Neoantigen Identification, Manufacture, and Use

Abstract

Disclosed herein is a system and methods for determining the alleles, neoantigens, and vaccine composition as determined on the basis of an individual's tumor mutations. Also disclosed are systems and methods for obtaining high quality sequencing data from a tumor. Further, described herein are systems and methods for identifying somatic changes in polymorphic genome data. Finally, described herein are unique cancer vaccines.

1 . A method for identifying one or more neoantigens from a tumor cell of a subject that are likely to be presented on the tumor cell surface, comprising the steps of:
 obtaining at least one of exome, transcriptome or whole genome tumor nucleotide sequencing data from the tumor cell of the subject, wherein the tumor nucleotide sequencing data is used to obtain data representing peptide sequences of each of a set of neoantigens, and wherein the peptide sequence of each neoantigen comprises at least one alteration that makes it distinct from the corresponding wild-type, parental peptide sequence;   inputting the peptide sequence of each neoantigen into one or more presentation models to generate a set of numerical likelihoods that each of the neoantigens is presented by one or more MHC alleles on the tumor cell surface of the tumor cell of the subject, the set of numerical likelihoods having been identified at least based on received mass spectrometry data; and   selecting a subset of the set of neoantigens based on the set of numerical likelihoods to generate a set of selected neoantigens.   
     
     
         2 . The method of  claim 1 , wherein a number of the set of selected neoantigens is 20. 
     
     
         3 . The method of any of  claims 1 - 2 , wherein the presentation model represents dependence between:
 presence of a pair of a particular one of the MHC alleles and a particular amino acid at a particular position of a peptide sequence; and   likelihood of presentation on the tumor cell surface, by the particular one of the MHC alleles of the pair, of such a peptide sequence comprising the particular amino acid at the particular position.   
     
     
         4 . The method of any of  claims 1 - 3 , wherein inputting the peptide sequence comprises:
 applying the one or more presentation models to the peptide sequence of the corresponding neoantigen to generate a dependency score for each of the one or more MHC alleles indicating whether the MHC allele will present the corresponding neoantigen based on at least positions of amino acids of the peptide sequence of the corresponding neoantigen.   
     
     
         5 . The method of  claim 4 , further comprising:
 transforming the dependency scores to generate a corresponding per-allele likelihood for each MHC allele indicating a likelihood that the corresponding MHC allele will present the corresponding neoantigen; and   combining the per-allele likelihoods to generate the numerical likelihood.   
     
     
         6 . The method of  claim 5 , wherein the transforming the dependency scores model the presentation of the peptide sequence of the corresponding neoantigen as mutually exclusive. 
     
     
         7 . The method of any of  claims 4 - 6 , further comprising:
 transforming a combination of the dependency scores to generate the numerical likelihood.   
     
     
         8 . The method of  claim 7 , wherein the transforming the combination of the dependency scores models the presentation of the peptide sequence of the corresponding neoantigen as interfering between MHC alleles. 
     
     
         9 . The method of any of  claims 4 - 8 , wherein the set of numerical likelihoods are further identified by at least an allele noninteracting feature, and further comprising:
 applying an allele noninteracting one of the one or more presentation models to the allele noninteracting features to generate a dependency score for the allele noninteracting features indicating whether the peptide sequence of the corresponding neoantigen will be presented based on the allele noninteracting features.   
     
     
         10 . The method of  claim 9 , further comprising:
 combining the dependency score for each MHC allele in the one or more MHC alleles with the dependency score for the allele noninteracting feature;   transforming the combined dependency scores for each MHC allele to generate a corresponding per-allele likelihood for the MHC allele indicating a likelihood that the corresponding MHC allele will present the corresponding neoantigen; and   combining the per-allele likelihoods to generate the numerical likelihood.   
     
     
         11 . The method of any of  claims 9 - 10 , further comprising:
 transforming a combination of the dependency scores for each of the MHC alleles and the dependency score for the allele noninteracting features to generate the numerical likelihood.   
     
     
         12 . The method of any of  claims 1 - 11 , wherein a set of numerical parameters for the presentation model is trained based on a training data set including at least a set of training peptide sequences identified as present in a plurality of samples and one or more MHC alleles associated with each training peptide sequence, wherein the training peptide sequences are identified through mass spectrometry on isolated peptides eluted from MHC alleles derived from the plurality of samples. 
     
     
         13 . The method of  claim 12 , wherein the training data set further includes data on mRNA expression levels of the tumor cell. 
     
     
         14 . The method of any of  claims 12 - 13 , wherein the samples comprise cell lines engineered to express a single MHC class I or class II allele. 
     
     
         15 . The method of any of  claims 12 - 14 , wherein the samples comprise cell lines engineered to express a plurality of MHC class I or class II alleles. 
     
     
         16 . The method of any of  claims 12 - 15 , wherein the samples comprise human cell lines obtained or derived from a plurality of patients. 
     
     
         17 . The method of any of  claims 12 - 16 , wherein the samples comprise fresh or frozen tumor samples obtained from a plurality of patients. 
     
     
         18 . The method of any of  claims 12 - 17 , wherein the samples comprise fresh or frozen tissue samples obtained from a plurality of patients. 
     
     
         19 . The method of any of  claims 12 - 18 , wherein the samples comprise peptides identified using T-cell assays. 
     
     
         20 . The method of any of  claims 12 - 19 , wherein the training data set further comprises data associated with:
 peptide abundance of the set of training peptides present in the samples;   peptide length of the set of training peptides in the samples.   
     
     
         21 . The method of any of  claims 12 - 20 , wherein the training data set is generated by comparing the set of training peptide sequences via alignment to a database comprising a set of known protein sequences, wherein the set of training protein sequences are longer than and include the training peptide sequences. 
     
     
         22 . The method of any of  claims 12 - 21 , wherein the training data set is generated based on performing or having performed mass spectrometry on a cell line to obtain at least one of exome, transcriptome, or whole genome peptide sequencing data from the cell line, the peptide sequencing data including at least one protein sequence including an alteration. 
     
     
         23 . The method of any of  claims 12 - 22 , wherein the training data set is generated based on obtaining at least one of exome, transcriptome, and whole genome normal nucleotide sequencing data from normal tissue samples. 
     
     
         24 . The method of any of  claims 12 - 23 , wherein the training data set further comprises data associated with proteome sequences associated with the samples. 
     
     
         25 . The method of any of  claims 12 - 24 , wherein the training data set further comprises data associated with MHC peptidome sequences associated with the samples. 
     
     
         26 . The method of any of  claims 12 - 25 , wherein the training data set further comprises data associated with peptide-MHC binding affinity measurements for at least one of the isolated peptides. 
     
     
         27 . The method of any of  claims 12 - 26 , wherein the training data set further comprises data associated with peptide-MHC binding stability measurements for at least one of the isolated peptides. 
     
     
         28 . The method of any of  claims 12 - 27 , wherein the training data set further comprises data associated with transcriptomes associated with the samples. 
     
     
         29 . The method of any of  claims 12 - 28 , wherein the training data set further comprises data associated with genomes associated with the samples. 
     
     
         30 . The method of any of  claims 12 - 29 , wherein the training peptide sequences are of lengths within a range of k-mers where k is between 8-15, inclusive. 
     
     
         31 . The method of any of  claims 12 - 30 , further comprising encoding the peptide sequence using a one-hot encoding scheme. 
     
     
         32 . The method of  claim 31 , further comprising encoding the training peptide sequences using a left-padded one-hot encoding scheme. 
     
     
         33 . A method of treating a subject having a tumor, comprising performing any of the steps of  claims 1 - 32 , and further comprising obtaining a tumor vaccine comprising the set of selected neoantigens, and administering the tumor vaccine to the subject. 
     
     
         34 . A method of manufacturing a tumor vaccine, comprising performing any of the steps of  claims 1 - 33 , and further comprising producing or having produced a tumor vaccine comprising the set of selected neoantigens. 
     
     
         35 . A tumor vaccine comprising a set of selected neoantigens of any of  claims 1 - 32 , selected by performing the method of any of  claims 1 - 32 . 
     
     
         36 . The vaccine of  claim 35 , wherein the tumor vaccine comprises one or more of a nucleotide sequence, a polypeptide sequence, RNA, DNA, a cell, a plasmid, or a vector. 
     
     
         37 . The vaccine of any of  claims 35 - 36 , wherein the tumor vaccine comprises one or more neoantigens presented on the tumor cell surface. 
     
     
         38 . The vaccine of any of  claims 35 - 37 , wherein the tumor vaccine comprises one or more neoantigens that is immunogenic in the subject. 
     
     
         39 . The vaccine of any of  claims 35 - 38 , wherein the tumor vaccine does not comprise one or more neoantigens that induce an autoimmune response against normal tissue in the subject. 
     
     
         40 . The vaccine of any of  claims 35 - 39 , wherein the tumor vaccine further comprises an adjuvant. 
     
     
         41 . The vaccine of any of  claims 35 - 40 , wherein the tumor vaccine further comprises an excipient. 
     
     
         42 . The method of any of  claims 1 - 41 , wherein selecting the set of selected neoantigens comprises selecting neoantigens that have an increased likelihood of being presented on the tumor cell surface relative to unselected neoantigens based on the presentation model. 
     
     
         43 . The method of any of  claims 1 - 42 , wherein selecting the set of selected neoantigens comprises selecting neoantigens that have an increased likelihood of being capable of inducing a tumor-specific immune response in the subject relative to unselected neoantigens based on the presentation model. 
     
     
         44 . The method of any of  claims 1 - 43 , wherein selecting the set of selected neoantigens comprises selecting neoantigens that have an increased likelihood of being capable of being presented to naïve T cells by professional antigen presenting cells (APCs) relative to unselected neoantigens based on the presentation model, optionally wherein the APC is a dendritic cell (DC). 
     
     
         45 . The method of any of  claims 1 - 44 , wherein selecting the set of selected neoantigens comprises selecting neoantigens that have a decreased likelihood of being subject to inhibition via central or peripheral tolerance relative to unselected neoantigens based on the presentation model. 
     
     
         46 . The method of any of  claims 1 - 45 , wherein selecting the set of selected neoantigens comprises selecting neoantigens that have a decreased likelihood of being capable of inducing an autoimmune response to normal tissue in the subject relative to unselected neoantigens based on the presentation model. 
     
     
         47 . The method of any of  claims 1 - 46 , wherein exome or transcriptome nucleotide sequencing data is obtained by performing sequencing on the tumor tissue. 
     
     
         48 . The method of any of  claims 1 - 47 , wherein the sequencing is next generation sequencing (NGS) or any massively parallel sequencing approach. 
     
     
         49 . The method of any of  claims 1 - 48 , wherein the set of numerical likelihoods are further identified by at least MHC-allele interacting features comprising at least one of:
 a. The predicted affinity with which the MHC allele and the neoantigen encoded peptide bind.   b. The predicted stability of the neoantigen encoded peptide-MHC complex.   c. The sequence and length of the neoantigen encoded peptide.   d. The probability of presentation of neoantigen encoded peptides with similar sequence in cells from other individuals expressing the particular MHC allele as assessed by mass-spectrometry proteomics or other means.   e. The expression levels of the particular MHC allele in the subject in question (e.g. as measured by RNA-seq or mass spectrometry).   f. The overall neoantigen encoded peptide-sequence-independent probability of presentation by the particular MHC allele in other distinct subjects who express the particular MHC allele.   g. The overall neoantigen encoded peptide-sequence-independent probability of presentation by MHC alleles in the same family of molecules (e.g., HLA-A, HLA-B, HLA-C, HLA-DQ, HLA-DR, HLA-DP) in other distinct subjects.   
     
     
         50 . The method of any of  claims 1 - 49 , wherein the set of numerical likelihoods are further identified by at least MHC-allele noninteracting features comprising at least one of:
 a. The C- and N-terminal sequences flanking the neoantigen encoded peptide within its source protein sequence.   b. The presence of protease cleavage motifs in the neoantigen encoded peptide, optionally weighted according to the expression of corresponding proteases in the tumor cells (as measured by RNA-seq or mass spectrometry).   c. The turnover rate of the source protein as measured in the appropriate cell type.   d. The length of the source protein, optionally considering the specific splice variants (“isoforms”) most highly expressed in the tumor cells as measured by RNA-seq or proteome mass spectrometry, or as predicted from the annotation of germline or somatic splicing mutations detected in DNA or RNA sequence data.   e. The level of expression of the proteasome, immunoproteasome, thymoproteasome, or other proteases in the tumor cells (which may be measured by RNA-seq, proteome mass spectrometry, or immunohistochemistry).   f. The expression of the source gene of the neoantigen encoded peptide (e.g., as measured by RNA-seq or mass spectrometry).   g. The typical tissue-specific expression of the source gene of the neoantigen encoded peptide during various stages of the cell cycle.   h. A comprehensive catalog of features of the source protein and/or its domains as can be found in e.g. uniProt or PDB http://www.rcsb.org/pdb/home/home.do.   i. Features describing the properties of the domain of the source protein containing the peptide, for example: secondary or tertiary structure (e.g., alpha helix vs beta sheet); Alternative splicing.   j. The probability of presentation of peptides from the source protein of the neoantigen encoded peptide in question in other distinct subjects.   k. The probability that the peptide will not be detected or over-represented by mass spectrometry due to technical biases.   l. The expression of various gene modules/pathways as measured by RNASeq (which need not contain the source protein of the peptide) that are informative about the state of the tumor cells, stroma, or tumor-infiltrating lymphocytes (TILs).   m. The copy number of the source gene of the neoantigen encoded peptide in the tumor cells.   n. The probability that the peptide binds to the TAP or the measured or predicted binding affinity of the peptide to the TAP.   o. The expression level of TAP in the tumor cells (which may be measured by RNA-seq, proteome mass spectrometry, immunohistochemistry).   p. Presence or absence of tumor mutations, including, but not limited to:
 i. Driver mutations in known cancer driver genes such as EGFR, KRAS, ALK, RET, ROS1, TP53, CDKN2A, CDKN2B, NTRK1, NTRK2, NTRK3 
 ii. In genes encoding the proteins involved in the antigen presentation machinery (e.g., B2M, HLA-A, HLA-B, HLA-C, TAP-1, TAP-2, TAPBP, CALR, CNX, ERP57, HLA-DM, HLA-DMA, HLA-DMB, HLA-DO, HLA-DOA, HLA-DOBHLA-DP, HLA-DPA1, HLA-DPB1, HLA-DQ, HLA-DQA1, HLA-DQA2, HLA-DQB1, HLA-DQB2, HLA-DR, HLA-DRA, HLA-DRB1, HLA-DRB3, HLA-DRB4, HLA-DRB5 or any of the genes coding for components of the proteasome or immunoproteasome). Peptides whose presentation relies on a component of the antigen-presentation machinery that is subject to loss-of-function mutation in the tumor have reduced probability of presentation. 
   q. Presence or absence of functional germline polymorphisms, including, but not limited to:
 i. In genes encoding the proteins involved in the antigen presentation machinery (e.g., B2M, HLA-A, HLA-B, HLA-C, TAP-1, TAP-2, TAPBP, CALR, CNX, ERP57, HLA-DM, HLA-DMA, HLA-DMB, HLA-DO, HLA-DOA, HLA-DOBHLA-DP, HLA-DPA1, HLA-DPB1, HLA-DQ, HLA-DQA1, HLA-DQA2, HLA-DQB1, HLA-DQB2, HLA-DR, HLA-DRA, HLA-DRB1, HLA-DRB3, HLA-DRB4, HLA-DRB5 or any of the genes coding for components of the proteasome or immunoproteasome) 
   r. Tumor type (e.g., NSCLC, melanoma).   s. Clinical tumor subtype (e.g., squamous lung cancer vs. non-squamous).   t. Smoking history.   u. The typical expression of the source gene of the peptide in the relevant tumor type or clinical subtype, optionally stratified by driver mutation.   
     
     
         51 . The method of any of  claims 1 - 50 , wherein the at least one mutation is a frameshift or nonframeshift indel, missense or nonsense substitution, splice site alteration, genomic rearrangement or gene fusion, or any genomic or expression alteration giving rise to a neoORF. 
     
     
         52 . The method of any of  claims 1 - 51 , wherein the tumor cell is selected from the group consisting of: lung cancer, 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 myelogenous leukemia, chronic myelogenous leukemia, chronic lymphocytic leukemia, and T cell lymphocytic leukemia, non-small cell lung cancer, and small cell lung cancer. 
     
     
         53 . The method of any of  claims 1 - 52 , further comprising obtaining a tumor vaccine comprising the set of selected neoantigens or a subset thereof, optionally further comprising administering the tumor vaccine to the subject. 
     
     
         54 . The method of any of  claims 1 - 53 , wherein at least one of neoantigens in the set of selected neoantigens, when in polypeptide form, comprises at least one of: a binding affinity with MHC with an IC50 value of less than 1000 nM, for MHC Class 1 polypeptides a length of 8-15, 8, 9, 10, 11, 12, 13, 14, or 15 amino acids, presence of sequence motifs within or near the polypeptide in the parent protein sequence promoting proteasome cleavage, and presence of sequence motifs promoting TAP transport. 
     
     
         55 . A method for generating a model for identifying one or more neoantigens that are likely to be presented on a tumor cell surface of a tumor cell, comprising executing the steps of:
 receiving mass spectrometry data comprising data associated with a plurality of isolated peptides eluted from major histocompatibility complex (MHC) derived from a plurality of samples;   obtaining a training data set by at least identifying a set of training peptide sequences present in the samples and one or more MHCs associated with each training peptide sequence;   training a set of numerical parameters of a presentation model using the training data set comprising the training peptide sequences, the presentation model providing a plurality of numerical likelihoods that peptide sequences from the tumor cell are presented by one or more MHC alleles on the tumor cell surface.   
     
     
         56 . The method of  claim 55 , wherein the presentation model represents dependence between:
 presence of a particular amino acid at a particular position of a peptide sequence; and   likelihood of presentation, by one of the MHC alleles on the tumor cell, of the peptide sequence containing the particular amino acid at the particular position.   
     
     
         57 . The method of any of  claims 55 - 56 , wherein the samples comprise cell lines engineered to express a single MHC class I or class II allele. 
     
     
         58 . The method of any of  claims 55 - 57 , wherein the samples comprise cell lines engineered to express a plurality of MHC class I or class II alleles. 
     
     
         59 . The method of any of  claims 55 - 58 , wherein the samples comprise human cell lines obtained or derived from a plurality of patients. 
     
     
         60 . The method of any of  claims 55 - 59 , wherein the samples comprise fresh or frozen tumor samples obtained from a plurality of patients. 
     
     
         61 . The method of any of  claims 55 - 60 , wherein the samples comprise peptides identified using T-cell assays. 
     
     
         62 . The method of any of  claims 55 - 61 , wherein the training data set further comprises data associated with:
 peptide abundance of the set of training peptides present in the samples;   peptide length of the set of training peptides in the samples.   
     
     
         63 . The method of any of  claims 55 - 62 , wherein obtaining the training data set comprises:
 obtaining a set of training protein sequences based on the training peptide sequences by comparing the set of training peptide sequences via alignment to a database comprising a set of known protein sequences, wherein the set of training protein sequences are longer than and include the training peptide sequences.   
     
     
         64 . The method of any of  claims 55 - 63 , wherein obtaining the training data set comprises:
 performing or having performed mass spectrometry on a cell line to obtain at least one of exome, transcriptome, or whole genome nucleotide sequencing data from the cell line, the nucleotide sequencing data including at least one protein sequence including a mutation.   
     
     
         65 . The method of any of  claims 55 - 64 , wherein training the set of parameters of the presentation model comprises:
 encoding the training peptide sequences using a one-hot encoding scheme.   
     
     
         66 . The method of any of  claims 55 - 65 , further comprising:
 obtaining at least one of exome, transcriptome, and whole genome normal nucleotide sequencing data from normal tissue samples; and   training the set of parameters of the presentation model using the normal nucleotide sequencing data.   
     
     
         67 . The method of any of  claims 55 - 66 , wherein the training data set further comprises data associated with proteome sequences associated with the samples. 
     
     
         68 . The method of any of  claims 55 - 67 , wherein the training data set further comprises data associated with MHC peptidome sequences associated with the samples. 
     
     
         69 . The method of any of  claims 55 - 68 , wherein the training data set further comprises data associated with peptide-MHC binding affinity measurements for at least one of the isolated peptides. 
     
     
         70 . The method of any of  claims 55 - 69 , wherein the training data set further comprises data associated with peptide-MHC binding stability measurements for at least one of the isolated peptides. 
     
     
         71 . The method of any of  claims 55 - 70 , wherein the training data set further comprises data associated with transcriptomes associated with the samples. 
     
     
         72 . The method of any of  claims 55 - 71 , wherein the training data set further comprises data associated with genomes associated with the samples. 
     
     
         73 . The method of any of  claims 55 - 72 , wherein training the set of numerical parameters further comprises:
 logistically regressing the set of parameters.   
     
     
         74 . The method of any of  claims 55 - 73 , wherein the training peptide sequences are of lengths within a range of k-mers where k is between 8-15, inclusive. 
     
     
         75 . The method of any of  claims 55 - 74 , wherein training the set of numerical parameters of the presentation model comprises:
 encoding the training peptide sequences using a left-padded one-hot encoding scheme.   
     
     
         76 . The method of any of  claims 55 - 75 , wherein training the set of numerical parameters further comprises:
 determining values for the set of parameters using a deep learning algorithm.   
     
     
         77 . A method for generating a model for identifying one or more neoantigens that are likely to be presented on a tumor cell surface of a tumor cell, comprising executing the steps of:
 receiving mass spectrometry data comprising data associated with a plurality of isolated peptides eluted from major histocompatibility complex (MHC) derived from a plurality of fresh or frozen tumor samples;   obtaining a training data set by at least identifying a set of training peptide sequences present in the tumor samples and presented on one or more MHC alleles associated with each training peptide sequence;   obtaining a set of training protein sequences based on the training peptide sequences; and   training a set of numerical parameters of a presentation model using the training protein sequences and the training peptide sequences, the presentation model providing a plurality of numerical likelihoods that peptide sequences from the tumor cell are presented by one or more MHC alleles on the tumor cell surface.   
     
     
         78 . The method of  claim 77 , wherein the presentation model represents dependence between:
 presence of a pair of a particular one of the MHC alleles and a particular amino acid at a particular position of a peptide sequence; and   likelihood of presentation on the tumor cell surface, by the particular one of the MHC alleles of the pair, of such a peptide sequence comprising the particular amino acid at the particular position.