chain neoantigens. In some embodiments, neoantigens having about 15–30 amino acids (e.g., 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, or 30 amino acids) are defined as long-chain neoantigens, and neoantigens having about 8–11 amino acids (e.g., 8, 9, 10, or 11 amino acids) are defined as short-chain neoantigens. In other embodiments / implementations of the technology, long-chain and short-chain neoantigens having different numbers of amino acids may be defined.

[0033] Figure 4 illustrates an exemplary method (400) for ranking tumor-specific neoantigens from a subject's tumor with respect to a subject-specific immunogenic composition. First, multiple somatic mutations present within the tumor are identified (410). Next, for each individual somatic mutation, the first multiple short-chain neoantigens and the first multiple long-chain neoantigens associated with that somatic mutation are identified or obtained by other means (420). The first multiple short-chain neoantigens may include short-chain polypeptides containing at least one MHC class I epitope associated with the subject. The first multiple long-chain neoantigens may include long-chain polypeptides containing at least one MHC class I epitope and at least one MHC class II epitope associated with the subject.

[0034] From the initial group of short-chain neoantigens, the one with the highest immunogenicity score is selected or determined (430) and added to the short-chain neoantigen candidate list (440). The selected short-chain neoantigen is also called the best short-chain neoantigen for a particular somatic mutation. Similarly, from the initial group of long-chain neoantigens, the long-chain neoantigen with the highest immunogenicity score is selected or determined (460) and added to the long-chain neoantigen candidate list (470). The selected long-chain neoantigen is also called the best long-chain neoantigen for a particular somatic mutation. The immunogenicity score may be any form of evaluation or value, such as numerical or non-numerical, used to represent the quality of a neoantigen based on one or more data points with respect to one or more criteria. The immunogenicity score of a neoantigen can be determined according to the techniques described in detail below. The step of selecting the best short-chain neoantigen and best long-chain neoantigen is performed for each somatic mutation among multiple somatic mutations, and the completed short-chain neoantigen candidate list contains the best short-chain neoantigen for each somatic mutation, and the completed long-chain neoantigen candidate list contains the best long-chain neoantigen for each somatic mutation. In other words, each short-chain neoantigen in the short-chain neoantigen candidate list may be the best short-chain neoantigen for each of the multiple identified somatic mutations. Similarly, each long-chain neoantigen in the long-chain neoantigen candidate list may be the best long-chain neoantigen for each of the multiple identified somatic mutations. For each somatic mutation, the best short-chain neoantigen and best long-chain neoantigen are identified. The short-chain neoantigen candidate list and the long-chain neoantigen candidate list are then sorted and ranked in descending order of immunogenicity score (450, 480).

[0035] In some embodiments, a sorted list of long-chain neoantigen candidates is trimmed to a predetermined number of top-ranked candidates. For example, the list may be trimmed to the top 30 long-chain neoantigen candidates. In other words, a predetermined number of top-ranked long-chain neoantigens from the sorted list are selected for manufacturability analysis and determination. In some embodiments, the trimmed list of long-chain neoantigen candidates (i.e., a predetermined number of top-ranked long-chain neoantigens) is provided to the manufacturer for determining manufacturability. The manufacturability of a particular neoantigen may be expressed as a numerical or non-numerical score, value, classification, etc. Manufacturability may be based on one or more criteria or data that can be calculated, weighted, or otherwise processed in various ways. The determination of manufacturability may be based on an analysis performed on actual neoantigens or on reference materials. The manufacturer then selects a subgroup of long-chain neoantigen candidates from the trimmed list (i.e., the top group of manufacturability scores) based on manufacturability. For example, the lower group includes the top nine long-chain neoantigens with the highest manufacturability scores. In this specification, "manufacturer" refers to an entity that conducts the manufacturability analysis and selects the lower group, but may also refer to a similar entity or third party responsible for the remaining technologies.

[0036] Once a subgroup of long-chain neoantigen candidates is obtained based on manufacturability, any neoantigens in the short-chain neoantigen candidate list that are included in any of the neoantigens within the long-chain neoantigen subgroup are removed from the short-chain neoantigen candidate list to avoid duplication. In other words, for each long-chain neoantigen within the long-chain neoantigen candidate subgroup, its mutation is identified, and the corresponding short-chain neoantigen is removed from the short-chain neoantigen candidate list. The short-chain neoantigens remaining in the short-chain neoantigen candidate list are trimmed to a predetermined number based on their immunogenicity score. For example, the list may be trimmed to approximately 15 neoantigens. These short-chain candidates are then assessed for manufacturability, and a subgroup of short-chain neoantigen candidates selected for their manufacturability is obtained. The subgroups of short-chain neoantigen candidates and long-chain neoantigen candidates are used to form or generate target-specific immunogenic compositions that can be administered to subjects.

[0037] Acquisition of the first multiple short-chain neoantigens: To obtain the first defined set of short neoantigens for individual somatic mutations, the longest neoantigen sequence containing the mutated amino acid, neoT, is first identified. The germline sibling neoG of this neoantigen is also identified. Next, all neoantigen sequences containing mutations with a minimum (e.g., 8) to maximum (e.g., 11) amino acid counts are identified using a sliding window across the longest neoantigen sequence. This yields the initial set of short neoantigens, neoT_1. For example, in embodiments where the minimum number of amino acids is 8 and the maximum is 11, all neoantigen sequences containing mutations and having lengths of 8, 9, 10, and 11 amino acids within the longest neoantigen neoT are identified and designated as members of the initial set of short neoantigens neoT_1. In some embodiments, individual alleles in multiple HLA class I alleles a1 of the subject are identified.

number

number

number

number

[0038] In some embodiments, the probability may be determined at least in part based on data from an MHC class I machine learning model trained to determine the probability that a given allele among multiple HLA class I alleles presents a specific antigen.

[0039] The initial set of short chain neoantigens is further filtered to exclude neoantigens that are nested within one of the initial set of short chain neoantigens, or within another neoantigen. This filtering may be performed by identifying pairs of neoantigens in which one sequence is nested within the other, resulting in a higher probability score, as calculated using equation (1) above.

number

[0040] Acquisition of the first multiple long-chain neoantigens: Once the initial set of short neoantigens is defined, the short subsequence T1 is identified from the longest neoantigen sequence neoT. The short subsequence T1 is identified as the shortest subsequence of the longest neoantigen sequence neoT, which contains all the neoantigens in the initial set of short neoantigens neoT_1. As described above, the filtered initial set of short neoantigens neoT_1_filt either does not contain any of the neoantigens present in the initial set of short neoantigens, or it contains other neoantigens. The extended sequence T1_long can then be identified. The extended sequence T1_long is obtained by adding amino acids to both sides of the short subsequence T1, according to the longest neoantigen neoT. This results in a first maximum number of amino acids adjacent to each side of the mutant amino acid. For example, the first maximum number is 29. In some embodiments, the second maximum number of amino acids may be 9 to 50 (for example, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50). Accordingly, in these embodiments, the extended sequence T1_long includes the short subsequence T1 and 29 amino acids adjacent to both sides of the mutant amino acid.

[0041] All possible subsequences may be identified from long subsequences having a length between the length of a short subsequence [length(T1)] and the length of the second maximum number of amino acids, and designated as the first multiple long-chain neoantigen neoT_2. For example, the second maximum number is 30. In some embodiments, the second maximum number of amino acids may be 9 to 50 (e.g., 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50). In some embodiments, the first multiple long-chain neoantigens may be filtered based on one or more manufacturability conditions.

[0042] Obtaining immunogenicity scores for neoantigens: The immunogenicity score of individual short-chain neoantigens used for selection and ranking (sorting) short-chain neoantigens can be determined at least in part based on the probability that at least one allele in a subject's multiple HLA class I alleles presents the individual short-chain neoantigen and does not present germline siblings of the individual short-chain neoantigen. This probability can be expressed as follows:

number

[0043] In some embodiments, the above calculated values ​​are called the pan-HLAI allele scores for each short-chain neoantigen in the initial group of short-chain neoantigens before filtering for nesting pairs.

number

[0044] Furthermore, in some embodiments, the allele score of a peptide sequence containing mutant amino acids and MHC class I epitopes for individual MHC class I alleles is determined based on the probability that the individual allele presents at least one neoantigen in the first multiple short-chain neoantigens after filtering for nest pairs, and does not present at least one germline sibling of the neoantigen. This can be expressed as follows:

number

[0045]

number

[0046] In some embodiments, the pan-allele HLA class I score is determined at least partially based on the probability that at least one allele in a subject's multiple HLA class I alleles presents at least one neoantigen in the short-chain neoantigen group and does not present germline siblings of at least one neoantigen. This can be expressed as follows:

number

[0047] The immunogenicity score of individual long-chain neoantigens used for the selection and ranking (sorting) of long-chain neoantigens can be determined at least partially based on the probability that at least one allele in a subject's multiple HLA class II alleles presents the individual neoantigen and does not present germline siblings of the individual neoantigen. Each neoantigen within the initial multiple long-chain neoantigens neoT_2,

number

number

number

number

number

[0048] Therefore, the probability that at least one allele in a subject's multiple HLAII alleles presents a specific neoantigen and does not present germline siblings of that specific neoantigen can be expressed as follows:

number

[0049] The probability is determined at least in part based on data from an MHC class II machine learning model trained to determine the probability that a given allele in multiple HLA class II alleles presents a specific antigen.

[0050] The probability that a mutated peptide sequence generates an immune response to one or more HLA class I alleles can be expressed as follows:

number

number

number

[0051] Short peptides can be directly incorporated into vaccines and are expected to compete with endogenously expressed peptides for binding to MHC-I. Therefore, for these peptides, the score S may be adjusted to also include the predicted binding probability of the peptide to a given MHC-I molecule. The modified score for short peptides can be expressed as follows.

Math.

[0052] Allele-specific CD8+ immunogenicity can be expressed as follows.

Math.

[0053] DistToSelf can be expressed as follows.

Math.

[0054] The probability of immunogenicity of mutant peptides similar to germline peptides can be suppressed using the following function.

number

number

[0055] In some embodiments, a method for ranking tumor-specific neoantigens from a subject's tumor with respect to a target-specific immunogenic composition involves identifying multiple somatic mutations present within the tumor, determining the best short-chain neoantigen from the first multiple short-chain neoantigens for each somatic mutation in the multiple somatic mutations, and determining the best long-chain neoantigen from the first multiple long-chain neoantigens based at least partially on the quality score of the best long-chain neoantigen. The best short-chain neoantigen for each somatic mutation is added to a short-chain neoantigen candidate list, and the best long-chain neoantigen for each somatic mutation is added to a long-chain neoantigen candidate list. The lists are then ranked in descending order of quality score. In some embodiments, the quality score is at least partially based on at least one of the predicted presentation probability, predicted binding affinity, and predicted immunogenic response. In some embodiments, the quality score is at least partially based on the predicted presentation probability. In some embodiments, the quality score is at least partially based on the predicted binding affinity. In some embodiments, the predicted binding affinity is determined at least in part on data from an MHC class II learning model trained to determine the binding affinity between a class II allele and a given peptide. In some embodiments, the quality score is determined at least in part on the predicted immunogenicity response. In some embodiments, the quality score is determined at least in part on a combination of the predicted presentation probability, predicted binding affinity, and predicted presentation probability. In some embodiments, the predicted presentation probability, predicted binding affinity, and predicted presentation probability are determined by one or more machine learning models.

[0056] In some embodiments, peptides may be filtered for consideration or incorporation into the final target-specific immunogenic composition based on any group of the following criteria: 1) RNA abundance of the gene to which the somatic mutation belongs (measured in transcripts per million, TPM). For example, RNA abundance can be determined by multiplying the RNATPM value of the gene to which the mutant belongs by the ratio of the number of reads superimposed on the mutant locus containing the mutant allele to the sum of (a) the number of reads superimposed on the mutant locus containing the mutant allele and (b) the total number of reads superimposed on the mutant locus. 2) Whether the somatic mutation is in an essential gene or a driver gene. A driver gene is a gene in which a mutation may cause tumor growth. An essential gene is a gene essential for the survival of the organism. 3) Whether the peptide is expected to pass quality control thresholds for synthesization and solubility. 4) To what extent is the mutant peptide heterogeneous (i.e., different) from the corresponding germline peptide? In some embodiments, the peptides to be considered or incorporated require a minimum number of mutant amino acids, and peptides with a high degree of exogenousness may be preferred over those with low exogenousness. 5) The confidence score for the presence of a specific mutation in a specific subject. For example, rare somatic mutations are given a lower confidence score than more frequently occurring mutations. 6) Whether or not the peptide candidate contains a specific amino acid such as cysteine.

[0057] Somatic mutants can mutate 0, 1, or more amino acids. For example, silent mutations mutate 0 amino acids, single-nucleotide mutations typically mutate 1 amino acid, and frameshift or stop-loss mutations can mutate multiple amino acids. If the RNA upper read is found to be assembled upstream of the mutated locus, the longest consensus mRNA sequence overlapping with the mutated amino acid is assembled. mRNA sequence assembly is stopped when RNA read coverage is complete or a new stop codon is found. If no RNA upper read is found, or if no mutated amino acids exceeding the required protein sequence length are found, mRNA sequence assembly is stopped.

[0058] The predictive presentation data consists entirely of "positive" samples that may be presented on the cell surface. Therefore, one or more stochastic negative mining strategies can be used during training to train predictors that may require "negative" samples that cannot be presented on the cell surface. Such a process may involve HLA allele shuffling, where, given a positive sample (e.g., a peptide and its corresponding HLA allele), a different allele that does not belong to the supertype of that positive allele can be randomly sampled and replaced with a given allele. Each HLA allele may be classified into one or more HLA supertypes until only unclassified HLA alleles remain. Unclassified HLA alleles can be mapped to one or more "unclassified" supertype classes, and these groups can be processed in the same way as classified supertype classes.

[0059] Furthermore, predictors can also be trained using peptide shuffling. In this case, given a positive sample consisting of a peptide and its corresponding HLA allele, the peptide is replaced with an amino acid subsequence of the same length randomly sampled from the peptide's protein source.

[0060] To assist in training predictors, random peptides may be generated. According to this embodiment, random peptides can be generated sampled from an amino acid data distribution, and the target qualitative affinity is below a predetermined threshold and a negative presentation target. The length of the random peptide is determined such that there are an equal number of unbound data points for each peptide length for each allele. A predetermined ratio (e.g., 10:1) of negative presentation samples to positive presentation samples can be sampled, and sample weights can be applied to the negative samples for equilibrium loss. For each negative sample, the sampling method can be randomly selected with a uniform distribution.

[0061] Methods of sequencing Various sequencing methods are well known in the art, including, but are not limited to, PCR-based methods including real-time PC, whole exome sequencing, deep sequencing, high-throughput sequencing, or combinations thereof. In some embodiments, the aforementioned techniques and procedures are carried out according to the methods described, for example, Sambrook et al., Molecular Cloning: A Laboratory Manual 4th ed. (2012), Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY. Other references include Austell et al., Current Protocols in Molecular Biology, ed., Greene Publishing and Wiley-Interscience New York (1992) (regularly updated).

[0062] Sequence determination methods include high-throughput sequencing, single-cell RNA sequencing, RNA sequencing, pyro-sequencing, synthetic sequencing, single-molecule sequencing, nanopore sequencing, semiconductor sequencing, synthetic sequencing, ligation sequencing, hybridization sequencing, RNA-Sew (Illumina), Digital Gene Expression (Helicos), next-generation sequencing, synthetic single-molecule sequencing (SMSS) (Helicos), massively parallel sequencing, clonal single-molecule array (Solexa), shotgun sequencing, Maxham-Hilbury or Sanger sequencing, whole-genome sequencing, whole-exome sequencing, primer walking, sequencing using PacBio, SOLid, IonTorrent, or Nanopore platforms, and any other sequencing methods known in the art. The sequencing method used herein to obtain sequence data is preferably high-throughput sequencing. High-throughput sequencing techniques can sequence multiple nucleic acid molecules in parallel and can sequence millions of nucleic acid molecules at once. See Churko et al., Circ. Res. 112(12):1613-1623 (2013).

[0063] In some cases, high-throughput sequencing can be replaced with next-generation sequencing. Numerous next-generation platforms exist that utilize different sequencing techniques (e.g., using HiSeq or MiSeq instruments available from Illumina, San Diego, California). Any of these platforms can be used for sequencing the genetic material disclosed herein. Next-generation sequencing is based on the sequencing of a large number of independent reads, each representing a nucleic acid of 10 to 1000 base pairs. Synthetic sequencing is a common technique used in next-generation sequencing. Generally, sequencing involves hybridizing primers to a template to form a template / primer duplex, and then contacting the duplex with polymerase under conditions where polymerase can add nucleotides to the primer in a template-dependent manner in the presence of detectably labeled nucleotides. This step is then repeated sequentially to identify the incorporated bases using signals from the detectable labels and to determine the linear order of nucleotides in the template. Detectable labels include radiolabeling, fluorescent labeling, and enzymatic labeling. Many techniques are known for detecting sequences, such as the Illumina NextSeq platform using cycle-end sequencing.

[0064] Machine learning models Once sequence data representing the polypeptide sequences of one or more tumor-specific neoantigens is obtained, this sequence data can be input into a machine learning platform (i.e., a model) along with the target MHC molecule. The machine learning platform can generate one or more numerical probability scores that predict whether or not one or more tumor-specific neoantigens are immunogenic (e.g., whether or not they induce an immune response in the subject).

[0065] MHC molecules transport peptides and present them on the cell surface. MHC molecules are classified into class I and class II. MHC class I molecules are present on the surface of almost all cells in the body, including most tumor cells. MHC class I proteins are typically loaded with antigens derived from endogenous proteins or pathogens present within the cell and presented to cytotoxic T lymphocytes (i.e., CD8+). MHC class I molecules can include HLA-A, HLA-B, or HLA-C. Class II MHC molecules are present only on dendritic cells, B lymphocytes, macrophages, and other antigen-presenting cells. They primarily present peptides and are processed by helper T (Th) cells (i.e., CD4+) from an external antigen source, i.e., outside the cell. MHC class II molecules can include HLA-DPA1, HLA-DPB1, HLA-DQA1, HLA-DQB1, HLA-DRA, and HLA-DRB1. In some cases, MHC class II molecules may also be expressed on cancer cells.

[0066] MHC class I molecules and / or MHC class II molecules can be input into the machine learning platform. Typically, either MHC class I molecules or MHC class II molecules are input into the machine learning platform. In some embodiments, MHC class I molecules are input into the machine learning platform. In other embodiments, MHC class II molecules are input into the machine learning platform. In some embodiments, the MHC class I machine learning platform may be trained on MHC class I training data. In some embodiments, the MHC class II machine learning platform may be trained on MHC class II training data. In some embodiments, the same machine learning platform may be trained on training data for both MHC class I and MHC class II. In some embodiments, the machine learning platform may include MHC class I models and MHC class II modes.

[0067] MHC class I molecules bind to short-chain peptides. MHC class I molecules can typically accommodate peptides that are about 8 to 10 amino acids long. In one embodiment, the sequence data encoding one or more tumor-specific neoantigens is a short-chain peptide that is about 8 to 10 amino acids long. MHC class II molecules bind to longer-chain peptides. MHC class II molecules can typically accommodate peptides that are about 13 to 25 amino acids long. In one embodiment, the sequence data encoding one or more tumor-specific neoantigens is a long-chain peptide that is about 13 to 25 amino acids long.

[0068] Sequence data encoding one or more tumor-specific neoantigens may have lengths of approximately 5 amino acids, 6 amino acids, 7 amino acids, 8 amino acids, 9 amino acids, 10 amino acids, 11 amino acids, 12 amino acids, 13 amino acids, 14 amino acids, 15 amino acids, 16 amino acids, 17 amino acids, 18 amino acids, 19 amino acids, 20 amino acids, 21 amino acids, 22 amino acids, 23 amino acids, 24 amino acids, 25 amino acids, 26 amino acids, 27 amino acids, 28 amino acids, 29 amino acids, or 30 amino acids.

[0069] The machine learning platform can predict the likelihood that one or more tumor-specific neoantigens are immunogenic (e.g., they trigger an immune response).

[0070] Immunogenic tumor-specific neoantigens are not expressed in normal tissues. They can be presented to CD4+ and CD8+ T cells by antigen-presenting cells and trigger an immune response. In embodiments, the immune response in a subject induced by one or more tumor-specific neoantigens includes the presentation of one or more tumor-specific neoantigens to the surface of tumor cells. More specifically, the immune response in a subject induced by one or more tumor-specific neoantigens includes the presentation of one or more tumor-specific neoantigens by one or more MHC molecules on tumor cells. The immune response induced by one or more tumor-specific neoantigens is expected to be a T cell-mediated response. The immune response in a subject induced by one or more tumor-specific neoantigens may involve one or more tumor-specific neoantigens that can be presented to T cells by antigen-presenting cells such as dendritic cells. Preferably, one or more tumor-specific neoantigens can activate CD8+ T cells and / or CD4+ T cells.

[0071] In some embodiments, the machine learning platform can predict the likelihood that one or more tumor-specific neoantigens will activate CD8+ T cells. In other embodiments, the machine learning platform can predict the likelihood that one or more tumor-specific neoantigens will activate CD4+ T cells. In some cases, the machine learning platform can predict the antibody titer that one or more tumor-specific neoantigens can induce. In other examples, the machine learning platform can predict the frequency of CD8+ activation by one or more tumor-specific neoantigens.

[0072] A machine learning platform may include a model trained on training data. Training data can be obtained from different groups of subjects. Training data may include data from healthy subjects as well as subjects with cancer. The training data may include various data that can be used to generate probability scores indicating whether one or more tumor-specific neoantigens induce an immune response in subjects. Exemplary training data may include data representing nucleotide or polypeptide sequences from normal tissues and / or cells, data representing nucleotide or polypeptide sequences from tumor tissues, data representing MHC peptide dome sequences from normal and tumor tissues, peptide-MHC binding affinity measurements, or combinations thereof. Reference data may further include mass spectrometry data, DNA sequence data, RNA sequence data, clinical data from healthy individuals or cancer patients, cytokine profiling data, T-cell toxicity assay data, peptide-MHC monomer or multimer data, synthetic proteins manipulated to express a given MHC allele, normal and tumor human cell lines, fresh and frozen primary samples, and proteomics data from single-allele cell lines exposed to T-cell assays.

[0073] In some exemplary embodiments, binding affinity predictions for various samples, including a "weak" label for samples with unknown binding affinity predictions, may be extracted and added to the binding affinity training dataset. Samples whose binding affinity predictions exceed a predetermined threshold are removed, leaving a refined dataset for use in further training processes.

[0074] Machine learning platforms can be supervised, unsupervised, or semi-supervised. Using sequence-based approaches, machine learning platforms can generate numerical probabilities that one or more tumor-specific neoantigens can trigger an immune response (e.g., trigger a high / low antibody response or a CD8+ response). Sequence-based predictions include artificial neural networks (deep or other), support vector machines, k-nearest neighbors, logistic multiple network constraint regression (LogMiNeR), regression trees, random forests, adaboost, XGBoost, or hidden Markov models. These platforms require training datasets containing known MHC-binding peptides.

[0075] According to some embodiments, masked language modeling may be implemented in the pre-training phase, where a determined subset of peptide sequences is masked via a tokenization process. The classifier can then predict the original token values ​​based on the existing, unmasked tokens.

[0076] In another example, the next peptide in a sequence is determined according to a pre-training process where the input sequence may be a concatenation of two peptide sequences instead of peptide and allele sequences in the main training phase. The two peptide sequences are separated using special separation tokens, and each segment may have a different segment index and embedding. The segment sequences are provided as input to a network, and each token may indicate whether it belongs to the first sequence, the second sequence, or is a special token. The tokens can be used to train a classifier to predict whether the second peptide is the next peptide to appear in the protein. The peptides may be provided by two consecutive peptides of the same length from a human protein, or they may be randomly sampled from different proteins.

[0077] Numerous predictive programs are used to predict whether tumor-specific neoantigens can be presented on MHC molecules and trigger an immune response. Examples of prediction programs include, for instance, HLAminer (Warren et al., Genome Med., 4:95 (2012); HLA type predicted by oriented the assembly of shotgun sequence data and comparing it with the reference allele sequence database), VariantEffect Predictor Tool (McLaren et al., Genome Biol., 17:122 (2016)), NetMHCpan (Andreatta et al., Bioinformatics., 32:511-517 (2016); sequence comparison method based on artificial neural network, and predict the affinity of peptide-MHC-I type molecule), UCSC browser (Kent et al., Genome Res., 12:996-1006 (2002)), CloudNeo pipeline (Bais et al., Bioinformatics, 33:3110-2 (2017)), and OptiType (Szolek et al.). al., Bioinformatics, 30:3310-316 (2014)), ATHLATES (Liu C et al., Nucleic Acids Res. 41:e142 (2013)), pVAC-Seq (Hundal et al., Genome Med. 8:11 (2016)), MuPeXI (Bjerregaard et al., Cancer Immunol Immunother., 66:1123-30 (2017)), Strelka (Saunders et al., Bioinformatics. 28:1811-7 (2012)), Strelka2 (Kim et al., Nat Methods.2018;15:591-4.), VarScan2 (Koboldt et al., Genome Res., 22:568-76 (2012)), Somaticseq (Fang L et al., Genome Biol., 16:197 (2015)), SMMPMBEC (Kim et al., BMC Bioinformatics., 10:394 (2009)), NeoPredPipe (Schenck RO, BMC Bioinformatics., 20:264 (2019)), Weka (Witten et al., Data mining: practical machine-learning tools and techniques. 4. th Examples include *Ed. Elsevier*, ISBN: 97801280435578 (eBook) (2017), or *Orange* (Demsar et al., *Orange: Data Mining Toolbox in Python*, J. Mach Learn Res., 14:2349-2353 (2013)). Any known predictive program can be used as a machine learning platform to generate numerical probability scores indicating whether or not neoantigens trigger an immune response.

[0078] Depending on the machine learning platform used, additional filters can be applied to prioritize tumor-specific neoantigen candidates, including the removal of hypothesis-based (Riken) proteins, the use of antigen processing algorithms to remove epitopes that are unlikely to be proteolytically generated by constitutive or immunoproteasomes, and prioritizing neoantigens with higher predicted binding affinity than their corresponding wild-type sequences.

[0079] The numerical probability score may be a number between 0 and 1. In this embodiment, the numerical probability score may be 0, 0.0001, 0.0002, 0.0003, 0.0004, 0.0005, 0.0006, 0.0007, 0.0008, 0.0009, 0.001, 0.002, 0.003, 0.004, 0.005, 0.006, 0.007, 0.008, 0.009, 0.01, 0.02, 0.03, 0.04, 0.05, 0.06, 0.07, 0.08, 0.09, 0.10, 0.20, 0.30, 0.40, 0.50, 0.60, 0.70, 0.80, 0.90, or 1. Tumor-specific neoantigens with higher numerical probability scores compared to those with lower numerical probability scores indicate that they induce a greater immune response in subjects and are therefore likely to be suitable candidates for immunogenic compositions. For example, a tumor-specific neoantigen with a numerical probability score of 1 is more likely to induce a greater immune response in subjects than a tumor-specific neoantigen with a numerical probability score of 0.05. Similarly, a tumor-specific neoantigen with a numerical probability score of 0.5 is more likely to induce a greater immune response in subjects than a tumor-specific neoantigen with a numerical probability score of 0.1.

[0080] A higher numerical probability score is preferable to a lower numerical probability score. Preferably, tumor-specific neoantigens have a numerical probability score of at least 0.8, 0.81, 0.82, 0.83, 0.84, 0.85, 0.86, 0.87, 0.88, 0.89, 0.9, 0.95, 0.96, 0.97, 0.98, 0.99, or 1, which indicates a high probability of inducing an immune response in the subject.

[0081] A higher probability score is preferable, but even a low probability score may indicate that tumor-specific neoantigens can still induce a sufficient immune response and are likely to be suitable candidates.

[0082] In some cases, the machine learning platform described herein can also predict the likelihood that one or more tumor-specific neoantigens will be presented by MHC molecules on tumor cells. The machine learning platform can predict the likelihood that one or more tumor-specific neoantigens will be presented by MHC class I molecules or MHC class II molecules.

[0083] A method for selecting one or more tumor-specific neoantigens may further include the step of measuring the affinity of one or more tumor-specific neoantigens to MHC molecules in a subject in silico. Tumor-specific neoantigens with binding affinity to MHC molecules of less than approximately 1000 nM indicate that one or more tumor-specific neoantigens may be suitable for an immunogenic composition. Tumor-specific neoantigens with binding affinity to MHC molecules of less than approximately 500 nM, less than approximately 400 nM, less than approximately 300 nM, less than approximately 200 nM, less than approximately 100 nM, or less than approximately 50 nM also indicate that one or more tumor-specific neoantigens may be suitable for an immunogenic composition. The immunogenicity of tumor-specific neoantigens can be predicted from the affinity of one or more tumor-specific neoantigens to MHC molecules in a subject. Alternatively, the median affinity is an effective method for predicting the immunogenicity of tumor-specific neoantigens. The median affinity can be calculated using epitope prediction algorithms such as NetMHCpan, ANN, SMM, and SMMPMBEC.

[0084] RNA expression of one or more tumor-specific neoantigens is also quantified. Quantifying the RNA expression of one or more tumor-specific neoantigens identifies one or more neoantigens that induce an immune response in subjects. Various methods exist for measuring RNA expression. Known techniques for measuring RNA expression include RNA-seq, in situ hybridization (e.g., FISH), Northern blotting, DNA microarrays, tiling arrays, and quantitative polymerase chain reaction (qPCR). RNA expression can also be quantified using other known techniques in the art. RNA includes messenger RNA (mRNA), short interfering RNA (siRNA), microRNA (miRNA), circular RNA (circRNA), transfer RNA (tRNA), ribosomal RNA (rRNA), small nucleolar RNA (snRNA), Piwi-interacting RNA (piRNA), long non-coding RNA (long ncRNA), subgenomic RNA (sgRNA), RNA from embedded or non-embedded viruses, or other RNA. Preferably, mRNA expression is measured.

[0085] This technology can further reduce the likelihood of selecting tumor-specific neoantigens that may induce an autoimmune response in normal tissue. Tumor-specific neoantigens with sequences similar to normal antigens are expected to potentially induce an autoimmune response in normal tissue. For example, tumor-specific neoantigens that are at least 90%, at least 91%, at least 92%, at least 93%, at least 94%, at least 95%, at least 96%, at least 97%, at least 98%, or at least 99% similar to normal antigens may induce an autoimmune response. Tumor-specific neoantigens that are expected to induce an autoimmune response are not preferred as immunogenic compositions. Tumor-specific neoantigens that are expected to induce an autoimmune response are generally not selected as immunogenic compositions. This method may further include measuring the ability of one or more tumor-specific neoantigens to induce immune tolerance. Tumor-specific neoantigens that are expected to induce immune tolerance are not preferred as immunogenic compositions. Tumor-specific neoantigens that are expected to induce immune tolerance are not preferred as immunogenic compositions.

[0086] Finally, one or more tumor-specific neoantigens are selected based on the tumor-specific score for formulation of a target-specific immunogenic composition. In one embodiment, at least about 1, at least about 2, at least about 3, at least about 4, at least about 5, at least about 6, at least about 7, at least about 8, at least about 9, at least about 10, at least about 11, at least about 12, at least about 13, at least about 14, at least about 15, at least about 16, at least about 17, at least about 18, at least about 19, at least about 20, at least about 25, at least about 30, at least about 35, at least about 40, and at least about 50 or more tumor-specific neoantigens are selected as the immunogenic composition. Typically, at least about 10 tumor-specific neoantigens are selected. In other examples, at least about 20 tumor-specific neoantigens are selected.

[0087] II. Treatment method This disclosure also relates to a method for treating cancer in a subject requiring it, comprising administering a personalized immunogenic composition comprising one or more tumor-specific neoantigens selected using the method described herein.

[0088] Cancer is any solid tumor or any hematological malignancy. The methods disclosed herein are preferably suited to solid tumors. The tumor may be a primary tumor (e.g., a tumor located at the original site where the tumor first originated). Solid tumors include, but are not limited to, breast cancer tumors, ovarian cancer tumors, prostate cancer tumors, lung cancer tumors, kidney cancer tumors, gastric cancer tumors, testicular cancer tumors, head and neck cancer tumors, pancreatic cancer tumors, brain cancer tumors, melanoma tumors, and the like. Hematological malignancies include, but are not limited to, tumors derived from lymphoma (e.g., B-cell lymphoma) or leukemia (e.g., acute myeloid leukemia, chronic myeloid leukemia, chronic lymphocytic leukemia, and T-cell lymphocytic leukemia).

[0089] The methods disclosed herein can be used for appropriate cancerous tumors, including hematological malignancies, solid tumors, sarcomas, carcinomas, other solid tumors, and non-solid tumors. Examples of appropriate cancers include, for example, acute lymphoblastic leukemia (ALL), acute myeloid leukemia (AML), adrenocortical carcinoma, anal cancer, appendiceal cancer, astrocytoma, basal cell carcinoma, brain tumor, cholangiocarcinoma, bladder cancer, bone cancer, breast cancer, bronchial tumor, cancer of unknown primary origin, cardiac tumor, cervical cancer, chordoma, colon cancer, colorectal cancer, craniopharyngioma, ductal carcinoma, fetal tumor, endometrial cancer, ependymoma, esophageal cancer, sensory neuroblastoma, and fibrous histiocytoma. Tumor, Ewing's sarcoma, eye cancer, germ cell tumor, gallbladder cancer, stomach cancer, gastrointestinal carcinoid tumor, gastrointestinal stromal tumor, chorioblastic disease of pregnancy, glioma, head and neck cancer, hepatocellular carcinoma, histiocytosis, Hodgkin lymphoma, hypopharyngeal cancer, intraocular melanoma, islet cell tumor, Kaposi's sarcoma, kidney cancer, Langerhans cell histiocytosis, laryngeal cancer, lip and oral cancer, liver cancer, carcinoma in situ, lung cancer, macroglobulinemia, malignant fibrous histiocytoma, black Cancer, Merkel cell carcinoma, mesothelioma, metastatic squamous cell carcinoma with silent primary midline tract cancer involving the NUT gene, oral cancer, multiple endocrine neoplasia syndrome, multiple myeloma, mycosis fungoides, myelodysplastic syndrome, myelodysplastic / myeloproliferative neoplasm, nasal cavity and paranasal sinus cancer, nasopharyngeal cancer, neuroblastoma, non-small cell lung cancer, oropharyngeal cancer, osteosarcoma, ovarian cancer, pancreatic cancer, papillomatosis, paraganglioma, parathyroid cancer, penile cancer, pharyngeal cancer, pheochromocytoma Examples include pituitary tumors, pleuropulmonary blastoma, primary central nervous system lymphoma, prostate cancer, rectal cancer, renal cell carcinoma, renal pelvis and ureteral cancer, retinoblastoma, rhabdoid tumor, salivary gland cancer, Sézary syndrome, skin cancer, small cell lung cancer, small intestine cancer, soft tissue sarcoma, spinal cord tumor, gastric cancer, T-cell lymphoma, teratoma, testicular cancer, pharyngeal cancer, thymoma, thymic cancer, thyroid cancer, urethral cancer, uterine cancer, vaginal cancer, vulvar cancer, and Wilms' tumor. Preferably, the cancers are melanoma, breast cancer, ovarian cancer, prostate cancer, kidney cancer, gastric cancer, colon cancer, testicular cancer, head and neck cancer, pancreatic cancer, brain cancer, B-cell lymphoma, acute myeloid leukemia, chronic myeloid neoplasm, leukemia, chronic lymphocytic leukemia, T-cell lymphocytic leukemia, bladder cancer, or lung cancer. Melanoma is of particular interest. There is particular interest in breast cancer, lung cancer, and bladder cancer.

[0090] Immunogenic compositions stimulate the immune system of subjects, particularly the response of specific CD8+ T cells or CD4+ T cells. Interferon-gamma produced by CD8+ and T helper CD4+ cells regulates PD-L1 expression. PD-L1 expression in tumor cells is upregulated when attacked by T cells. Therefore, tumor vaccines may induce the production of specific T cells and simultaneously upregulate PD-L1 expression, which may limit the efficacy of the immunogenic composition. Furthermore, when the immune system is activated, the expression of the T cell surface reporter CTLA-4 increases accordingly. CTLA-4 binds to ligand B7-1 / B7-2 on antigen-presenting cells and exerts an immunosuppressive effect. Therefore, in some cases, subjects may be further administered immunosuppressants or immunostimulants, such as checkpoint inhibitors. Checkpoint inhibitors include, but are not limited to, anti-CTL4-A antibodies, anti-PD-1 antibodies, and anti-PD-L1 antibodies. These checkpoint inhibitors bind to immune checkpoint proteins on T cells, eliminating the inhibition of T cell function by tumor cells. Blocking CTLA-4 or PD-L1 with antibodies may enhance the patient's immune response to cancer cells. CTLA-4 has been shown to be effective when followed in a vaccination program.

[0091] An immunogenic composition containing one or more tumor-specific neoantigens can be administered to subjects diagnosed with cancer, subjects already suffering from cancer, subjects with recurrent cancer (i.e., recurrence), or subjects at risk of developing cancer. An immunogenic composition containing one or more tumor-specific neoantigens can be administered to subjects who are resistant to other forms of cancer treatment (e.g., chemotherapy, immunotherapy, radiation). An immunogenic composition containing one or more tumor-specific neoantigens can be administered to subjects before other standard cancer treatments (e.g., chemotherapy, immunotherapy, radiation). An immunogenic composition containing one or more tumor-specific neoantigens can be administered to subjects in combination with, after, or in conjunction with, other standard cancer treatments (e.g., chemotherapy, immunotherapy, radiation).

[0092] The subjects may be humans, dogs, cats, horses, or animals for which a tumor-specific response is desired.

[0093] The immunogenic composition is administered to a subject in an amount sufficient to induce an immune response to tumor-specific neoantigens and destroy or at least partially block symptoms and / or complications. In embodiments, the immunogenic composition can provide a long-lasting immune response. A long-lasting immune response can be established by administering additional immunogenic compositions to the subject. The immune response to the immunogenic composition can be prolonged by administering additional immunization doses to the subject. In embodiments, at least one, at least two, and at least three or more additional immunization doses may be administered to alleviate cancer. The first additional immunization dose can increase the immune response by at least 50%, at least 100%, at least 200%, at least 300%, at least 400%, at least 500%, or at least 1000%. The second additional immunization dose can increase the immune response by at least 50%, at least 100%, at least 200%, at least 300%, at least 400%, at least 500%, or at least 1000%. A third additional immune dose can increase the immune response by at least 50%, at least 100%, at least 200%, at least 300%, at least 400%, at least 500%, or at least 1000%.

[0094] The amount sufficient to induce an immune response is defined as the "therapeutic effective dose." This effective dose depends, for example, on the composition, method of administration, stage and severity of the disease being treated, the patient's weight and general health, and the judgment of the prescribing physician. It should be noted that immunogenic compositions are generally used in serious medical conditions, i.e., life-threatening or potentially life-threatening situations, particularly when cancer has metastasized. In such cases, considering the minimization of foreign substances and the relatively non-toxic nature of neoantigens, the attending physician may, and may even deem, that a substantially excessive dose of these immunogenic compositions is possible and desirable.

[0095] An immunogenic composition comprising one or more tumor-specific neoantigens can be administered to a target alone or in combination with other therapeutic agents. The therapeutic agent may be, for example, a chemotherapeutic agent, radiotherapy, or immunotherapy. Appropriate therapeutic treatment may be given for a specific cancer. Exemplary chemotherapeutic agents include aldesleukin, altretamine, amifostine, asparaginase, bleomycin, capecitabine, carboplatin, carmustine, cladribine, cisapride, cisplatin, cyclophosphamide, cytarabine, dacarbazine (DTIC), dactinomycin, docetaxel, doxorubicin, dronabinol, epoetin α, etoposide, filgrastim, fludarabine, fluorouracil, gemcitabine, granisetron, hydroxyurea, idarubicin, ifosfamide, Examples of such therapies include, but are not limited to, interferon-alpha, irinotecan, lansoprazole, levamisole, leucovorin, megestrol, mesna, methotrexate, metoclopramide, mitomycin, mitotane, mitoxantrone, omeprazole, ondansetron, paclitaxel (Taxol®), pilocarpine, prochloroperazine, rituximab, tamoxifen, Taxol, topotecan hydrochloride, trastuzumab, vinblastine, vincristine, and vinorelbine tartrate. The subjects may also be administered small molecule or targeted therapies (e.g., kinase inhibitors). Furthermore, the subjects may be administered anti-CTLA antibodies, anti-PD-1 antibodies, or anti-PD-L1 antibodies. Blocking CTLA-4 or PD-L1 with antibodies may enhance the patient's immune response to cancer cells.

[0096] III. Immunogenic composition The present invention further relates to personalized (i.e., target-specific) immunogenic compositions (e.g., cancer vaccines) comprising one or more tumor-specific antigens selected using the methods described herein. Such immunogenic compositions can be formulated according to standard procedures of the Art. The immunogenic compositions can evoke a specific immune response.

[0097] Immunogenic compositions can be formulated to adjust the selection and number of tumor-specific neoantigens to suit the specific cancer of the subject. For example, the selection of tumor-specific neoantigens may depend on the specific type of cancer, the stage of the cancer, the subject's immune status, and the subject's MHC type.

[0098] The immunogenic composition may contain at least 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 37, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50 or more tumor-specific neoantigens. The immunogenic composition may contain about 10 to 20 tumor-specific neoantigens, about 10 to 30 tumor-specific neoantigens, about 10 to 40 tumor-specific neoantigens, about 10 to 50 tumor-specific neoantigens, about 10 to 60 tumor-specific neoantigens, about 10 to 70 tumor-specific neoantigens, about 10 to 80 tumor-specific neoantigens, about 10 to 90 tumor-specific neoantigens, or about 10 to 100 tumor-specific neoantigens. Preferably, the immunogenic composition contains at least about 10 tumor-specific neoantigens. Furthermore, an immunogenic composition containing at least about 20 tumor-specific neoantigens is preferred.

[0099] Immunogenic compositions may further contain natural or synthetic antigens. Natural or synthetic antigens may enhance the immune response. Examples of natural or synthetic antigens include, but are not limited to, pan-DR epitopes (PADREs) and tetanus toxin antigens.

[0100] The immunogenic composition may be in any form, such as a synthetic long-chain peptide, RNA, DNA, cells, dendritic cells, nucleotide sequences, polypeptide sequences, plasmids, or vectors.

[0101] Tumor-specific neoantigens can also be incorporated into viral vector-based vaccine platforms such as vaccinia, fowlpox, self-replicating alphavirus, marabavirus, adenovirus (see, e.g., Tatsis et al., Molecular Therapy, 10:616-629 (2004)), or lentiviruses including, but not limited to, second-generation, third-generation, or hybrid second / third-generation lentiviruses, or recombinant lentiviruses of any generation designed to target specific cell types or receptors (see, e.g., Hu et al., Immunol Rev., 239(1): 45-61 (2011), Sakma et al, Biochem J., 443(3):603-18 (2012)). Depending on the packaging capabilities of the viral vector-based vaccine platforms described above, this approach can deliver one or more nucleotide sequences encoding one or more tumor-specific neoantigen peptides. The sequence may be adjacent to a non-mutant sequence, separated by a linker, or preceded by one or more sequences targeting an intracellular compartment (see, for example, Gros et al., Nat Med., 22 (4):433-8 (2016), Stronen et al., Science., 352(6291): 1337-1341 (2016), Lu et al., Clin Cancer Res., 20(13):3401-3410 (2014)). Upon introduction into the host, infected cells express one or more tumor-specific neoantigens, thereby inducing a host immune response (e.g., CD8+ or CD4+) to one or more tumor-specific neoantigens. Useful vaccinia vectors and methods for immunization protocols are described, for example, in U.S. Patent No. 4,722,848. Another vector is BCG (Bacille Calmette Guerin). The BCG vector is described in Stover et al. (Nature 351:456-460 (1991)).A wide variety of other vaccine vectors useful for the therapeutic administration or immunization of neoantigens, which are obvious to those skilled in the art from the description herein, can also be used.

[0102] Immunogenic compositions may contain personalized ingredients to meet the individual needs of a specific subject.

[0103] The immunogenic compositions described herein may further contain an adjuvant. The adjuvant is a substance that, when mixed with the immunogenic composition, increases, or otherwise enhances and / or boosts, the immune response to tumor-specific neoantigens, but does not produce an immune response to tumor-specific neoantigens when administered alone. It is preferable that the adjuvant generates an immune response to neoantigens without causing allergies or other adverse reactions. In this specification, it is intended that the immunogenic composition may be administered before, concurrently with, or after the administration of the immunogenic composition.

[0104] Adjuvants can enhance the immune response through some mechanism, including, for example, lymphocyte recruitment, B cell and / or T cell stimulation, and macrophage stimulation. When the immunogenic composition of the present invention contains an adjuvant, or is administered together with one or more adjuvants, the adjuvants that can be used include, but are not limited to, mineral salt adjuvants or mineral salt gel adjuvants, particle adjuvants, microparticle adjuvants, mucosal adjuvants, and immunostimulant adjuvants. Examples of adjuvants include aluminum salts (alum) (aluminum hydroxide, aluminum phosphate, aluminum sulfate, etc.), 3-deoxylated monophosphoryl lipid A (MPL) (see GB2220211), MF59 (Novartis), AS03 (Glaxo SmithKline), AS04 (Glaxo SmithKline), polysorbate 80 (Tween 80; ICL Americas), imidazopyridine compounds (see international application number PCT / US2007 / 064857, published as international publication number WO2007 / 109812), imidazoquinoxaline compounds (see international application number PCT / US2007 / 064858, published as international publication number WO2007 / 109813), and saponins such as QS21 (Kensil et al, in Vaccine Design: The Subunit and Adjuvant Approach (eds.)). Examples include, but are not limited to, Powell & Newman (Plenum Press, NY, 1995) and U.S. Patent No. 5,057,540. In some embodiments, the adjuvant is a Freund's adjuvant (complete or incomplete). Other adjuvants are oil-in-water emulsions (such as squalene or peanut oil) and may be combined with immunostimulants such as monophosphoryl lipid A as needed (see Stoute et al, N. Engl. J. Med. 336, 86-91 (1997)).

[0105] CpG immunostimulatory oligonucleotides have also been reported to enhance the adjuvant effect in vaccine settings. Other TLR-binding molecules, such as RNAs that bind to TLR7, TLR8, and / or TLR9, can also be used.

[0106] Other examples of useful adjuvants include, but are not limited to, chemically modified CpG (e.g., CpR, Idera), poly(I:C) (e.g., polyi:CI2U), polyICLC, non-CpG bacterial DNA or RNA, immunoactive small molecules and antibodies, such as cyclophosphamide, sunitomib, bevacizumab, Celebrex (celecoxib), NCX-4016, sildenafil, tadalafil, vardenafil, sorafinib, XL-999, CP-547632, pazopam, ZD2171, AZD2171, ipilimumab, tremelimumab, SC58175, etc., that act therapeutically and / or as adjuvants. In embodiments, polyICLC is a preferred adjuvant.

[0107] An immunogenic composition may contain one or more tumor-specific neoantigens described herein, either alone or in combination with a pharmaceutically acceptable carrier. A suspension or dispersion of one or more tumor-specific neoantigens, particularly an isotonic aqueous suspension, dispersion, or amphiphilic solvent, may be used. The immunogenic composition may be sterilized and / or contain excipients, such as preservatives, stabilizers, wetting agents and / or emulsifiers, solubilizers, salts and / or buffers for adjusting osmotic pressure, and may be prepared by known methods, such as conventional dispersion and suspension processes. In certain embodiments, such dispersions or suspensions may contain viscosity modifiers. Suspensions or dispersions are stored at a temperature of about 2°C to 8°C, and may be frozen and thawed immediately before use if longer storage is preferred. For injection, the vaccine or immunogenic preparation may be formulated in an aqueous solution, preferably in a physiologically suitable buffer such as Hanks' solution, Ringer's solution, or physiological saline buffer. This solution may contain compounding agents such as suspending agents, stabilizers and / or dispersants.

[0108] In certain embodiments, the compositions described herein further comprise a preservative, such as the mercury derivative thimerosal. In certain embodiments, the pharmaceutical compositions described herein comprise 0.001% to 0.01% thimerosal. In other embodiments, the pharmaceutical compositions described herein do not contain a preservative.

[0109] Excipients can exist independently of adjuvants. The functions of excipients include, for example, increasing the molecular weight of an immunogenic composition, increasing its activity or immunogenicity, conferring stability, increasing biological activity, or increasing its serum half-life. Excipients can also be used to assist in the presentation of one or more tumor-specific neoantigens to T cells (e.g., CD4+ or CD8+ T cells). Excipients may be, but are not limited to, carrier proteins such as keyhole limpet hemocyanin, serum proteins such as transferrin, bovine serum albumin, human serum albumin, thyroglobulin or ovalbumin, immunoglobulins, or hormones such as insulin or palmitic acid. In the case of human immunization, the carrier is generally a physiologically acceptable carrier that is well-received and safe in the human body. Alternatively, the carrier may be dextran, such as cepharose.

[0110] Cytotoxic T cells recognize antigens not as intact foreign antigens themselves, but in the form of peptides bound to MHC molecules. MHC molecules themselves are present on the cell surface of antigen-presenting cells. Therefore, activation of cytotoxic T cells is possible when a trimer complex of peptide antigen, MHC molecule, and antigen-presenting cell (APC) is present. Beyond using one or more tumor-specific antigens to activate cytotoxic T cells, adding additional APCs containing each MHC molecule may enhance the immune response. Therefore, in some embodiments, the immunogenic composition further comprises at least one APC.

[0111] The immunogenic composition may contain an acceptable carrier (e.g., an aqueous carrier). Various aqueous carriers can be used, such as water, buffer water, 0.9% physiological saline, 0.3% glycine, hyaluronic acid, etc. These compositions may be sterilized by conventional and well-known sterilization techniques or aseptically filtered. The resulting aqueous solution may be packaged for immediate use or freeze-dried, and the freeze-dried preparation may be combined with a sterile solution before administration. The composition may contain pharmaceutically acceptable adjuncts necessary to approximate physiological conditions, such as pH adjusters, buffers, tonics, and humectants, such as sodium acetate, sodium lactate, sodium chloride, potassium chloride, calcium chloride, sorbitan monolaurate, and triethanolamine oleate.

[0112] Neoantigens can also be administered via liposomes that target specific cell tissues, such as lymphoid tissue. Liposomes also help to extend the half-life. Liposomes include emulsions, foams, micelles, insoluble monolayers, liquid crystals, phospholipid dispersions, and lamellar layers. In these formulations, the neoantigen to be delivered is incorporated as part of the liposome, either alone or together with molecules that bind to receptors widely present on lymphoid cells, such as monoclonal antibodies that bind to the CD45 antigen, or together with other therapeutic or immunogenic compositions. Thus, liposomes filled with the desired neoantigen can be directed to a site on lymphoid cells, where the liposomes deliver the selected immunogenic composition. Liposomes can be formed from standard vesicle-forming lipids, generally containing neutral or negatively charged phospholipids and sterols such as cholesterol. Lipids are generally selected considering, for example, liposome size, acid instability, and liposome stability in the bloodstream. Various methods can be used to prepare liposomes, including, for example, the methods described in Szoka et al., An. Rev. Biophys. Bioeng. 9;467 (1980), and U.S. Patents No. 4,235,871, No. 4,501,728, No. 4,501,728, No. 4,837,028, and No. 5,019,369.

[0113] To target immune cells, the ligands incorporated into the liposomes may include, for example, antibodies or fragments thereof that are specific to the cell surface determinants of desired immune system cells. The liposome suspension can be administered intravenously, topically, or otherwise, in doses that vary depending on the method of administration, the peptides delivered, and the stage of the disease being treated.

[0114] Another method for targeting immune cells is to incorporate components of an immunogenic composition, such as an antigen (i.e., a tumor-specific novel antigen), ligand, or adjuvant (e.g., a TLR), into polylactic glycolic acid copolymer microspheres. These polylactic glycolic acid copolymer microspheres can then capture components of the immunogenic composition as endosome delivery devices.

[0115] Nucleic acids encoding tumor-specific neoantigens as described herein may also be administered to patients for therapeutic or immunization purposes. Many methods have been employed to deliver nucleic acids to patients. For example, nucleic acids may be delivered directly as "naked DNA." This approach is described, for example, in Wolff et al., Science 247: 1465-1468 (1990), and in U.S. Patents 5,580,859 and 5,589,466. Nucleic acids may also be delivered using ballistic delivery methods, for example, as described in U.S. Patent 5,204,253. Particles consisting solely of DNA may also be delivered. Alternatively, DNA may be attached to particles such as gold particles. Approaches to nucleic acid sequence delivery include viral vectors, mRNA vectors, and DNA vectors, with or without the use of electroporation. Nucleic acids may also be delivered by forming complexes with cationic compounds, such as cationic lipids.

[0116] The immunogenic compositions provided herein can be administered to subjects by methods including, but not limited to, oral, intradermal, intratumoral, intramuscular, intraperitoneal, intravenous, topical, subcutaneous, percutaneous, nasal, inhalation, and scar formation (e.g., by scratching the uppermost layer of skin using a bifurcated needle). The immunogenic compositions can be administered to tumor sites to induce a local immune response against the tumor.

[0117] The dosage of one or more tumor-specific neoantigens depends on the type of composition, the subject's age, weight, body surface area, individual condition, individual pharmacokinetic data, and mode of administration.

[0118] Also disclosed herein are methods for producing immunogenic compositions comprising one or more tumor-specific neoantigens selected by the methods disclosed herein. The immunogenic compositions described herein can be produced using methods known in the art. For example, a method for producing tumor-specific neoantigens or vectors (e.g., a vector comprising at least one sequence encoding one or more tumor-specific neoantigens) disclosed herein comprises culturing host cells under conditions suitable for expressing the neoantigen or vector. The host cells contain at least one polynucleotide encoding the neoantigen or vector and for purifying the neoantigen or vector. Standard purification methods include chromatography, electrophoresis, immunology, precipitation, dialysis, filtration, concentration, and chromatofocusing techniques.

[0119] The host cells may include Chinese hamster ovary (CHO) cells, NS0 cells, yeast, or HEK293 cells. The host cells may be transformed with one or more polynucleotides comprising at least one nucleic acid sequence encoding one or more tumor-specific neoantigens or vectors disclosed herein. In certain embodiments, the isolated polynucleotide is cDNA.

[0120] IV. Samples A method disclosed herein includes ranking one or more tumor-specific neoantigens derived from a tumor. The method for ranking one or more tumor-specific neoantigens includes obtaining sequence data derived from the tumor. Such sequence data can be obtained from a tumor sample of a subject. The tumor sample can be obtained from a tumor biopsy.

[0121] Tumor samples can be obtained from human or non-human subjects. Preferably, tumor samples are obtained from humans. Tumor samples can be obtained from a variety of biological sources, including cancerous tumors. Tumors may originate from the tumor site or from circulating tumor cells in the blood. Examples of samples include, but are not limited to, body fluids, tissue biopsies, blood samples, serum plasma, stool, and skin samples. The source of the sample may be a solid tissue sample, such as a tumor tissue biopsy. Tissue biopsy samples may be obtained, for example, from the lung, prostate, colon, skin, breast tissue, or lymph nodes. Samples may also be bone marrow samples, including bone marrow aspirates and bone marrow biopsies. Samples may also be liquid biopsy samples, such as circulating tumor cells, cell-free circulating tumor DNA, or exosomes. Blood samples may be whole blood, partially purified blood, or a portion of whole or partially purified blood, such as peripheral blood mononuclear cells (PBMCs).

[0122] The tumor samples described herein may be obtained directly from the subject, derived from the subject, or derived from a sample obtained from the subject, such as cultured cells derived from a body fluid or tissue sample. Tumor biopsy samples may be fresh samples. Fresh samples can be fixed after removal from the subject using a known fixative (e.g., formalin, Zenker fixative, or B-5 fixative). Tumor biopsy samples may also be preserved samples, such as frozen samples or cryopreserved samples of cells obtained directly from the subject or derived from cells obtained from the subject. Preferably, tumor samples obtained from the subject are fresh tumor biopsy samples.

[0123] Tumor samples can be obtained from a subject by means including, but not limited to, tumor biopsy, needle aspiration, curettage, surgical excision, surgical incision, venipuncture, or means known in the art. Tumor biopsy is a preferred method for obtaining tumors. Tumor biopsies can be performed at any cancer site, such as a primary tumor or a secondary tumor. Generally, tumor biopsies from primary tumors are preferred. Those skilled in the art will also be able to conceive of other suitable techniques for obtaining tumor samples.

[0124] Tumor samples can be collected from subjects in a single procedure. Tumor samples can also be collected repeatedly from subjects over a period of time. For example, tumor samples may be collected once a day, once a week, monthly, every six months, or annually. Obtaining a large number of samples over a period of time is helpful when identifying and selecting new tumor-specific neoantigens. Tumor samples can be obtained from the same tumor or from different tumors.

[0125] Tumor samples can be obtained from the primary tumor, one or more metastases, and / or individual sites of tumor growth (e.g., bone marrow from different skeletal parts such as the hip, bone, or vertebra). Tumor samples can be taken from the same site or from different sites.

[0126] All or part of the above can be implemented on a computing environment as shown in Figures 1 to 3. Figure 1 shows an exemplary provider network (or “Service Provider System”) environment according to one embodiment. The provider network 900 may provide resource virtualization to customers through one or more virtualization services 910, thereby enabling customers to purchase, rent, or acquire instances 912 of virtualized resources, including but not limited to computing and storage resources, that are implemented on devices within the provider network (or more) in one or more data centers. A local Internet Protocol (IP) address 916 may be associated with a resource instance 912, where the local IP address is the internal network address of the resource instance 912 on the provider network 900. In some embodiments, the provider network 900 also provides public IP addresses 914 and / or public IP address ranges (e.g., Internet Protocol version 4 (IPv4) or Internet Protocol version 6 (IPv6) addresses) that customers can acquire from the provider 900.

[0127] Conventionally, the provider network 900 enables a service provider's customer (for example, a customer operating one or more customer networks 950A-950C, including one or more customer devices 952) to dynamically map at least several public IP addresses 914 assigned or allocated to the customer to specific resource instances 912 assigned to the customer, via the virtualization service 910. The provider network 900 also enables the customer to remap a public IP address 914 previously mapped to one virtualized computing resource instance 912 allocated to the customer to another virtualized computing resource instance 912, also allocated to the customer. A service provider's customer, such as an operator of customer networks 950A-950C(or more), uses the virtualized computing resource instances 912 and public IP addresses 914 provided by the service provider to, for example, implement customer-specific applications and present customer applications over an intermediate network 940, such as the internet. Next, another network entity 920 on the intermediate network 940 may generate traffic to a destination public IP address 914 exposed by customer networks 950A-950C(or more). This traffic is routed to the service provider data center, where it is routed via the network infrastructure to the local IP address 916 of the virtualized computing resource instance 912, which is currently mapped to the destination public IP address 914. Similarly, response traffic from the virtualized computing resource instance 912 is routed back to the intermediate network 940 via the network infrastructure and then routed to the source entity 920.

[0128] In this specification, a local IP address refers to a network address within a provider network, for example, an internal, i.e., "private," resource instance. Local IP addresses may reside within address blocks reserved by Internet Engineering Task Force (IETF) Request for Comment (RFC) 1918 and / or address blocks in address formats specified by IETF RFC 4193, and are modifiable within the provider network. Network traffic originating from outside the provider network is not directly routed to local IP addresses; instead, the traffic uses public IP addresses mapped to the resource instance's local IP addresses. The provider network includes network devices or appliances that provide Network Address Translation (NAT) or similar functionality to perform the mapping from public IP addresses to local IP addresses and vice versa.

[0129] A public IP address is a network address that is changeable on the internet and is assigned to a resource instance by a service provider or customer. Traffic routed to a public IP address is translated, for example, via 1:1 NAT and forwarded to the local IP address of each resource instance.

[0130] Some public IP addresses may be assigned to specific resource instances by the provider network infrastructure, and these public IP addresses are sometimes referred to as standard public IP addresses, or simply standard IP addresses. In some embodiments, mapping a standard IP address to the resource instance's local IP address is the default startup configuration for all resource instance types.

[0131] At least some public IP addresses may be assigned to or acquired by customers of Provider Network 900, and customers may then assign their assigned public IP addresses to specific resource instances assigned to them. These public IP addresses are referred to as customer public IP addresses, or simply customer IP addresses. Instead of being assigned to resource instances by Provider Network 900 as in the case of standard IP addresses, customer IP addresses are assigned to resource instances by the customer, for example, via an API provided by the service provider. Unlike standard IP addresses, customer IP addresses are assigned to customer accounts and can be remapped to other resource instances by each customer as needed or desired. Customer IP addresses are associated with customer accounts, not specific resource instances, and customers control those IP addresses until they choose to release them. Unlike traditional static IP addresses, customer IP addresses allow customers to mask resource instance or availability zone failures by remapping their public IP addresses to any resource instance associated with their account. Customer IP addresses allow customers to address issues with their resource instances or software by, for example, remapping their customer IP addresses to alternative resource instances.

[0132] Figure 2 is a block diagram of an exemplary provider network that provides storage services and hardware virtualization services to a customer, according to one embodiment. The hardware virtualization service 1020 provides a number of computing resources 1024 (e.g., VMs) to the customer. The computing resources 1024 may be rented or leased, for example, to a customer of the provider network 1000 (e.g., a customer implementing a customer network 1050). Each computing resource 1024 is provided with one or more local IP addresses. The provider network 1000 is configured to route packets from the local IP addresses of the computing resources 1024 to public internet destinations and from public internet sources to the local IP addresses of the computing resources 1024.

[0133] The provider network 1000 provides the customer network 1050, which is connected to the intermediate network 1040 via, for example, the local network 1056, with the ability to implement a virtual computing system 1092 via a hardware virtualization service 1020 connected to the intermediate network 1040 and the provider network 1000. In some embodiments, the hardware virtualization service 1020 may provide one or more APIs 1002, such as a web service interface, through which the customer network 1050 can access the functions provided by the hardware virtualization service 1020, for example, through a console 1094 (e.g., a web-based application, a standalone application, a mobile application, etc.). In some embodiments, each virtual computing system 1092 in the customer network 1050 corresponds to computing resources 1024 leased, rented, or provided to the customer network 1050 in the provider network 1000.

[0134] From an instance of the virtual computing system 1092 and / or another customer device 1090 (e.g., via a console 1094), the customer accesses the functions of the storage service 1010, for example via one or more APIs 1002, to access and store data in the storage resources 1018A-1018N of the virtual data store 1016 (e.g., folders or "buckets", virtualized volumes, databases, etc.) provided by the provider network 1000. In some embodiments, a virtualized data store gateway (not shown) may be provided in the customer network 1050, which may locally cache at least some data, such as frequently accessed data or critical data, and communicates with the storage service 1010 via one or more communication channels to upload new or modified data from the local cache so that the primary store of data (virtual data store 1016) is maintained. In some embodiments, a user can mount and access volumes of a virtual datastore 1016 via a virtual computing system 1092 and / or on another customer device 1090, via a storage service 1010 which functions as a storage virtualization service, and these volumes are presented to the user as local (virtualized) storage 1098.

[0135] Although not shown in Figure 2, the virtualization service(s) can also be accessed from resource instances within provider network 1000 via API(s) 1002(s). For example, a customer, appliance service provider, or other entity can access the virtualization service(s) within their respective virtual network on provider network 1000 via API 1002 to request the allocation of one or more resource instances within their virtual network or another virtual network.

[0136] Exemplary system In some embodiments, a system implementing some or all of the techniques described herein may include a general-purpose computer system configured to include or access one or more computer-accessible media, such as the computer system 1100 shown in Figure 3. In the illustrated embodiment, the computer system 1100 includes one or more processors 1110 coupled to system memory 1120 via an input / output (I / O) interface 1130. The computer system 1100 further includes a network interface 1140 connected to the I / O interface 1130. While Figure 3 shows the computer system 1100 as a single computing device, in various embodiments, the computer system 1100 may include one computing device or any number of computing devices configured to cooperate as a single computer system 1100.

[0137] In various embodiments, the computer system 1100 is a uniprocessor system including one processor 1110, or a multiprocessor system including several (e.g., two, four, eight, or another preferred number) processors 1110. The processors 1110 may be any preferred processor capable of executing instructions. For example, in various embodiments, the processors 1110 may be general-purpose or embedded processors implementing any of the following instruction set architectures (ISAs), such as x86, ARM, PowerPC, SPARC, or MIPS-ISA, or any other preferred ISA. In a multiprocessor system, each of the processors 1110 typically, but not necessarily, implements the same ISA.

[0138] The system memory 1120 stores instructions and data accessible by the processor 1110(or more). In various embodiments, the system memory 1120 is implemented by any suitable memory technology, such as random access memory (RAM), static RAM (SRAM), synchronous dynamic RAM (SDRAM), non-volatile / flash memory, or any other type of memory. In the shown embodiments, it is shown that program instructions and data that perform one or more desired functions, such as the aforementioned methods, techniques, and data, are stored in the system memory 1120 as enzyme substrate prediction service code 1125 and data 1126.

[0139] In one embodiment, the I / O interface 1130 is configured to coordinate I / O traffic between the processor 1110, the system memory 1120, and any peripheral devices in the device, including the network interface 1140 or other peripheral interfaces. In some embodiments, the I / O interface 1130 performs any necessary protocol conversion, timing conversion, or other data conversion to convert data signals from one component (e.g., the system memory 1120) into a format suitable for use by another component (e.g., the processor 1110). In some embodiments, the I / O interface 1130 includes support for devices attached via various types of peripheral buses, such as a variation of the Peripheral Component Interconnect (PCI) bus standard or the General Purpose Serial Bus (USB) standard. In some embodiments, the functionality of the I / O interface 1130 is divided into two or more separate components, such as a northbridge and a southbridge. Also, in some embodiments, some or all of the functionality of the I / O interface 1130, such as the interface to the system memory 1120, is directly incorporated into the processor 1110.

[0140] The network interface 1140 is configured to enable data exchange between the computer system 1100 and other devices 1160 connected to the network 1150(or more). In various embodiments, the network interface 1140 supports communication over any suitable wired or wireless general-purpose data network, such as Ethernet networks. Furthermore, the network interface 1140 supports communication over telecommunications / telephone networks, such as analog voice networks or digital fiber optic networks, or over storage area networks (SANs), such as Fibre Channel SANs, or over other suitable types of networks and / or protocols.

[0141] In some embodiments, the computer system 1100 includes one or more offload cards 1170 (including one or more processors 1175 and possibly one or more network interfaces 1140) connected using an I / O interface 1130 (e.g., a version of the Peripheral Components Interconnection Express (PCI-E) standard, or a bus implementing another interconnect such as QuickPath Interconnection (QPI) or UltraPath Interconnection (UPI)). For example, in some embodiments, the computer system 1100 may function as a host electronic device hosting compute instances (e.g., operating as part of a hardware virtualization service), and one or more offload cards 1170 run a virtualization manager capable of managing compute instances running on the host electronic device. As an example, in some embodiments, the offload card 1170(s) may perform compute instance management operations, such as pausing and / or unpausing compute instances, starting and / or terminating compute instances, and performing memory transfer / copy operations. These management operations are performed by the offload card 1170(or more) in cooperation with the hypervisor running on other processors 1110A-1110N of the computer system 1100 (for example, in response to requests from the hypervisor). However, in some embodiments, the virtualization manager implemented by the offload card 1170(or more) may respond to requests from other entities (for example, from the compute instance itself) and may not cooperate with (or be unable to service) any separate hypervisor.

[0142] In some embodiments, the system memory 1120 may be an embodiment of a computer-accessible medium configured to store the aforementioned program instructions and data. However, in other embodiments, the program instructions and / or data may be received, transmitted, or stored in different types of computer-accessible media. Generally, computer-accessible media include non-temporary storage media or memory media such as magnetic or optical media such as disks or DVDs / CDs connected to the computer system 1100 via an I / O interface 1130. Non-temporary computer-accessible storage media also include any volatile or non-volatile media that may be included in some embodiments of the computer system 1100 as system memory 1120 or another type of memory, such as RAM (e.g., SDRAM, double data rate (DDR) SDRAM, SRAM, etc.) and read-only memory (ROM). Furthermore, computer-accessible media include transmission media or signals such as electrical signals, electromagnetic signals, or digital signals that are transmitted via communication media such as networks and / or wireless links, which may be implemented via a network interface 1140.

[0143] The various embodiments discussed or proposed herein can be implemented in a variety of operating environments, which in some cases include one or more user computers, computing devices, or processing devices that can be used to operate any number of applications. User or client devices include any of many general-purpose personal computers, such as desktop or laptop computers running standard operating systems, as well as cellular, wireless, and portable devices that can run mobile software and support a wide range of networking and messaging protocols. Such systems also include several workstations that run any of a variety of commercially available operating systems and other known applications for purposes such as development and database management. These devices may also include other electronic devices, such as dummy terminals, thin clients, gaming systems, and / or other devices that can communicate over a network.

[0144] Most embodiments utilize at least one network well known to those skilled in the art to support communication using one of a variety of widely available protocols, such as Transmission Control Protocol / Internet Protocol (TCP / IP), File Transfer Protocol (FTP), Universal Plug and Play (UPnP), Network File System (NFS), Common Internet File System (CIFS), Extensible Messaging and Presence Protocol (XMPP), and AppleTalk. Networks may include, for example, local area networks (LANs), wide area networks (WANs), virtual private networks (VPNs), the Internet, intranets, extranets, public switched telephone networks (PSTNs), infrared networks, wireless networks, and combinations thereof.

[0145] In embodiments utilizing a web server, the web server can run any of a variety of server or middle-tier applications, including HTTP servers, File Transfer Protocol (FTP) servers, Common Gateway Interface (CGI) servers, data servers, Java servers, and business application servers. The server(s) can also execute programs or scripts in response to requests from user devices, such as by running one or more web applications that can be implemented as one or more scripts or programs written in any programming language such as Java®, C, C#, or C++, or scripting languages ​​such as Perl, Python, PHP, or TCL, or combinations thereof. The server(s) may also include, but are not limited to, database servers commercially available from Oracle®, Microsoft®, Sybase®, IBM®, and others. The database server may be relational or non-relational (e.g., "NoSQL"), distributed or undistributed, etc.

[0146] The environments disclosed herein may include various data stores and other memory and storage media, as described above. These may reside in various locations, such as local to (and / or residing within) one or more computers, or remote from any or all computers across a network. In certain sets of embodiments, information may reside within a Storage Area Network ("SAN") as is well known to those skilled in the art. Similarly, any necessary files for performing functions belonging to a computer, server, or other network device may be stored locally and / or remotely as appropriate. Where the system includes computerized devices, each such device may include hardware elements that can be electrically connected via a bus, the elements including, for example, at least one central processing unit (CPU), at least one input device (e.g., mouse, keyboard, controller, touchscreen, or keypad), and / or at least one output device (e.g., display device, printer, or speaker). Such a system also includes one or more storage devices, such as disk drives, optical storage devices, random access memory (RAM) or read-only memory (ROM), and solid-state storage devices such as removable storage devices, memory cards, and flash cards.

[0147] Such devices also include, as described above, a computer-readable storage medium reader, a communication device (e.g., a modem, a network card (wireless or wired), an infrared communication device, etc.), and working memory. The computer-readable storage medium reader can connect to, or be configured to accept, computer-readable storage media representing remote storage devices, local storage devices, fixed storage devices, and / or removable storage devices, as well as storage media, for temporarily and / or more permanently containing, storing, transmitting, and retrieving computer-readable information. The system and various devices also typically include an operating system and several software applications, modules, services, or other elements located within at least one working memory device, including application programs such as client applications or web browsers. It should be understood that alternative embodiments include variations or improvements on those described above. For example, customized hardware may also be used, and / or certain elements may be implemented in hardware, software (including portable software such as applets), or both. Furthermore, connections to other computing devices, such as network input / output devices, may be used.

[0148] Storage media and computer-readable media for containing code or part of code may include, but are not limited to, any suitable media known or used in the art, including storage media and communication media, such as RAM, ROM, electrically erasable programmable read-only memory ("EEPROM"), flash memory or other memory technologies, compact disc read-only memory ("CD-ROM"), digital versatile discs (DVDs) or other optical storage, magnetic cassettes, magnetic tapes, magnetic disk storage or other magnetic storage devices, or any other media that can be used to store desired information and can be accessed by a system device, including volatile and non-volatile media, removable and non-removable media, etc., which are implemented in any way or technique for storing and / or transmitting information such as computer-readable instructions, data structures, program modules or other data. Based on the disclosure and teachings provided herein, those skilled in the art will understand other forms and / or methods for implementing various embodiments.

[0149] Various embodiments are described above. For explanatory purposes, and to provide a complete understanding of the embodiments, specific configurations and details are provided. However, those skilled in the art will understand that embodiments can be implemented without specific details. Furthermore, well-known features may be omitted or simplified to avoid obscuring the embodiments described.

[0150] In this specification, optional actions that add additional features to certain embodiments are described using text enclosed in parentheses and blocks enclosed in dashed lines (e.g., large dashed lines, small dashed lines, dotted lines, etc.). However, such notation should not be understood to mean that these are the only options or optional actions, and / or that blocks enclosed in solid lines are not optional in certain embodiments.

[0151] Reference numbers with suffix characters may indicate that there may be one or more instances of an entity referred to in various embodiments, and that if multiple instances exist, they do not need to be identical, but instead may share some common characteristics or behave similarly. Furthermore, the specific suffix used does not imply the existence of a particular number of entities unless otherwise indicated. Thus, two entities using the same or different suffix characters may or may not have the same number of instances in various embodiments.

[0152] References to "one embodiment," "embodiment," or "exemplary embodiment" indicate that the embodiment described may include certain features, structures, or characteristics, but not all embodiments necessarily include such features, structures, or characteristics. Furthermore, such phrases do not necessarily refer to the same embodiment. Moreover, if certain features, structures, or characteristics are described in relation to an embodiment, it is considered that other embodiments may also possess such features, structures, or characteristics, whether explicitly stated or not, to be within the knowledge of those skilled in the art.

[0153] Furthermore, in the various embodiments described above, unless otherwise specified, disjunctive expressions such as the phrase "at least one of A, B, or C" are intended to be understood to mean any one of A, B, or C, or any combination thereof (e.g., A, B, and / or C). Therefore, disjunctive expressions are not intended, nor should they be understood, to mean that there must be at least one A, at least one B, or at least one C in a given embodiment.

[0154] Accordingly, the specification and drawings should be considered illustrative rather than restrictive. However, it is clear that various modifications and changes can be made herein without departing from the broader intent and scope of the disclosure as set forth in the claims.

[0155] Equal portions

[0156] Other suitable modifications and adaptations of the methods of the present invention described herein are obvious and will be readily apparent to those skilled in the art that can be carried out using suitable equivalents without departing from the scope of this disclosure or embodiments. While specific compositions and methods have been described in detail so far, the same will be more clearly understood by referring to the following examples. These examples are presented for illustrative purposes only and are not intended to limit the scope of the invention. [Examples]

[0157] The following examples are for illustrative purposes only and are not intended to limit the scope of the present invention.

[0158] Example 1: [Table 1] [Table 2] [Table 3] [Table 4]

[0159] Example 1 shows predicted mutant epitopes for a short-chain MHC class I vaccine peptide candidate and an exemplary variant according to an exemplary embodiment. In this example, the letter "H" enclosed in a square represents a mutated subsequence of the vaccine peptide sequence "FVLQHLVFL". One or more mutant epitopes can be predicted and an immunogenicity score can be generated by one or more methods described herein. According to this example, the immunogenicity score indicates the probability that at least one epitope in the long-chain peptide sequence is immunogenic to at least one of the subject's MHC class I alleles. The MHC class I binding score may indicate the probability that the peptide binds to at least one of the subject's MHC class I alleles. Furthermore, the length may indicate the number of amino acids in the sequence, which can be used to distinguish between short-chain and long-chain neoantigens. The MHC class I immunogenicity binding score can be determined by multiplying the MHC class I immunogenicity score by the MHC class I binding score. RNA TPM is the number of normalized RNA reads per gene length and sequence depth, expressed in TPM. Maximum coding sequence coverage may indicate the number of RNA reads that cover the vaccine peptide sequence.

[0160] Example 2: [Table 5] [Table 6] [Table 7] [Table 8]

[0161] Example 2 shows predicted mutant epitopes for another short-chain MHC class I vaccine peptide candidate and exemplary variants according to an exemplary embodiment. In this example, the letter "C" enclosed in a square represents a mutated subsequence of the vaccine peptide sequence "KACHYHSYNGW".

[0162] Example 3: [Table 9] [Table 10] [Table 11] [Table 12]

[0163] Example 3 shows predicted mutant epitopes for another short-chain MHC class I vaccine peptide candidate and exemplary variants. In this example, the letter "H" enclosed in a square represents a mutated subsequence of the vaccine peptide sequence "REEENHSFL".

[0164] Example 4: [Table 13] [Table 14] [Table 15] [Table 16-1] [Table 16-2] [Table 16-3]

[0165] Example 5: [Table 17] [Table 18]

[0166] In Examples 4 and 5 described above, the long-chain MHC class I vaccine peptide of Example 4 and the long-chain MHC class II vaccine peptide of Example 5 were constructed using the short-chain sequence "FVLQHLVFL" from Example 1. Both sequences contain the same short subsequence at the center (e.g., the letter "H"). As described herein, amino acids are added to both sides of the short subsequence according to the longest-chain neoantigen. This results in the presence of a first maximum number of amino acids adjacent to each side of the mutant amino acid. Predicted mutant epitopes are generated or determined for both the MHC class I vaccine peptide and the MHC class II vaccine peptide, along with their corresponding immunogenicity scores. The MHC class I immunogenicity score indicates the probability that at least one epitope in the long-chain peptide sequence is immunogenic to at least one of the subject's MHC class I alleles. The MHC class II immunogenicity score indicates the probability that at least one epitope in the long-chain peptide sequence is immunogenic to at least one of the subject's MHC class II alleles.

[0167] Example 6: [Table 19] [Table 20] [Table 21] [Table 22-1] [Table 22-2] [Table 22-3]

[0168] Example 7: [Table 23] [Table 24]

[0169] In Examples 6 and 7 above, the long-chain MHC class I vaccine peptide of Example 6 and the long-chain MHC class II vaccine peptide of Example 7 were constructed using the short-chain sequence "KACHYHSYNGW" from Example 2. Both sequences contain the same short subsequence as Example 2 at their center (e.g., the letter "C"). Predicted mutant epitopes are generated or determined for both the MHC class I vaccine peptide and the MHC class II vaccine peptide, along with their corresponding immunogenicity scores. The MHC class I immunogenicity score indicates the probability that at least one epitope in the long-chain peptide sequence is immunogenic to at least one of the subject's MHC class I alleles. The MHC class II immunogenicity score indicates the probability that at least one epitope in the long-chain peptide sequence is immunogenic to at least one of the subject's MHC class II alleles.

[0170] Example 8: [Table 25] [Table 26] [Table 27] [Table 28-1] [Table 28-2]

[0171] Example 9: [Table 29] [Table 30-1] [Table 30-2]

[0172] In Examples 8 and 9 described above, the long-chain MHC class I vaccine peptide of Example 8 and the long-chain MHC class II vaccine peptide of Example 9 were constructed using the short-chain sequence "REEENHSFL" from Example 3. Both sequences contain the same short subsequence as Example 3 at their center (e.g., the letter "H"). Predicted mutant epitopes are generated or determined for both the MHC class I vaccine peptide and the MHC class II vaccine peptide, along with their corresponding immunogenicity scores. The MHC class I immunogenicity score indicates the probability that at least one epitope in the long-chain peptide sequence is immunogenic to at least one of the subject's MHC class I alleles. The MHC class II immunogenicity score indicates the probability that at least one epitope in the long-chain peptide sequence is immunogenic to at least one of the subject's MHC class II alleles.

Claims

1. A method for ranking tumor-specific neoantigens from a subject's tumor with respect to a target-specific immunogenic composition, a) Identifying multiple somatic mutations present within the tumor, b) For individual somatic mutations among the plurality of somatic mutations, i) Determining the best short-chain neoantigen from a group of first short-chain neoantigens, at least in part, based on the immunogenicity score of the best short-chain neoantigen, wherein each individual neoantigen in the first group of short-chain neoantigens has 8, 9, 10, or 11 amino acids, and the immunogenicity score represents the quality of the best short-chain neoantigen. ii) Determining the best long-chain neoantigen from a first plurality of long-chain neoantigens, at least in part, based on the immunogenicity score of the best long-chain neoantigen, wherein the individual neoantigens in the first plurality of long-chain neoantigens have a length in the range of 15 to 30 amino acids, and the immunogenicity score represents the quality of the best long-chain neoantigen. iii) Add the best short-chain neoantigen to the short-chain neoantigen candidate list, iv) Add the best long-chain neoantigen to the long-chain neoantigen candidate list, c) Performing b) on the plurality of somatic mutations, wherein the completed short-chain neoantigen candidate list includes the best short-chain neoantigen for each of the plurality of somatic mutations, and the completed long-chain neoantigen candidate list includes the best long-chain neoantigen for each of the plurality of somatic mutations, d) Ranking the short-chain neoantigen candidate list in descending order of immunogenicity score, e) Ranking the long-chain neoantigen candidate list in descending order of immunogenicity score, wherein the short-chain neoantigen candidate list and the long-chain neoantigen candidate list include at least one MHC class I epitope and at least one MHC class II epitope, Regarding the aforementioned individual somatic mutations, the longest chain neoantigen sequence containing the mutated amino acid is identified, The method further comprises identifying the first plurality of short-chain neoantigens from the longest-chain neoantigen sequence, wherein each individual neoantigen in the first plurality of short-chain neoantigens has the mutant amino acid and a minimum to maximum number of amino acids.

2. The method according to claim 1, further comprising determining the neoantigen-allelic score for each of the first plurality of short-chain neoantigens for each individual allele in the plurality of HLA class I alleles present in the subject.

3. The method according to claim 2, wherein the neoantigen-allelic score between individual neoantigens and individual alleles in the first plurality of short-chain neoantigens is at least partially based on the probability that the individual neoantigen is presented by the individual allele, but the germline siblings of the individual neoantigen are not presented by the individual allele, the probability being determined at least partially based on data from an MHC class I machine learning model trained to determine the probability that a given allele in the plurality of HLA class I alleles presents a specific antigen.

4. The method according to claim 2, further comprising removing from the initial plurality of short chain neoantigens the neoantigen with a lower neoantigen allele score for the individual allele, where one of the two neoantigens in the initial plurality of short chain neoantigens includes the other.

5. The method according to claim 4, further comprising identifying a short subsequence, the short subsequence being the shortest subsequence of the longest neoantigen sequence containing all the neoantigens in the first plurality of short-chain neoantigens, wherein the neoantigens in the first plurality of short-chain neoantigens are not included in any other neoantigen in the first plurality of short-chain neoantigens.

6. The method according to claim 5, further comprising determining the probability that the individual allele presents at least one neoantigen in the short-chain neoantigen group and does not present a germline sibling of the at least one neoantigen.

7. Determining the immunogenicity score of individual neoantigens in the short-chain neoantigen group, further comprising determining the immunogenicity score based at least in part on the probability that at least one allele in the subject's multiple HLA class I alleles presents the individual neoantigen and does not present germline siblings of the individual neoantigen, or The method according to claim 1, further comprising determining the probability that at least one allele in a plurality of HLA class I alleles of the subject presents at least one neoantigen in the short-chain neoantigen group and does not present germline siblings of the at least one neoantigen.

8. Identifying an extended sequence, wherein the extended sequence is a partial sequence of the longest chain neoantigen comprising the short subsequence and the first maximum number of amino acids on each side of the mutant amino acid, The method according to claim 5, further comprising identifying a group of long-chain neoantigens from the extended sequence, wherein the group of long-chain neoantigens has a length between the length of the short subsequence and the length of the second largest number of amino acids.

9. The method according to claim 8, wherein the first maximum number is 29, or the second maximum number is 30, or the method further comprises removing neoantigens that do not meet the conditions for production from the long-chain neoantigen group, or the method further comprises determining the immunogenicity score of an individual neoantigen in the long-chain neoantigen group, the immunogenicity score being at least in part on the probability that at least one allele in a plurality of HLA class II alleles of the subject presents the individual neoantigen and does not present germline siblings of the individual neoantigen, and the probability being arbitrarily determined at least in part on data from an MHC class II machine learning model trained to determine the probability that a given allele in the plurality of HLA class II alleles presents a particular antigen.

10. The method according to claim 1, further comprising trimming the list of long-chain neoantigen candidates to a predetermined number of high-ranking candidates based on their immunogenicity scores.

11. Selecting a subgroup of long-chain neoantigen candidates from the long-chain neoantigen candidate list based at least partially on manufacturability, From the short-chain neoantigen candidate list, remove any neoantigen that belongs to any of the subgroups of the long-chain neoantigen candidate list. The method according to claim 1 or 10, further comprising forming a target-specific immunogenic composition comprising one or more neoantigens obtained from the long-chain neoantigen candidate list.

12. The method according to claim 1, further comprising trimming the list of short-chain neoantigen candidates to a predetermined number of high-ranking candidates based on an immunogenicity score, and selecting a lower group of short-chain neoantigen candidates from the list of short-chain neoantigen candidates at least in part on manufacturability.

13. The method according to claim 1 or 12, further comprising forming a target-specific immunogenic composition comprising one or more neoantigens obtained from the short-chain neoantigen candidate list.

14. The aforementioned first plurality of short-chain neoantigens comprise short-chain polypeptides containing at least one MHC class I epitope related to the subject, or The first plurality of long-chain neoantigens comprises a long-chain polypeptide comprising at least one MHC class I epitope and at least one MHC class II epitope related to the subject, or The first plurality of short-chain neoantigens and the first plurality of long-chain neoantigens originate from the tumor and include the individual somatic mutations, or The immunogenicity score is determined at least in part based on data from a machine learning model, or The best short-chain neoantigen for the individual somatic mutation is the short-chain neoantigen having the highest immunogenicity score among the first plurality of short-chain neoantigens for the individual somatic mutation, or The method according to claim 1, wherein the best long-chain neoantigen for the individual somatic mutation is the long-chain neoantigen having the highest immunogenicity score among the first plurality of long-chain neoantigens for the individual somatic mutation.

Images