Classifications

• • G—PHYSICS
  • G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
  • G16B—BIOINFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR GENETIC OR PROTEIN-RELATED DATA PROCESSING IN COMPUTATIONAL MOLECULAR BIOLOGY
  • G16B30/00—ICT specially adapted for sequence analysis involving nucleotides or amino acids
• • G—PHYSICS
  • G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
  • G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
  • G16H20/00—ICT specially adapted for therapies or health-improving plans, e.g. for handling prescriptions, for steering therapy or for monitoring patient compliance
  • G16H20/10—ICT specially adapted for therapies or health-improving plans, e.g. for handling prescriptions, for steering therapy or for monitoring patient compliance relating to drugs or medications, e.g. for ensuring correct administration to patients
• • A—HUMAN NECESSITIES
  • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
  • A61K—PREPARATIONS FOR MEDICAL, DENTAL OR TOILETRY PURPOSES
  • A61K35/00—Medicinal preparations containing materials or reaction products thereof with undetermined constitution
  • A61K35/12—Materials from mammals; Compositions comprising non-specified tissues or cells; Compositions comprising non-embryonic stem cells; Genetically modified cells
  • A61K35/14—Blood; Artificial blood
  • A61K35/15—Cells of the myeloid line, e.g. granulocytes, basophils, eosinophils, neutrophils, leucocytes, monocytes, macrophages or mast cells; Myeloid precursor cells; Antigen-presenting cells, e.g. dendritic cells
• • A—HUMAN NECESSITIES
  • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
  • A61K—PREPARATIONS FOR MEDICAL, DENTAL OR TOILETRY PURPOSES
  • A61K39/00—Medicinal preparations containing antigens or antibodies
  • A61K39/0005—Vertebrate antigens
  • A61K39/0011—Cancer antigens
• • A—HUMAN NECESSITIES
  • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
  • A61K—PREPARATIONS FOR MEDICAL, DENTAL OR TOILETRY PURPOSES
  • A61K39/00—Medicinal preparations containing antigens or antibodies
  • A61K39/39—Medicinal preparations containing antigens or antibodies characterised by the immunostimulating additives, e.g. chemical adjuvants
• • A—HUMAN NECESSITIES
  • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
  • A61P—SPECIFIC THERAPEUTIC ACTIVITY OF CHEMICAL COMPOUNDS OR MEDICINAL PREPARATIONS
  • A61P35/00—Antineoplastic agents
• • A—HUMAN NECESSITIES
  • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
  • A61P—SPECIFIC THERAPEUTIC ACTIVITY OF CHEMICAL COMPOUNDS OR MEDICINAL PREPARATIONS
  • A61P35/00—Antineoplastic agents
  • A61P35/02—Antineoplastic agents specific for leukemia
• • C—CHEMISTRY; METALLURGY
  • C12—BIOCHEMISTRY; BEER; SPIRITS; WINE; VINEGAR; MICROBIOLOGY; ENZYMOLOGY; MUTATION OR GENETIC ENGINEERING
  • C12Q—MEASURING OR TESTING PROCESSES INVOLVING ENZYMES, NUCLEIC ACIDS OR MICROORGANISMS; COMPOSITIONS OR TEST PAPERS THEREFOR; PROCESSES OF PREPARING SUCH COMPOSITIONS; CONDITION-RESPONSIVE CONTROL IN MICROBIOLOGICAL OR ENZYMOLOGICAL PROCESSES
  • C12Q1/00—Measuring or testing processes involving enzymes, nucleic acids or microorganisms; Compositions therefor; Processes of preparing such compositions
  • C12Q1/68—Measuring or testing processes involving enzymes, nucleic acids or microorganisms; Compositions therefor; Processes of preparing such compositions involving nucleic acids
  • C12Q1/6869—Methods for sequencing
• • C—CHEMISTRY; METALLURGY
  • C12—BIOCHEMISTRY; BEER; SPIRITS; WINE; VINEGAR; MICROBIOLOGY; ENZYMOLOGY; MUTATION OR GENETIC ENGINEERING
  • C12Q—MEASURING OR TESTING PROCESSES INVOLVING ENZYMES, NUCLEIC ACIDS OR MICROORGANISMS; COMPOSITIONS OR TEST PAPERS THEREFOR; PROCESSES OF PREPARING SUCH COMPOSITIONS; CONDITION-RESPONSIVE CONTROL IN MICROBIOLOGICAL OR ENZYMOLOGICAL PROCESSES
  • C12Q1/00—Measuring or testing processes involving enzymes, nucleic acids or microorganisms; Compositions therefor; Processes of preparing such compositions
  • C12Q1/68—Measuring or testing processes involving enzymes, nucleic acids or microorganisms; Compositions therefor; Processes of preparing such compositions involving nucleic acids
  • C12Q1/6876—Nucleic acid products used in the analysis of nucleic acids, e.g. primers or probes
  • C12Q1/6883—Nucleic acid products used in the analysis of nucleic acids, e.g. primers or probes for diseases caused by alterations of genetic material
  • C12Q1/6886—Nucleic acid products used in the analysis of nucleic acids, e.g. primers or probes for diseases caused by alterations of genetic material for cancer
• • G—PHYSICS
  • G01—MEASURING; TESTING
  • G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
  • G01N27/00—Investigating or analysing materials by the use of electric, electrochemical, or magnetic means
  • G01N27/62—Investigating or analysing materials by the use of electric, electrochemical, or magnetic means by investigating the ionisation of gases, e.g. aerosols; by investigating electric discharges, e.g. emission of cathode
• • G—PHYSICS
  • G01—MEASURING; TESTING
  • G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
  • G01N33/00—Investigating or analysing materials by specific methods not covered by groups G01N1/00 - G01N31/00
  • G01N33/48—Biological material, e.g. blood, urine; Haemocytometers
  • G01N33/50—Chemical analysis of biological material, e.g. blood, urine; Testing involving biospecific ligand binding methods; Immunological testing
  • G01N33/53—Immunoassay; Biospecific binding assay; Materials therefor
  • G01N33/575—Immunoassay; Biospecific binding assay; Materials therefor for cancer
  • G01N33/5758—Immunoassay; Biospecific binding assay; Materials therefor for cancer involving compounds serving as markers for tumours, cancers or neoplasias, e.g. cellular determinants, receptors, heat shock/stress proteins, A-protein, oligosaccharides or metabolites
• • G—PHYSICS
  • G01—MEASURING; TESTING
  • G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
  • G01N33/00—Investigating or analysing materials by specific methods not covered by groups G01N1/00 - G01N31/00
  • G01N33/48—Biological material, e.g. blood, urine; Haemocytometers
  • G01N33/50—Chemical analysis of biological material, e.g. blood, urine; Testing involving biospecific ligand binding methods; Immunological testing
  • G01N33/68—Chemical analysis of biological material, e.g. blood, urine; Testing involving biospecific ligand binding methods; Immunological testing involving proteins, peptides or amino acids
  • G01N33/6803—General methods of protein analysis not limited to specific proteins or families of proteins
  • G01N33/6848—Methods of protein analysis involving mass spectrometry
• • G—PHYSICS
  • G09—EDUCATION; CRYPTOGRAPHY; DISPLAY; ADVERTISING; SEALS
  • G09F—DISPLAYING; ADVERTISING; SIGNS; LABELS OR NAME-PLATES; SEALS
  • G09F19/00—Advertising or display means not otherwise provided for
  • G09F19/12—Advertising or display means not otherwise provided for using special optical effects
  • G09F19/16—Advertising or display means not otherwise provided for using special optical effects involving the use of mirrors
• • G—PHYSICS
  • G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
  • G16B—BIOINFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR GENETIC OR PROTEIN-RELATED DATA PROCESSING IN COMPUTATIONAL MOLECULAR BIOLOGY
  • G16B20/00—ICT specially adapted for functional genomics or proteomics, e.g. genotype-phenotype associations
• • G—PHYSICS
  • G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
  • G16B—BIOINFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR GENETIC OR PROTEIN-RELATED DATA PROCESSING IN COMPUTATIONAL MOLECULAR BIOLOGY
  • G16B20/00—ICT specially adapted for functional genomics or proteomics, e.g. genotype-phenotype associations
  • G16B20/20—Allele or variant detection, e.g. single nucleotide polymorphism [SNP] detection
• • G—PHYSICS
  • G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
  • G16B—BIOINFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR GENETIC OR PROTEIN-RELATED DATA PROCESSING IN COMPUTATIONAL MOLECULAR BIOLOGY
  • G16B40/00—ICT specially adapted for biostatistics; ICT specially adapted for bioinformatics-related machine learning or data mining, e.g. knowledge discovery or pattern finding
• • G—PHYSICS
  • G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
  • G16B—BIOINFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR GENETIC OR PROTEIN-RELATED DATA PROCESSING IN COMPUTATIONAL MOLECULAR BIOLOGY
  • G16B40/00—ICT specially adapted for biostatistics; ICT specially adapted for bioinformatics-related machine learning or data mining, e.g. knowledge discovery or pattern finding
  • G16B40/10—Signal processing, e.g. from mass spectrometry [MS] or from PCR
• • G—PHYSICS
  • G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
  • G16B—BIOINFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR GENETIC OR PROTEIN-RELATED DATA PROCESSING IN COMPUTATIONAL MOLECULAR BIOLOGY
  • G16B50/00—ICT programming tools or database systems specially adapted for bioinformatics
  • G16B50/40—Encryption of genetic data
• • G—PHYSICS
  • G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
  • G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
  • G16H20/00—ICT specially adapted for therapies or health-improving plans, e.g. for handling prescriptions, for steering therapy or for monitoring patient compliance
• • A—HUMAN NECESSITIES
  • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
  • A61K—PREPARATIONS FOR MEDICAL, DENTAL OR TOILETRY PURPOSES
  • A61K39/00—Medicinal preparations containing antigens or antibodies
  • A61K2039/51—Medicinal preparations containing antigens or antibodies comprising whole cells, viruses or DNA/RNA
  • A61K2039/515—Animal cells
  • A61K2039/5152—Tumor cells
• • A—HUMAN NECESSITIES
  • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
  • A61K—PREPARATIONS FOR MEDICAL, DENTAL OR TOILETRY PURPOSES
  • A61K39/00—Medicinal preparations containing antigens or antibodies
  • A61K2039/51—Medicinal preparations containing antigens or antibodies comprising whole cells, viruses or DNA/RNA
  • A61K2039/515—Animal cells
  • A61K2039/5154—Antigen presenting cells [APCs], e.g. dendritic cells or macrophages
• • A—HUMAN NECESSITIES
  • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
  • A61K—PREPARATIONS FOR MEDICAL, DENTAL OR TOILETRY PURPOSES
  • A61K39/00—Medicinal preparations containing antigens or antibodies
  • A61K2039/51—Medicinal preparations containing antigens or antibodies comprising whole cells, viruses or DNA/RNA
  • A61K2039/53—DNA (RNA) vaccination
• • A—HUMAN NECESSITIES
  • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
  • A61K—PREPARATIONS FOR MEDICAL, DENTAL OR TOILETRY PURPOSES
  • A61K39/00—Medicinal preparations containing antigens or antibodies
  • A61K2039/57—Medicinal preparations containing antigens or antibodies characterised by the type of response, e.g. Th1, Th2
  • A61K2039/577—Medicinal preparations containing antigens or antibodies characterised by the type of response, e.g. Th1, Th2 tolerising response
• • A—HUMAN NECESSITIES
  • A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
  • A61K—PREPARATIONS FOR MEDICAL, DENTAL OR TOILETRY PURPOSES
  • A61K39/00—Medicinal preparations containing antigens or antibodies
  • A61K2039/58—Medicinal preparations containing antigens or antibodies raising an immune response against a target which is not the antigen used for immunisation
  • A61K2039/585—Medicinal preparations containing antigens or antibodies raising an immune response against a target which is not the antigen used for immunisation wherein the target is cancer
• • G—PHYSICS
  • G01—MEASURING; TESTING
  • G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
  • G01N2333/00—Assays involving biological materials from specific organisms or of a specific nature
  • G01N2333/435—Assays involving biological materials from specific organisms or of a specific nature from animals; from humans
  • G01N2333/705—Assays involving receptors, cell surface antigens or cell surface determinants
  • G01N2333/70503—Immunoglobulin superfamily, e.g. VCAMs, PECAM, LFA-3
  • G01N2333/70539—MHC-molecules, e.g. HLA-molecules
• • Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
  • Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
  • Y02A—TECHNOLOGIES FOR ADAPTATION TO CLIMATE CHANGE
  • Y02A90/00—Technologies having an indirect contribution to adaptation to climate change
  • Y02A90/10—Information and communication technologies [ICT] supporting adaptation to climate change, e.g. for weather forecasting or climate simulation

Definitions

• Neoantigen identification manufacture, and use.
• FIG. 1 A shows current clinical approaches to neoantigen identification.
• FIG. IB shows that ⁇ 5% of predicted bound peptides are presented on tumor ceils.
• FIG. IE shows probability of MHC -I presentation as a function of peptide length.
• FIG. 1 G shows how the addition of features increases the model positive predictive value.
• FIG. 2A is an overview of an environment for identifying likelihoods of peptide presentation in patients, in accordance with an embodiment.
• FIG. 3 is a high-level block diagram illustrating the computer logic components of the presentation identification system, according to one embodiment.
• FIG. 7 illustrates generating a presentation likelihood for a peptide in association with an MHC allele using an example network model.
• FIG. 9 illustrates generating a presentation likelihood for a peptide in association with MHC alleles using example network models.
• FIG. 10 illustrates generating a presentation likelihood for a peptide in association with MHC alleles using example network models.
• FIG. 12 illustrates generating a presentation likelihood for a peptide in association with MHC alleles using example network models.
• FIG. 13 illustrates performance results of various example presentation models.
• FIG. 14 illustrates an example computer for implementing the entities shown in FIGS. 1 and 3.
• tumor neoantigen is a neoantigen present in a subject's tumor cell or tissue but not in the subject's corresponding normal cell or tissue.
• neoantigen-based vaccine is a vaccine construct based on one or more neoantigens, e.g., a plurality of neoantigens.
• candidate neoantigen is a mutation or other aberration giving rise to a new sequence that may represent a neoantigen.
• coding region is the portion(s) of a gene that encode protein.
• coding mutation is a mutation occurring in a coding region.
• ORF means open reading frame
• NEO-ORF is a tumor-specific ORF arising from a mutation or other aberration such as splicing.
• missense mutation is a mutation causing a substitution from one amino acid to another.
• nonsense mutation is a mutation causing a substitution from an amino acid to a stop codon.
• frameshift mutation is a mutation causing a change in the frame of the protein.
• the term “indel” is an insertion or deletion of one or more nucleic acids.
• the term percent "identity,” in the context of two or more nucleic acid or polypeptide sequences, refer to two or more sequences or subsequences that have a specified percentage of nucleotides or amino acid residues that are the same, when compared and aligned for m aximum correspondence, as measured using one of the sequence comparison algorithms described below (e.g., BLASTP and BLASTN or other algorithms available to persons of skill) or by visual inspection. Depending on the application, the percent “identity” can exist over a region of the sequence being compared, e.g., over a functional domain, or, alternatively, exist over the full length of the two sequences to be compared.
• sequence comparison typically one sequence acts as a reference sequence to which test sequences are compared.
• test and reference sequences are input into a computer, subsequence coordinates are designated, if necessary, and sequence algorithm program parameters are designated.
• sequence comparison algorithm then calculates the percent sequence identity for the test sequence(s) relative to the reference sequence, based on the designated program
• Optimal alignment of sequences for comparison can be conducted, e.g., by the local homology algorithm of Smith & Waterman, Adv. Appl. Math. 2:482 (1981 ), by the homology alignment algorithm of Needleman & Wunsch, J. Mol. Biol. 48:443 (1970), by the search for similarity method of Pearson & Lipman, Proc. Natl. Acad. Sci. USA 85:2444 (1988), by computerized implementations of these algorithms (GAP, BESTFIT, FASTA, and TFASTA in the Wisconsin Genetics Software Package, Genetics Computer Group, 575 Science Dr., Madison, Wis.), or by visual inspection (see generally Ausubel et al., infra).
• BLAST algorithm is described in Altschul et al., J. Mol. Biol. 215:403-410 (1990).
• Software for perforating BLAST analyses is publicly available through the National Center for Biotechnology Information.
• non-stop or read-through is a mutation causing the removal of the natural stop codon.
• epitopope is the specific portion of an antigen typically bound by an antibody or T cell receptor.
• immunogenic is the ability to elicit an immune response, e.g., via T cells, B cells, or both.
• HLA binding affinity means affinity of binding between a specific antigen and a specific MHC allele.
• the term "bait” is a nucleic acid probe used to enrich a specific sequence of DNA or RNA from a sample.
• variant is a difference between a subject's nucleic acids and the reference human genome used as a control.
• variant call is an algorithmic determination of the presence of a variant, typically from sequencing.
• polymorphism is a germline variant, i.e., a variant found in all DNA-bearing cells of an individual.
• allele is a version of a gene or a version of a genetic sequence or a version of a protein.
• HLA type is the complement of HLA gene alleles.
• peptidome is the set of all peptides presented by MHC-I or MHC-I1 on the cell surface.
• the peptidome may refer to a property of a ceil or a collection of cells (e.g., the tumor peptidome, meaning the union of the peptidomes of all cells that comprise the tumor).
• EL1SPOT Enzyme-linked immunosorbent spot assay - which is a common method for monitoring immune responses in humans and animals.
• clinical factor refers to a measure of a condition of a subject, e.g., disease activity or severity.
• “Clinical factor” encompasses all markers of a subject's health status, including non-sample markers, and/or other characteristics of a subject, such as, without limitation, age and gender.
• a clinical factor can be a score, a value, or a set of values that can be obtained from evaluation of a sample (or population of samples) from a subject or a subject under a determined condition.
• a clinical factor can also be predicted by markers and/or other parameters such as gene expression surrogates.
• Clinical factors can include tumor type, tumor sub-type, and smoking history.
• MHC major histocompatibility complex
• HLA human leukocyte antigen, or the human MHC gene locus
• NGS next-generation sequencing
• PPV positive predictive value
• TSNA tumor-specific neoantigen
• FFPE formalin-fixed, paraffin- embedded
• NMD nonsense-mediated decay
• NSCLC non-small-cell lung cancer
• DC dendritic cell.
• one such method may comprise the steps of: obtaining at least one of exome, transcriptome or whole genome tumor nucleotide sequencing data from the tumor ceil 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 numeri cal 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 or cells present in the tumor, the set of numerical likelihoods
• the presentation model can comprise a statistical regression or a machine learning (e.g., deep learning) model trained on a set of reference data (also referred to as a training data set) comprising a set of corresponding labels, wherein the set of reference data is obtained from each of a plurality of distinct subjects where optionally some subjects can have a tumor, and wherein the set of reference data comprises at least one of: data representing exome nucleotide sequences from tumor tissue, data representing exome nucleotide sequences from normal tissue, data representing transcriptome nucleotide sequences from tumor tissue, data representing proteome sequences from tumor tissue, and data representing MHC peptidome sequences from tumor tissue, and data representing MHC peptidome sequences from normal tissue.
• a machine learning e.g., deep learning
• the reference data can further comprise mass spectrometry data, sequencing data, RNA sequencing data, and proteomics data for single-aliele cell lines engineered to express a predetermined MHC allele that are subsequently exposed to synthetic protein, normal and tumor human cell lines, and fresh and frozen primary samples, and T cell assays (e.g., ELISPOT).
• the set of reference data includes each form of reference data.
• the presentation model can comprise a set of features derived at least in part from the set of reference data, and wherein the set of features comprises at least one of alleie dependent-features and allele-independent features. In certain aspects each feature is included.
• the samples can also include cell lines engineered to express a plurality of MHC class I or class II alleles.
• the training data set may be generated based on performing or having performed nucleotide sequencing on a cell line to obtain at least one of exome, transcriptome, or whole genome sequencing data from the cell line, the sequencing data including at least one nucleotide sequence including an alteration.
• the training data set may further include data associated with peptide-MHC binding stability measurements for at least one of the isolated peptides.
• a method disclosed herein can also include encoding the peptide sequence using a one-hot encoding scheme.
• a tumor vaccine including a set of selected neoantigens selected by performing the method comprising the stpes 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 mutation 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 numeri cal 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
• the samples can also include cell lines engineered to express a single MHC class 1 or class II allele.
• a method disclosed herein can also include 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.
• the training data set may further include data associated with transcriptomes associated with the samples.
• the training data set may further include data associated with genomes associated with the samples.
• a method disclosed herein may also include logistically regressing the set of parameters.
• a method disclosed herein may also include encoding the training peptide sequences using a left-padded one-hot encoding scheme.
• a method disclosed herein can also include selecting a subset of neoantigens, wherein the subset of neoantigens is selected because each has an increased likelihood that it is presented on the cell surface of the tumor relative to one or more distinct tumor neoantigens.
• a method disclosed herein can also include selecting a subset of neoantigens, wherein the subset of neoantigens is selected because each has a decreased likelihood that it is subject to inhibition via central or peripheral tolerance relative to one or more distinct tumor neoantigens.
• a method disclosed herein can also include selecting a subset of neoantigens, wherein the subset of neoantigens is selected because each has a decreased likelihood that it is capable of inducing an autoimmune response to normal tissue in the subject relative to one or more distinct tumor neoantigens.
• Peptides with mutations or mutated polypeptides arising from for example, splice- site, frameshift, readthrough, or gene fusion mutations in tumor cells can be identified by sequencing DNA, RNA or protein in tumor versus normal cells.
• mutations can include previously identified tumor specific mutations. Known tumor mutations can be found at the Catalogue of Somatic Mutations in Cancer (COSMIC) database.
• DASH dynamic allele-specific hybridization
• MADGE microplate array diagonal gel electrophoresis
• pyrosequencing oligonucleotide-specific ligation
• TaqMan system as well as various DNA "chip” technologies
• Affymetrix SNP chips These methods utilize amplification of a target genetic region, typically by PCR. Still other methods, based on the generation of small signal molecules by invasive cleavage followed by mass spectrometry or immobilized padlock probes and rolling- circle amplification. Several of the methods known in the art for detecting specific mutations are summarized below.
• PCR based detection means can include multiplex amplification of a plurality of markers simultaneously. For example, it is well known in the art to select PCR primers to generate PCR products that do not overlap in size and can be analyzed simultaneously.
• RNA molecules can be detected by using a specialized exonuclease-resistant nucleotide, as disclosed, e.g., in Mundy, C. R. (U.S. Pat. No. 4,656,127).
• a primer complementary to the allelic sequence immediately 3' to the polymorphic site is permitted to hybridize to a target molecule obtained from a particular animal or human.
• the polymorphic site on the target molecule contains a nucleotide that is complementary to the particular exonuclease- resistant nucleotide derivative present, then that derivative will be incorporated onto the end of the hybridized primer. Such incorporation renders the primer resistant to exonuclease, and thereby permits its detection. Since the identity of the exonuclease-resistant derivative of the sample is known, a finding that the primer has become resistant to exonucleases reveals that the nucleotide(s) present in the polymorphic site of the target molecule is complementary to that of the nucleotide derivative used in the reaction. This method has the advantage that it does not require the determination of large amounts of extraneous sequence data.
• a solution-based method can be used for determining the identity of a nucleotide of a polymorphic site.
• WO91/02087 As in the Mundy method of U.S. Pat. No. 4,656, 127, a primer is employed that is complementary to allelic sequences immediately 3' to a polymorphic site. The method determines the identity of the nucleotide of that site using labeled dideoxynucleotide derivatives, which, if complementary to the nucleotide of the polymorphic site will become incorporated onto the terminus of the primer.
• Goelet, P. et al. PCT Appln. No. 92/157112.
• the method of Goelet, P. et al. uses mixtures of labeled terminators and a primer that is complementary to the sequence 3' to a polymorphic site.
• the labeled terminator that is incorporated is thus determined by, and complementary to, the nucleotide present in the polymorphic site of the target molecule being evaluated.
• Cohen et al. Fernch Patent 2,650,840; PCT Appln. No.
• die method of Goelet, P. et al. can be a heterogeneous phase assay, in which the primer or the target molecule is immobilized to a solid phase.
• GBA in that they utilize incorporation of labeled deoxynucieotides to discriminate between bases at a polymorphic site.
• the signal is proportional to the number of deoxynucieotides incorporated, polymorphisms that occur in runs of the same nucleotide can result in signals that are proportional to the length of the run (Syvanen, A.-C, et al., Amer. J. Hum. Genet. 52:46-59 (1993)).
• oligonucleotides 30-50 bases in length are covalently anchored at the 5' end to glass cover slips. These anchored strands perform two functions. First, they act as capture sites for the target template strands if the templates are configured with capture tails complementar ' to the surface-bound oligonucleotides. They also act as primers for the template directed primer extension that forms the basis of the sequence reading.
• the capture primers function as a fixed position site for sequence determination using multiple cycles of synthesis, detection, and chemical cleavage of the dye-linker to remove the dye. Each cycle consists of adding the polymerase/labeied nucleotide mixture, rinsing, imaging and cleavage of dye.
• polymerase is modified with a fluorescent donor molecule and immobilized on a glass slide, while each nucleotide is color-coded with an acceptor fluorescent moiety attached to a gamma-phosphate.
• the system detects the interaction between a fluorescently-tagged polymerase and a fluorescently modified nucleotide as the nucleotide becomes incorporated into the de novo chain .
• Other sequencing-by-synthesis technologies also exist.
• a plurality of nucleic acid molecules being sequenced is bound to a support (e.g., solid support).
• a capture sequence/universal priming site can be added at the 3' and/or 5' end of the template.
• the nucleic acids can be bound to the support by hybridizing the capture sequence to a complementary sequence covalently attached to the support.
• the capture sequence (also referred to as a universal capture sequence) is a nucleic acid sequence complementary to a sequence attached to a support that may dually serve as a universal primer.
• a member of a coupling pair (such as, e.g., antibody/antigen, receptor/ligand, or the avidin-biotin pair as described in, e.g., US Patent Application No. 2006/0252077) can be linked to each fragment to be captured on a surface coated with a respective second member of that coupling pair.
• sequence of the template is determined by the order of labeled nucleotides incorporated into the 3' end of the growing chain. This can be done in real time or can be done in a step-and-repeat mode. For real-time analysis, different optical labels to each nucleotide can be incorporated and multiple lasers can be utilized for stimulation of incorporated nucleotides.
• Sequencing can also include oilier massively parallel sequencing or next generation sequencing (NGS) techniques and platforms. Additional examples of massively parallel sequencing techniques and platforms are the Iliumina HiSeq or MiSeq, Thermo PGM or Proton, the Pac Bio RS II or Sequel, Qiagen's Gene Reader, and the Oxford Nanopore MinlON. Additional similar current massively parallel sequencing technologies can be used, as well as future generations of these technologies.
• a DNA or RNA sample can be obtained from a tumor or a bodily fluid, e.g., blood, obtained by known techniques (e.g. venipuncture) or saliva.
• nucleic acid tests can be performed on dry samples (e.g. hair or skin).
• a sample can be obtained for sequencing from a tumor and another sample can be obtained from normal tissue for sequencing where the normal tissue is of the same tissue type as the tumor.
• a sample can be obtained for sequencing from a tumor and another sample can be obtained from normal tissue for sequencing where the normal tissue is of a distinct tissue type relative to the tumor.
• Tumors can include one or more of lung cancer, melanoma, breast cancer, o varian 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.
• protein mass spectrometry can be used to identify or validate the presence of mutated peptides bound to MHC proteins on tumor cells.
• Peptides can be acid- eiuted from tumor ceils or from HLA molecules that are immunoprecipitated from tumor, and then identified using mass spectrometry.
• Neoantigens can include nucleotides or polypeptides.
• a neoantigen can be an RNA sequence that encodes for a polypeptide sequence.
• Neoantigens useful in vaccines can therefore include nucleotide sequences or polypeptide sequences.
• Disclosed herein are isolated peptides that comprise tumor specific mutations identified by the methods disclosed herein, peptides that comprise known tumor specific mutations, and mutant polypeptides or fragments thereof identified by methods disclosed herein.
• Neoantigen peptides can be described in the context of their coding sequence where a neoantigen includes the nucleotide sequence (e.g., DNA or RNA) that codes for the related polypeptide sequence.
• One or more polypeptides encoded by a neoantigen nucleotide sequence can comprise at least one of: a binding affinity with MHC with an IC50 value of less than lOOOnM, for MHC Class 1 peptides a length of 8-15, 8, 9, 10, 11, 12, 13, 14, or 15 ammo acids, presence of sequence motifs within or near the peptide promoting proteasome cleavage, and presence or sequence motifs promoting TAP transport.
• One or more neoantigens can be presented on the surface of a tumor.
• One or more neoantigens can be is immunogenic in a subject having a tumor, e.g., capable of eliciting a T cell response or a B ceil response in the subject.
• One or more neoantigens that induce an autoimmune response in a subject can be excluded from consideration in the context of vaccine generation for a subject having a tumor.
• the size of at least one neoantigenic peptide molecule can comprise, but is not limited to, about 5, about 6, about 7, about 8, about 9, about 10, about 11, about 12, about 13, about 14, about 15, about 16, about 17, about 18, about 19, about 20, about 21, about 22, about 23, about 24, about 25, about 26, about 27, about 28, about 29, about 30, about 31 , about 32, about 33, about 34, about 35, about 36, about 37, about 38, about 39, about 40, about 41, about 42, about 43, about 44, about 45, about 46, about 47, about 48, about 49, about 50, about 60, about 70, about 80, about 90, about 100, about 1 10, about 120 or greater amino molecule residues, and any range derivable therein.
• the neoantigenic peptide molecules are equal to or less than 50 amino acids.
• Neoantigenic peptides and polypeptides can be: for MHC Class I 15 residues or less in length and usually consist of between about 8 and about 11 residues, particularly 9 or 10 residues; for MHC Class II, 15-24 residues.
• a longer peptide can be designed in several ways.
• a longer peptide could consist of either: (1) individual presented peptides with an extensions of 2-5 amino acids toward the N- and C-terminus of each corresponding gene product; (2) a concatenation of some or all of the presented peptides with extended sequences for each.
• sequencing reveals a long (> 10 residues) neoepitope sequence present in the tumor (e.g.
• a longer peptide would consist of: (3) the entire stretch of novel tumor-specific amino acids— thus bypassing the need for compu tational or in vitro test-based selection of the strongest HLA -presented shorter peptide.
• use of a longer peptide allows endogenous processing by patient ceils and may lead to more effective antigen presentation and induction of T ee 13 responses.
• Neoantigenic peptides and polypeptides can be presented on an HLA protein.
• neoantigenic peptides and polypeptides are presented on an HLA protein with greater affinity than a wild-type peptide.
• a neoantigenic peptide or polypeptide can have an IC50 of at least less than 5000 nM, at least less than 1000 nM, at least less than 500 nM, at least less than 250 nM, at least less than 200 nM, at least less than 150 nM, at least less than 100 nM, at least less than 50 nM or less.
• neoantigenic peptides and polypeptides do not induce an autoimmune response and/or invoke immunological tolerance when administered to a subject.
• compositions comprising at least two or more neoantigenic peptides.
• the composition contains at least two distinct peptides. At least two distinct peptides can be derived from the same polypeptide. By distinct polypeptides is meant that the peptide vary by length, amino acid sequence, or both.
• the peptides are derived from any polypeptide known to or have been found to contain a tumor specific mutation. Suitable polypeptides from which the neoantigenic peptides can be derived can be found for example in the COSMIC database. COSMIC curates comprehensive information on somatic mutations in human cancer.
• the peptide contains the tumor specific mutation. In some aspects the tumor specific mutation is a driver mutation for a particular cancer type.
• Neoantigenic peptides and polypeptides having a desired activity or property can be modified to provide certain desired attributes, e.g., improved pharmacological
• neoantigenic peptide and polypeptides can be subject to various changes, such as substitutions, either conservative or non-conservative, where such changes might provide for certain advantages in their use, such as improved MHC binding, stability or presentation.
• conservative substitutions is meant replacing an amino acid residue with another which is biologically and/or chemically similar, e.g., one hydrophobic residue for another, or one polar residue for another.
• substitutions include combinations such as Gly, Ala; Val, He, Leu, Met; Asp, Glu; Asn, Gin; Ser, Thr; Lys, Arg; and Phe, Tyr.
• the effect of single amino acid substitutions may also be probed using D-amino acids.
• Such modifications can be made using well known peptide synthesis procedures, as described in e.g., Merrifield, Science 232:341-347 (1986), Barany & Merrifield, The Peptides, Gross & Meienhofer, eds. ( .Y., Academic Press), pp. 1-284 (1979); and Stewart & Young, Solid Phase Peptide Synthesis, (Rockford, 111, Pierce), 2d Ed. (1984),
• Modifications of peptides and polypeptides with various amino acid rnirnetics or unnatural amino acids can be particularly useful in increasing the stability of the peptide and polypeptide in vivo. Stability can be assayed in a number of ways. For instance, peptidases and various biological media, such as human plasma and serum, have been used to test stability. See, e.g., Verhoef et al., Eur. J. Drug Metab Pharmacokin. 11 :291-302 (1986). Half- life of the peptides can be conveniently determined using a 25% human serum (v/v) assay. The protocol is generally as follows.
• pooled human serum (Type AB, non-heat inactivated) is delipidated by centrifugation before use. The serum is then diluted to 25% with RPMI tissue culture media and used to test peptide stability. At predetermined time intervals a small amount of reaction solution is removed and added to either 6% aqueous trichloracetic acid or ethanol . The cloudy reaction sample is cooled (4 degrees C) for 15 minutes and then spun to pellet the precipitated serum proteins. The presence of the peptides is then determined by reversed-phase HPLC using stability-specific chromatography conditions.
• the peptides and polypeptides can be modified to provide desired attributes other than improved serum half-life. For instance, the ability of the peptides to induce CTL activity can be enhanced by linkage to a sequence which contains at least one epitope that is capable of inducing a T helper cell response.
• Immunogenic peptides/T helper conjugates can be linked by a spacer molecule.
• the spacer is typically comprised of relatively small, neutral molecules, such as amino acids or amino acid mimetics, which are substantially uncharged under physiological conditions.
• the spacers are typically selected from, e.g., Ala, Gly, or other neutral spacers of nonpolar amino acids or neutral polar amino acids.
• the optionally present spacer need not be comprised of the same residues and thus can be a hetero- or homo-oligomer.
• the spacer will usually be at least one or two residues, more usually three to six residues.
• the peptide can be linked to the T helper peptide without a spacer,
• a neoantigenic peptide can be linked to the T helper peptide either directly or via a spacer either at the amino or carboxy terminus of the peptide.
• the amino terminus of either the neoantigenic peptide or the T helper peptide can be acyiated.
• Exemplar ⁇ ' T helper peptides include tetanus toxoid 830-843, influenza 307-319, malaria circumsporozoite 382- 398 and 378-389,
• Proteins or peptides can be made by any technique known to those of skill in the art, including the expression of proteins, polypeptides or peptides through standard molecular biological techniques, the isolation of proteins or peptides from natural sources, or the ch emical synthesis of proteins or peptides.
• the nucleotide and protein, polypeptide and peptide sequences corresponding to various genes have been previously disclosed, and can be found at computerized databases known to those of ordinary skill in the art.
• One such database is the National Center for Biotechnology Information's Genbank and GenPept databases located at the National Institutes of Health website.
• the coding regions for known genes can be amplified and/or expressed using the techniques disclosed herein or as would be known to those of ordinary skill in the art.
• various commercial preparations of proteins, polypeptides and peptides are known to those of skill in the art.
• a neoantigen includes a nucleic acid (e.g. polynucleotide) that encodes a neoantigenic peptide or portion thereof.
• the polynucleotide can be, e.g., DNA, cDNA, PNA, CNA, RNA (e.g., mRNA), either single- and/or double-stranded, or native or stabilized forms of polynucleotides, such as, e.g., polynucleotides with a phosphorothiate backbone, or combinations thereof and it may or may not contain introns.
• a still further aspect provides an expression vector capable of expressing a polypeptide or portion thereof.
• Expression vectors for different cell types are well known in the art and can be selected without undue experimentation.
• DNA is inserted into an expression vector, such as a plasmid, in proper orientation and correct reading frame for expression. If necessary, DNA can be linked to the appropriate transcriptional and translational regulatory control nucleotide sequences recognized by the desired host, although such controls are generally available in the expression vector.
• the vector is then introduced into the host through standard techniques. Guidance can be found e.g. in Sambrook et ai. (1989) Molecular Cloning, A Laboratory Manual, Cold Spring Harbor Laboratory, Cold Spring Harbor, ⁇ . ⁇ .
• an immunogenic composition e.g., a vaccine composition, capable of raising a specific immune response, e.g., a tumor-specific immune response.
• Vaccine compositions typically comprise a plurality of neoantigens, e.g., selected usmg a method described herein.
• Vaccine compositions can also be referred to as vaccines.
• a vaccine can contain between 1 and 30 peptides, 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, or 30 different peptides, 6, 7, 8, 9, 10 11, 12, 13, or 14 different peptides, or 12, 13 or 14 different peptides.
• a vaccine can contain between I and 100 or more nucleotide sequences, 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, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54, 55, 56, 57, 58, 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 or more different nucleotide sequences, 6, 7, 8, 9, 10 11, 12, 13, or 14 different nucleot
• a vaccine can contain between 1 and 30 neoantigen sequences, 2, 3, 4, 5, 6, 7, 8, 9, 10, 1 1 , 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 ?
• different peptides and/or polypeptides or nucleotide sequences encoding them are selected so that the peptides and/or polypeptides capable of associating with different MHC molecules, such as different MHC class I molecule.
• one vaccine composition comprises coding sequence for peptides and/or polypeptides capable of associating with the most frequently occurring MHC class I molecules.
• vaccine compositions can comprise different fragments capable of associating with at least 2 preferred, at least 3 preferred, or at least 4 preferred MHC class I molecules.
• the vaccine composition can be capable of raising a specific cytotoxic T-ce!ls response and/or a specific helper T-cell response.
• a vaccine composition can further comprise an adjuvant and/or a carrier.
• a composition can be associated with a carrier such as e.g. a protein or an antigen-presenting cell such as e.g. a dendritic cell (DC) capable of presenting the peptide to a T-cell.
• a carrier such as e.g. a protein or an antigen-presenting cell such as e.g. a dendritic cell (DC) capable of presenting the peptide to a T-cell.
• a carrier such as e.g. a protein or an antigen-presenting cell such as e.g. a dendritic cell (DC) capable of presenting the peptide to a T-cell.
• DC dendritic cell
• Adjuvants are any substance whose admixture into a vaccine composition increases or otherwise modifies the immune response to a neoantigen.
• Carriers can be scaffold structures, for example a polypeptide or a polysaccharide, to which a neoantigen, is capable of being associated.
• adjuvants are conjugated covarrirely or non- covalently.
• Tlie ability of an adjuvant to increase an immune response to an antigen is typically manifested by a significant or substantial increase in an immune-mediated reaction, or reduction in disease symptoms.
• an increase in humoral immunity is typically manifested by a significant increase in the titer of antibodies raised to the antigen
• an increase in T-cell activity is typically manifested in increased cell proliferation, or cellular cytotoxicity, or cytokine secretion.
• An adjuvant may also alter an immune response, for example, by changing a primarily humoral or Th response into a primarily cellular, or Th response.
• Suitable adjuvants include, but are not limited to 1018 ISS, alum, aluminium salts, Amplivax, AS 15, BCG, CP-870,893, CpG7909, CyaA, dSLIM, GM-CSF, IC30, IC31, Imiqumiod, ImuFact IMP321, IS Patch, ISS, ISCOMATRIX, Juvlmmune, LipoVac, MF59, monophosphoryl lipid A, Montanide IMS 1312, Montanide ISA 206, Montanide ISA 50V, Montanide ISA-51, OK-432, OM-174, OM-197-MP-EC, ONTAK, PepTel vector system, PLG microparticles, resiquimod, SRL172, Virosornes and other Virus-like particles, YF-17D, VEGF trap, R848, beta-glucan, Pam3Cys, Aquila's QS21 stimulon (Aqui
• cytokines have been directly linked to influencing dendritic cell migration to lymphoid tissues (e.g., TNF-alpha), accelerating the maturation of dendritic cells into efficient antigen-presenting cells for T- lymphocytes (e.g., GM-CSF, IL-l and IL-4) (U.S. Pat. No. 5,849,589, specifically incorporated herein by reference in its entirety) and acting as immunoadjuvants (e.g., IL-l 2) (Gabriiovich D I, et al., J Immunother Emphasis Tumor Immunol. 1996 (6):414-418).
• CpG immunostimulatory oligonucleotides have also been reported to enhance the effects of adjuvants in a vaccine setting.
• Other TLR binding molecules such as RNA binding TLR 7, TLR 8 and/or TLR 9 may also be used.
• CpGs e.g. CpR, Idera
• Poly(I:C) e.g. polyi:CI2U
• non-CpG bacterial DNA e.g., bacterial DNA
• RNA as well as immunoactive small molecules and antibodies such as cyclophosphamide, sunitmib, bevacizumab, celebrex, NCX-4016, sildenafil, tadalafil, vardenafil, sorafinib, XL- 999, CP-547632, pazopamb, ZD2171, AZD2171, ipilimumab, tremelimumab, and SC58175, which may act therapeutically and/or as a adjuvant.
• the amounts and concentrations of adjuvants and additives can readily be determined by the skilled artisan without undue experimentation. Additional adjuvants include colony-stimulating factors, such as
• Granulocyte Macrophage Colony Stimulating Factor GM-CSF, sargramostim.
• a vaccine composition can comprise more than one different adjuvant.
• a therapeutic composition can comprise any adjuvant substance including any of the above or combinations thereof. It is also contemplated that a vaccine and an adjuvant can be administered together or separately in any appropriate sequence.
• a earner can be present independently of an adju vant.
• the function of a carrier can for example be to increase the molecular weight of in particular mutant to increase activity or immunogenicity, to confer stability, to increase the biological activity , or to increase serum half-life.
• a carrier can aid presenting peptides to T-celis.
• a carrier can be any suitable carrier known to the person skilled in the art, for example a protein or an antigen presenting cell.
• a carrier protein could be but is not limited to 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.
• the carrier is generally a physiologically acceptable carrier acceptable to humans and safe.
• tetanus toxoid and/or diptheria toxoid are suitable carriers.
• the carrier can be dextrans for example sepharose.
• Cytotoxic T-cells recognize an antigen in the form of a peptide bound to an MHC molecule rather than the intact foreign antigen itself.
• the MHC molecule itself is located at the cell surface of an antigen presenting cell.
• an activation of CTLs is possible if a trimeric complex of peptide antigen, MHC molecule, and APC is present.
• a vaccine composition additionally contains at least one antigen presenting cell.
• Neoantigens can also be included in viral vector-based vaccine platforms, such as vaccinia, fowlpox, self-replicating alpha vims, marabavirus, adenovirus (See, e.g., Tatsis et al., Adenoviruses, Molecular Therapy (2004) 10, 616— 629), or lenti virus, including but not limited to second, third or hybrid second/third generation lentivirus and recombinant lentivirus of any generation designed to target specific cell types or receptors (See, e.g., Hu et al., Immunization Delivered by Lenti iral Vectors for Cancer and Infectious Diseases, Immunol Rev.
• viral vector-based vaccine platforms such as vaccinia, fowlpox, self-replicating alpha vims, marabavirus, adenovirus (See, e.g., Tatsis et al., Adenoviruses, Molecular Therapy (2004) 10, 616— 629),
• this approach can deliver one or more nucleotide sequences that encode one or more neoantigen peptides.
• the sequences may be flanked by non-mutated sequences, may be separated by linkers or may be preceded with one or more sequences targeting a subcellular compartment (See, e.g., Gros et al.. Prospective identification of neoantigen-specific lymphocytes in the peripheral blood of melanoma patients, Nat Med. (2016) 22 (4):433-8, Stronen et al., Targeting of cancer neoantigens with donor-derived T cell receptor repertoires, Science.
• Truncal peptides meaning those presented by all or most tumor subclones, will be prioritized for inclusion into the vaccine. 3" '
• further peptides can be prioritized by estimating the number and identi ty of tum or subclones and choosing peptides so as to maximize the number of tumor subclones covered by the vaccine. 54 IV.A.2. Neoantigen prioritization
• neoantigen filters After ail of the above above neoantigen filters are applied, more candidate neoantigens may still be available for vaccine inclusion than the vaccine technology can support. Additionally, uncertainty about various aspects of the neoantigen analysis may remain and tradeoffs may exist between different properties of candidate vaccine neoantigens.
• an integrated multi-dimensional model can be considered that places candidate neoantigens in a space with at least the following axes and optimizes selection using an integrative approach.
• presen tation of a set of neoantigens may lower the probability that a tumor will escape immune attack via downregulation or mutation of HLA molecules
• a subject has been diagnosed with cancer or is at risk of developing cancer.
• a subject can be a human, dog, cat, horse or any animal in which a tumor specific immune response is desired.
• a tumor can be any solid tumor such as breast, ovarian, prostate, lung, kidney, gastric, colon, testicular, head and neck, pancreas, brain, melanoma, and other tumors of tissue organs and hematological tumors, such as lymphomas and leukemias, including acute myelogenous leukemia, chronic myelogenous leukemia, chronic lymphocytic leukemia, T cell lymphocytic leukemia, and B cell lymphomas.
• a neoantigen can be administered in an amount sufficient to induce a CTL response.
• a neoantigen can be administered alone or in combination with other therapeutic agents.
• the therapeutic agent is for example, a chemotherapeutic agent, radiation, or immunotherapy . Any suitable therapeutic treatment for a particular cancer can be administered.
• a subject can be further administered an anti- immunosuppressive/immunostmiuiaiory agent such as a checkpoint inhibitor.
• the subject can be further administered an anti-CTLA antibody or anti-PD-i or anti-PD-Ll .
• Blockade of CTLA-4 or PD-LI by antibodies can enhance the immune response to cancerous cells in the patient.
• CTLA-4 blockade has been shown effective when following a vaccination protocol.
• a neoantigen or its variant can be prepared for intravenous (i.v.) injection, sub-cutaneous (s.c.) injection, intradermal (i.d.) injection, intraperitoneal (i.p.) injection, intramuscular (i.m.) injection.
• Methods of injection include s.c, i.d., i .p., i.m ., and i.v.
• Methods of DNA or RNA injection include i.d., i.m., s.c, i.p. and i.v.
• Other methods of administration of the vaccine composition are known to those skilled in the art.
• a vaccine can be compiled so that the selection, number and/or amount of neoantigens present in the composition is/are tissue, cancer, and/or patient-specific For instance, the exact selection of peptides can be guided by expression patterns of the parent proteins in a given tissue. The selection can be dependent on the specific type of cancer, the status of the disease, earlier treatment regimens, the immune status of the patient, and, of course, the HLA-haplotype of the patient. Furthermore, a vaccine can contain individualized components, according to personal needs of the particular patient. Examples include varying the selection of neoantigens according to the expression of the neoantigen in the particular patient or adjustments for secondary treatments following a first round or scheme of treatment.
• neoantigens with similar normal self-peptides that are expressed in high amounts in normal tissues can be avoided or be present in low amounts in a composition described herein.
• the respective pharmaceutical composition for treatment of this cancer can be present in high amounts and/or more than one neoantigen specific for this particularly neoantigen or pathway of this neoantigen can be included.
• compositions comprising a neoantigen can be administered to an individual already suffering from cancer.
• compositions are administered to a patient in an amount sufficient to elicit an effective CTL response to the tumor antigen and to cure or at least partially arrest symptoms and/or complications.
• An amount adequate to accomplish this is defined as "therapeutically effective dose.” Amounts effective for this use will depend on, e.g., the composition, the manner of administration, the stage and severity of the disease being treated, the weight and general state of health of the patient, and the judgment of the prescribing physician. It should be kept in mind that compositions can generally be employed in serious disease states, that is, life-threatening or potentially life threatening situations, especially when the cancer has metastasized. In such cases, in view of the minimization of extraneous substances and the relative nontoxic nature of a neoantigen, it is possible and can be felt desirable by the treating physician to administer substantial excesses of these compositions.
• administration can begin at the detection or surgical removal of tumors. This is followed by boosting doses until at least symptoms are substantially abated and for a period thereafter.
• compositions for therapeutic treatment are intended for parenteral, topical, nasal, oral or local administration.
• a pharmaceutical compositions can be administered parenterally, e.g., intravenously, subcutaneously, intraderrnally, or intramuscularly.
• the compositions can be administered at the site of surgical exiseion to induce a local immune response to the tumor.
• compositions for parenteral administration which comprise a solution of the neoantigen and vaccine compositions are dissolved or suspended in an acceptable carrier, e.g., an aqueous carrier.
• aqueous carriers can be used, e.g., water, buffered water, 0.9% saline, 0.3% glycine, hyaluronic acid and the like. These compositions can be sterilized by conventional, well known sterilization techniques, or can be sterile filtered. The resulting aqueous solutions can be packaged for use as is, or lyophilized, the lyophilized preparation being combined with a sterile solution prior to administration.
• compositions may contain pharmaceutically acceptable auxiliary substances as required to approximate physiological conditions, such as pH adjusting and buffering agents, tonicity adjusting agents, wetting agents and the like, for example, sodium acetate, sodium lactate, sodium chloride, potassium chloride, calcium chloride, sorbitan monolaurate, triethanolamine oleate, etc.
• auxiliary substances such as pH adjusting and buffering agents, tonicity adjusting agents, wetting agents and the like, for example, sodium acetate, sodium lactate, sodium chloride, potassium chloride, calcium chloride, sorbitan monolaurate, triethanolamine oleate, etc.
• Neoantigens can also be administered via liposomes, which target them to a particular cells tissue, such as lymphoid tissue. Liposomes are also useful in increasing half- life. Liposomes include emulsions, foams, micelles, insoluble monolayers, liquid crystals, phospholipid dispersions, lamellar layers and the like. In these preparations the neoantigen to be delivered is incorporated as part of a liposome, alone or in conjunction with a molecule which binds to, e.g., a receptor prevalent among lymphoid cells, such as monoclonal antibodies which bind to the CD45 antigen, or with other therapeutic or immunogenic compositions.
• a receptor prevalent among lymphoid cells such as monoclonal antibodies which bind to the CD45 antigen, or with other therapeutic or immunogenic compositions.
• liposomes filled with a desired neoantigen can be directed to the site of lymphoid cells, where the liposomes then deliver the selected therapeutic/immunogenic compositions.
• Liposomes can be formed from standard vesicle-forming lipids, which generally include neutral and negatively charged phospholipids and a sterol, such as cholesterol. The selection of lipids is generally guided by consideration of, e.g., liposome size, acid lability and stability of the liposomes in the blood stream. A variety of methods are available for preparing liposomes, as described in, e.g., Szoka et al., Ann. Rev. Biophys. Bioeng. 9; 467 ( 1980), U.S. Pat. Nos. 4,235,871, 4,501,728, 4,501,728, 4,837,028, and 5,019,369,
• a ligand to be incorporated into the liposome can include, e.g., antibodies or fragments thereof specific for cell surface determinants of the desired immune system ceils.
• a liposome suspension can be administered intravenously, locally, topically, etc. in a dose which varies according to, inter alia, the manner of administration, the peptide being delivered, and the stage of the disease being treated.
• nucleic acids encoding a peptide and optionally one or more of the peptides described herein can also be administered to the patient.
• a number of methods are conveniently used to deliver the nucleic acids to the patient.
• the nucleic acid can be delivered directiy, as "naked DNA". This approach is described, for instance, in Wolff et al.. Science 247: 1465-1468 (1990) as well as U.S. Pat. Nos. 5,580,859 and 5,589,466.
• the nucleic acids can also be administered using ballistic delivery as described, for instance, in U.S. Pat. No. 5,204,253.
• Particles comprised solely of DNA can be administered.
• DNA can be adhered to particles, such as gold particles.
• Approaches for delivering nucleic acid sequences can include viral vectors, mRNA vectors, and DNA vectors with or without electroporation.
• nucleic acids can also be delivered complexed to cationic compounds, such as cationic lipids.
• cationic compounds such as cationic lipids.
• Lipid-mediated gene delivery methods are described, for instance, in
• Neoantigens can also be included in viral vector-based vaccine platforms, such as vaccinia, fowlpox, self-replicating alphavirus, marabavirus, adenovirus (See. e.g., Tatsis et al., Adenoviruses, Molecular Therapy (2004) 10, 616— 629), or ienti virus, including but not limited to second, third or hybrid second/third generation lenti virus and recombinant lentivirus of any generation designed to target specific cell types or receptors (See, e.g., Hu et al., Immunization Delivered by Lentiviral Vectors for Cancer and Infectious Diseases, Immunol Rev.
• viral vector-based vaccine platforms such as vaccinia, fowlpox, self-replicating alphavirus, marabavirus, adenovirus (See. e.g., Tatsis et al., Adenoviruses, Molecular Therapy (2004) 10, 616— 629), or ienti virus,
• this approach can deliver one or more nucleotide sequences that encode one or more neoantigen peptides.
• the sequences may be flanked by non-mutated sequences, may be separated by linkers or may be preceded with one or more sequences targeting a subcellular compartment (See, e.g., Gros et al., Prospective identification of neoantigen-specific lymphocytes in the peripheral blood of melanoma patients, Nat Med. (2016) 22 (4):433-8, Stronen et al..
• the minigene sequence is converted to DNA by assembling oligonucleotides that encode the plus and minus strands of the minigene. Overlapping oligonucleotides (30-100 bases long) are synthesized, phosphorylated, purified and annealed under appropriate conditions using well known techniques. The ends of the oligonucleotides are joined using T4 DNA ligase. This synthetic minigene, encoding the CTL epitope polypeptide, can then cloned into a desired expression vector.
• Single-nucleotide variants and indels will be detected from tumor DNA, tumor RNA and normal DNA with a suite of tools including: programs based on comparisons of tumor and normal DNA, such as Strelka 21 and Mutect 2 ; and programs that incorporate tumor DNA, tumor RNA and normal DNA, such as IJNCeqR, which is particularly advantageous in low-purity samples J .
• IP immunoprecipitation
• FIG. 2A is an overview of an environment 100 for identifying likelihoods of peptide presentation in patients, in accordance with an embodiment.
• the environment 100 provides context in order to introduce a presentation identification system 160, itself including a presentation information store 165.
• the presentation identification system 160 is able to receive nucleotide sequences of candidate neoantigens associated with a set of MHC alleles from tumor cells of a patient 110 and determine likelihoods that the candidate neoantigens will be presented by one or more of the associated MHC alleles of the tumor and/or induce immunogenic responses in the immune system of the patient 110.
• Those candidate neoantigens with high likelihoods as determined by system 160 can be selected for inclusion in a vaccine 118, such an anti-tumor immune response can be elicited from the immune system of the patient 110 providing the tumor cells.
• the presentation identification system 160 determines presentation likelihoods through one or more presentation models. Specifically, the presentation models generate likelihoods of whether given peptide sequences will be presented for a set of associated MHC alleles, and are generated based on presentation information stored in store 165. For example, the presentation models may generate likelihoods of whether a peptide sequence
• the presentation information 165 contains information on whether peptides bind to different types of MHC alleles such that those peptides are presented by MHC alleles, which in the models is determined depending on positions of amino acids in the peptide sequences.
• the presentation model can predict whether an unrecognized peptide sequence will be presented in association with an associated set of MHC alleles based on the presentation information 165.
• FIG. 2 illustrates a method of obtaining presentation information, in accordance with an embodiment.
• the presentation information 165 includes two general categories of information: ailele-interacting information and allele-noninteracting information.
• Allele- interacting information includes information that influence presentation of peptide sequences that are dependent on the type of MHC allele.
• Allele-noninteracting information includes information that influence presentation of peptide sequences that are independent on the type of MHC allele.
• Ailele-interacting information primarily includes identified peptide sequences that are known to have been presented by one or more identified MHC molecules from humans, mice, etc. Notably, this may or may not include data obtained from tumor samples.
• the presented peptide sequences may be identified from cells that express a single MHC allele. In tins case the presented peptide sequences are generally collected from smgle-aiiele cell lines that are engineered to express a predetermined MHC allele and that are subsequently- exposed to synthetic protein. Peptides presented on the MHC allele are isolated by techniques such as acid-elution and identified through mass spectrometry. FIG.
• the direct association between a presented peptide and the MHC protein to which it was bound to is definitively known.
• Allele-interacting information can also include measurements or predictions of binding affinity between a given MHC allele and a given peptide.
• One or more affinity models can generate such predictions.
• presentation information 165 may include a binding affinity prediction of ⁇ between the peptide YEMFNDKSF and the allele HLA-A*01 :01. Few peptides with IC50 > lOOOnm are presented by the MHC, and lower IC5G values increase the probability of presentation.
• Allele-interacting information can also include the measured or predicted rate of the formation reaction for the peptide-MHC complex. Complexes that form at a higher rate are more likely to be presented on the cell surface at high concentration.
• Allele-interacting information can also include the expression or activity levels of proteins involved in the process of post-translational modification, e.g., kinases (as measured or predicted from. RNA seq, mass spectrometry, or other methods).
• proteins involved in the process of post-translational modification e.g., kinases (as measured or predicted from. RNA seq, mass spectrometry, or other methods).
• Allele-interacting information can also include the expression levels of the particular MHC allele in the individual in question (e.g. as measured by RNA-seq or mass spectrometry)- Peptides that bind most strongly to an MHC allele that is expressed at high levels are more likely to be presented than peptides that bind most strongly to an MHC allele that is expressed at a low level .
• Allele-interacting information can also include the overall neoantigen encoded peptide-sequence-independent probability of presentation by the particular MHC allele in other individuals who express the particular MHC allele.
• Allele-interacting information can also include the overall 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 individuals.
• HLA-C molecules are typically expressed at lower levels than HLA-A or HLA-B molecules, and consequently, presentation of a peptide by HLA-C is a priori less probable than presentation by HLA-A or HLA-B 11.
• Any MHC allele-noninteracting information listed in the below section can also be modeled as an MHC alieie-interacting information. VII.B.2. Allele-noninteracting Information
• Allele-noninteracting infonnation can include C-terminal sequences flanking the neoantigen encoded peptide within its source protein sequence.
• C-terminal flanking sequences may impact proteasomal processing of peptides.
• the C-terminal flanking sequence is cleaved from the peptide by the proteasome before the peptide is transported to the endoplasmic reticulum and encounters MHC alleles on the surfaces of cells. Consequently, MHC molecules receive no infonnation about the C-terminal flanking sequence, and thus, the effect of the C-terminal flanking sequence cannot vary depending on MHC allele type.
• presentation information 165 may include the C-terminal flanking sequence FOEIFNDKSLDKFJI of the presented peptide FJIEJFOESS identified from the source protein of the peptide.
• Allele-noninteracting information can also include mRNA quantification measurements.
• mRNA quantification data can be obtained for the same samples that provide the mass spectrometry- training data.
• RNA expression was identified to be a strong predictor of peptide presentation.
• the mRNA quantification measurements are identified from software tool RSEM. Detailed implementation of the RSEM software tool can be found at Bo Li and Colin N. Dewey, RSEM: accurate transcript quantification from RNA-Seq data with or without a reference genome. BMC Bioinformatics, 12:323, August 2011.
• the mRN A quantification is measured in units of fragments per kilobase of tran script per Million mapped reads (FPKM).
• Allele-noninteracting information can also include the N-terminal sequences flanking the peptide within its source protein sequence.
• Allele-noninteracting information can also include the level of expression of the proteasome, imniunoproteasome, thymoproteasome, or other proteases in the tumor cells (which may be measured by RNA-seq, proteome mass spectrometry, or
• Allele-noninteracting information can also include the expression of the source gene of the peptide (e.g., as measured by RMA-seq or mass spectrometry). Possible optimizations include adjusting the measured expression to account for the presence of stromal cells and tumor-infiltrating lymphocytes within the tumor sample. Peptides from more highly expressed genes are more likely to be presented. Peptides from genes with undetectable levels of expression can be excluded from consideration.
• Allele-noninteracting information can also include the probability that the source mRNA of the neoaritigen encoded peptide will be subject to nonsense-mediated decay as predicted by a model of nonsense-mediated decay, for example, the model from Rivas et al, Science 2015.
• Allele-noninteracting information can also include a comprehensive catalog of features of the source protein as given in e.g. uniProt or PDB
• These features may include, among others: the secondary and tertiary structures of the protein, subcellular localization 11, Gene ontology (GO) terms. Specifically, this information may contain annotations that act at the level of the protein, e.g., 5' UTR length, and annotations that act at the level of specific residues, e.g., helix motif between residues 300 and 310. These features can also include turn motifs, sheet motifs, and disordered residues.
• Allele-noninteracting information can also include features describing the presence or absence of a presentation hotspot at the position of the peptide in the source protein of the peptide.
• Allele-noninteracting information can also include the probability of presentation of peptides from the source protein of the peptide in question in other individuals (after adju sting for the expression level of the source protein in those individuals and the influence of the different HLA types of those individuals).
• Allele-noninteracting information can also include the copy number of the source gene of the peptide in the tumor cells.
• peptides from, genes that are subject to homozygous deletion in tumor cells can be assigned a probability of presentation of zero.
• Allele-noninteracting information can also include the expression level of TAP in the tumor cells (w hich may be measured by RNA-seq, proteome mass spectrometry, immunohistochemistry). Higher TAP expression levels increase the probability of presentation of all peptides.
• Allele-noninteracting information can also include known functionality of HLA alleles, as reflected by, for instance HLA allele suffixes.
• HLA allele suffixes For example, the N suffix in the allele name HLA-A*24:09N indicates a null allele that is not expressed and is therefore unlikely to present epitopes; the full HLA allele suffix nomenclature is described at ht1ps://ww ⁇ ebi.ac.u pd/imgt/hla/nomenclature/suffixes.html.
• Allele-noninteracting information can also include clinical tumor subtype (e.g., squamous lung cancer vs. non-squamous).
• clinical tumor subtype e.g., squamous lung cancer vs. non-squamous.
• Allele-noninteracting information can also include history of sunburn, sun exposure, or exposure to other mutagens.
• Allele-noninteracting information can also include the typical expression of the source gene of the peptide in the relevant tumor type or clinical subtype, optionally stratified by driver mutation. Genes that are typically expressed at high levels in the relevant tumor type are more likely to be presented. [00300] Allele-noninteracting information can also include the frequency of the mutation in all tumors, or in tumors of the same type, or in tumors from individuals with at least one shared MHC allele, or in tumors of the same type in individuals with at least one shared MHC allele,
• the list of features used to predict a probability of presentation may also include the annotation of the mutation (e.g., missense, read-through, frameshift, fusion, etc.) or whether the mutation is predicted to result in nonsense-mediated decay (NMD).
• NMD nonsense-mediated decay
• peptides from protein segments that are not translated in tumor cells due to homozygous early-stop mutations can be assigned a probability of presentation of zero. NMD results in decreased mRNA translation, which decreases the probability of presentation.
• FIG. 3 is a high-level block diagram illustrating the computer logic components of the presentation identification system 160, according to one embodiment.
• the presentation identification system 160 includes a data management module 312, an encoding module 314, a training module 3 6, and a prediction module 320.
• the presentation identification system 160 is also comprised of a training data store 170 and a presentation models store 175.
• Some embodiments of the model management system 160 have different modules than those described here. Similarly, the functions can be distributed among the modules in a different manner than is described here.
• the data management module 312 generates sets of training data 170 from the presentation information 165.
• Each set of training data contains a plurality of data instances, in which each data instance contains a set of independent variables t that include at least a presented or non-presented peptide sequence p one or more associated MHC alleles d associated with the peptide sequence p and a dependent variable V that represents information that the presentation identification system 160 is interested in predicting for new values of independent variables.
• the dependent variable V is a binary label indicating whether peptide p l was presented by the one or more associated MHC alleles d.
• the dependent variable f can represent any other kind of information that the presentation identification system 160 is interested in predicting dependent on the independent variables z l .
• the dependent variable / may also be a numerical value indicating the mass spectrometry ion current identified for the data instance,
• the peptide sequence p' for data instance is a sequence of k amino acids, in which ki may vary between data instances / ' within a range. For example, that range may be 8-15 for MHC class I or 9-30 for MHC class ⁇ , In one specific implementation of system 160, all peptide sequences p' in a training data set may have the same length, e.g. 9. The number of amino acids in a peptide sequence may vary depending on the type of MHC alleles (e.g., MHC alleles in humans, etc.). The MHC alleles d for data instance i indicate which MHC alleles were present in association with the corresponding peptide sequence p l ,
• the data management module 312 may also include additional allele-interacting variables, such as binding affinity b' and stability s' predictions in conjunction with the peptide sequences p 1 and associated MHC alleles d contained in the training data 170.
• the training data 170 may contain binding affinity predictions b l between a peptide p and each of the associated MHC molecules indicated in d.
• the training data 170 may contain stability predictions s l for each of the MHC alleles indicated in d.
• the data management module 312 may also include allele-noninteracting variables w such as C-terminai flanking sequences and mRNA quantification measurements in conjunction with the peptide sequences p
• the data management module 312 also identifies peptide sequences that are not presented by MHC alleles to generate the training data 170. Generally, this involves identify ing the 'longer" sequences of source protein that include presented peptide sequences prior to presentation. When the presentation information contains engineered cell lines, the data management module 312 identifies a series of peptide sequences in the synthetic protein to which the cells were exposed to that were not presented on MHC alleles of the ceils.
• the data management module 312 identifies source proteins from which presented peptide sequences originated from, and identifies a series of peptide sequences in the source protein that were not presented on MHC alleles of the tissue sample cells.
• the data management module 312 may also artificially generate peptides with random sequences of amino acids and identify the generated sequences as peptides not presented on MHC alleles. This can be accomplished by randomly generating peptide sequences allows the data management module 312 to easily generate large amounts of synthetic data for peptides not presented on MHC alleles. Since in reality, a small percentage of peptide sequences are presented by MHC alleles, the synthetically generated peptide sequences are highly likely not to have been presented by MHC alleles even if they were included in proteins processed by cells.
• FIG. 4 illustrates an example set of training data ⁇ 70 ⁇ , according to one embodiment.
• the first 3 data instances in the training data 170 A indicate peptide presentation information from a single-allele cell line involving the allele HLA-C*01 :03 and 3 peptide sequences QCEIOWARE, F1EUHFW1, and FEWRHRJTRUJR.
• the fourth data instance in the training data 170 A indicates peptide information from a multiple-allele cell line involving the alleles HLA-B*07:02, HLA-C*01 :03, HLA-A*01 :01and a peptide sequence QIEJOEIJE.
• the first data instance indicates that peptide sequence QCEIOWARE was not presented by the allele HLA-C*01 :03.
• the peptide sequence may be randomly generated by the data management module 312 or identified from source protein of presented peptides.
• the training data 170 A also includes a binding affinity prediction of lOOOnM and a stability prediction of a half-life of lh for the peptide sequence-allele pair.
• the training data ⁇ 70 ⁇ also includes allele-noninteracting variables, such as the C-terminal flanking sequence of the peptide FJELFISBOSJFIE, and a mRNA quantification measurement of 10 2 FPKM.
• the fourth data instance indicates thai peptide sequence QIEJOEIJE was presented by one of the alleles HLA-B*07:02, HLA- C*01 :03, or HLA-A* 01 :01.
• the training data 170 A also includes binding affinity predictions and stability predictions for each of the alleles, as well as the C-flanking sequence of the peptide and the mRNA quantification measurement for the peptide.
• the encoding module 314 encodes information contained in the training data 170 into a numerical representation that can be used to generate the one or more presentation models.
• the encoding module 314 one-hot encodes sequences (e.g., peptide sequences or C-terminal flanking sequences) over a predetermined 20-letter amino acid alphabet.
• sequences e.g., peptide sequences or C-terminal flanking sequences
• a peptide sequence p" with h, amino acids is represented as a ro vector of 20-k, elements, where a single element among p l 2o-(/-i)+j, ⁇ '2 ⁇ - ⁇ - ⁇ )+2, ...
• p'lo-j that corresponds to the alphabet of the amino acid at the j-th position of the peptide sequence has a value of 1. Otherwise, the remaining elements have a value of 0.
• a given alphabet ⁇ A, C, D, E, F, G, H, I, K, L, M, N, P, Q, R.
• the peptide sequence EAF of 3 amino acids for data instance may be represented by the row vector of 60 elements /
• the C-terminal flanking sequence d can be similarly encoded as described above, as well as the protein sequence i , for MHC alleles, and other sequence data in the presentation information.
• the encoding module 314 may further encode the peptides into equal-length vectors by adding a PAD character to extend the predetermined alphabet. For example, this may be performed by left-padding the peptide sequences with the PAD character until the length of the peptide sequence reaches the peptide sequence with the greatest length in the training data 170.
• the encoding module 314 numerically represents each sequence as a row vector of (20+1)- k max elements.
• the encoding module 314 also encodes the one or more MHC alleles d for data instance / as a row vector of m elements, in which each element /; /. 2, ... , «? corresponds to a unique identified MHC allele.
• the elements corresponding to the MHC alleles identified for the data instance i have a value of 1 . Otherwise, the remaining elements have a value of 0.
• HLA-A*01 :01, HLA-C*01 :08, HLA-B*07:02, HLA-C*01 :03 ⁇ may be represented by the row vector of 4 elements .
• the number of MHC allele types can be hundreds or thousands in practice.
• each data instance typically contains at most 6 different MHC allele types in association with the peptide sequence ⁇ ,.
• the encoding module 314 also encodes the label y, for each data instance as a binary variable having values from the set of ⁇ 0, 1 ⁇ , in which a value of 1 indicates that peptide x' was presented by one of the associated MHC alleles a and a value of 0 indicates that peptide x l was not presented by any of the associated MHC alleles a .
• the dependent variable v represents the mass spectrometry ion current
• the encoding module 314 may additionally scale the values using various functions, such as the log function having a range of [- ⁇ , ⁇ for ion current values between [0, ⁇ ] .
• the encoding module 314 may represent a pair of allele-interacting variables ⁇ 3 ⁇ 4 ! for peptide /?, and an associated MHC allele h as a row vector in which numerical representations of allele-interacting variables are concatenated one after the other.
• the encoding module 314 may represent ⁇ 1 ⁇ 4' as a row vector equal to ⁇ p p l >/ ], ⁇ p l S ] , or where b ⁇ is the binding affinity prediction for peptide p, and associated MHC allele h, and similarly for 3 ⁇ 4 for stability.
• one or more combination of allele-interacting variables may be stored individually (e.g., as individual vectors or matrices).
• the encoding module 314 represents binding affinity information by incorporating measured or predicted values for binding affinity in the allele-interacting variables ⁇ 3 ⁇ 4'.
• the encoding module 314 represents binding stability information by incorporating measured or predicted values for binding stability in the allele-interacting variables xn ,
• the encoding module 314 represents binding on-rate information by incorporating measured or predicted values for binding on-rate in the allele-interacting variables
• the vector 7* can be included in the allele-interacting variables x / .
• the encoding module 314 represents RNA expression information of MHC alleles by incorporating RNA-seq based expression levels of MHC alleles in the allele-interacting variables ⁇ 3 ⁇ 4'.
• the encoding module 314 may represent the allele-noninteracting variables w l as a row vector in which numerical representations of allele-noninteracting variables are concatenated one after the other.
• w l may be a row vector equal to
• w 1 is a row vector representing any other allele-noninteracting variables in addition to the C-terminai flanking sequence of peptide p' and the rnRNA quantification measurement m associated with the peptide.
• w 1 is a row vector representing any other allele-noninteracting variables in addition to the C-terminai flanking sequence of peptide p' and the rnRNA quantification measurement m associated with the peptide.
• one or more combination of allele-noninteracting variables may be stored individually (e.g., as individual vectors or matrices).
• the encoding module 314 represents turnover rate of source protein for a peptide sequence by incorporating the turnover rate or half-life in the allele- noninteracting variables w'.
• the encoding module 314 represents length of source protein or isoform by incorporating the protein length in the allele-noninteracting variables w'.
• the encoding module 314 represents activation of
• immunoproteasome by incorporating the mean expression of the immunoproteasome-specific proteasome subunits including the ⁇ 5,- subunits in the allele-noninteracting variables w
• the encoding module 314 represents the RNA-seq abundance of the source protein of the peptide or gene or transcript of a peptide (quantified in units of FPKM, TPM by techniques such as RSEM) can be incorporating the abundance of the source protein in the allele-noninteracting variables .
• the encoding module 314 represents the probability that the transcript of origin of a peptide will undergo nonsense-mediated decay (NMD) as estimated by the model in, for example, Rivas et. al. Science, 2015 by incorporating this probability in the allele-noninteracting variables w'.
• NMD nonsense-mediated decay
• the encoding module 314 represents the activation status of a gene module or pathway assessed via RNA-seq by, for example, quantifying expression of the genes in the pathway in units of TPM using e.g., RSEM for each of the genes in the pathway then computing a summary statistics, e.g., the mean, across genes in the pathway.
• the mean can be incorporated in the allele-noninteracting variables w
• the encoding module 314 represents the copy number of the source gene by incorporating the copy number in the allele-noninteracting variables w .
• the encoding module 314 represents the TAP binding affinity by including the measured or predicted TAP binding affinity (e.g., in nanomolar units) in the allele-noninteracting variables w
• the encoding module 314 represents TAP expression levels by including TAP expression levels measured by RNA-seq (and quantified in units of TPM by e.g., RSEM) in the allele-noninteracting variables w 1 .
• the encoding module 314 represents tumor type as a length-one one-hot encoded vector over the alphabet of tumor types (e.g., NSCLC, melanoma, colorectal cancer, etc). These one-hot-encoded variables can be included in the allele-noninteracting variables w'.
• the encoding module 314 represents MHC allele suffixes by treating 4-digit HLA alleles with different suffixes.
• HLA-A*24:09N is considered a different allele from HLA-A*24:09 for the purpose of the model.
• the probability of presentation by an N-suffixed MHC allele can be set to zero for all peptides, because HLA alleles ending in the N suffix are not expressed.
• the encoding module 314 represents tumor subtype as a length- one one-hot encoded vector over the alphabet of tumor subtypes (e.g., lung adenocarcinoma, lung squamous cell carcinoma, etc). These onehot-encoded variables can be included in the allele-noninteracting variables w ! .
• smoking history can be encoded as a length-one one-hot-enocded variable over an alphabet of smoking severity. For example, smoking status can be rated on a 1-5 scale, where 1 indicates nonsmokers, and 5 indicates current heavy smokers. Because smoking history is primarily relevant to lung tumors, when training a model on multiple tumor types, this variable can also be defined to be equal to 1 if the patient has a history of smoking and the tumor type is lung tumors and zero otherwise.
• the encoding module 314 represents distribution of expression levels of a particular gene or transcript for each gene or transcript in the human genome as summary statistics (e,g., mean, median) of distribution of expression levels by using reference databases such as TCGA.
• summary statistics e.g., mean, median
• TCGA reference databases
• the encoding module 314 represents mutation type as a length-one one-hot-encoded variable over the alphabet of mutation types (e.g., missense, frameshift, NMD-inducing, etc). These onehot-encoded variables can be included in the allele- nonmteracting variables w l .
• the encoding module 314 represents protein-level features of protein as the value of the annotation (e.g., 5' UTR length) of the source protein in the allele- noninteracting variables w
• the encoding module 314 represents residue- level annotations of the source protein for peptide p K by including an indicator variable, that is equal to 1 if peptide p k overlaps with a helix motif and 0 otherwise, or that is equal to I if peptide / is completely contained with within a helix motif in the allele-noninteracting variables w ! .
• a feature representing proportion of residues in peptide p k that are contained within a helix motif annotation can be included in the allele-noninteracting variables w'.
• the encoding module 314 represents type of proteins or isoforms in the human proteome as an indicator vector o k that has a length equal to the number of proteins or isoforms in the human proteome, and the corresponding element o h j is 1 if peptide p k comes from protein and 0 otherwise.
• the encoding module 314 may also represent the overall set of variables z! for peptide p' and an associated MHC allele h as a row vector in which numerical representations of the allele-interacting variables x l and the allele-noninteracting variables w" are
• the encoding module 314 may represent Z h as a row vector equal to j.v; w l ] or [wi .v;
• the training module 316 constructs one or more presentation models that generate likelihoods of whether peptide sequences will be presented by MHC alleles associated with the peptide sequences. Specifically, given a peptide sequence p k and a set of MHC alleles a associated with the peptide sequence p k , each presentation model generates an estimate 3 ⁇ 4 indicating a likelihood that the peptide sequence p k will be presented by one or more of the associated MHC alleles a k .
• the training module 3 6 constructs the one more presentation models based on the training data sets stored in store 170 generated from the presentation information stored in 165.
• all of the presentation models capture the dependence between independent variables and dependent variables in the training data 170 such that a loss function is minimized.
• the loss function (y iS s Hies, ⁇ ) represents discrepancies between values of dependent variables y, e s for one or more data instances S in the training data 170 and the estimated likelihoods u, e s for the data instances S generated by the presentation model.
• the loss function is the mean squared loss given by equation lb as follows:
• the presentation model may be a parametric model in which one or more parameters ⁇ mathematically specify the dependence between the independent variables and dependent variables.
• various parameters of parametric-type presentation models that minimize the loss function are determined through gradient-based numerical optimization algorithms, such as batch gradient algorithms, stochastic gradient algorithms, and the like.
• the presentation model may be a non-parametric model in winch the model structure is determined from, the training data 170 and is not strictly based on a fixed set of parameters.
• the training module 316 may construct the presentation models to predict presentation likelihoods of peptides on a per-allele basis. In this case, the training module 316 may train the presentation models based on data instances S in the training data 170 generated from cells expressing single MHC alleles.
• the training module 316 models the estimated presentation likelihood Uk for peptide p k for a specific allele h by:
• the values for the set of parameters ft for each MHC allele h can be determined by minimizing the loss function with respect to ft, where i is each instance in the subset S of training data 170 generated from cells expressing die single MHC allele h.
• the output of the dependency function gh(xh ,'0h) represents a dependency score for the MHC allele h indicating whether the MHC allele h will present die corresponding neoantigen based on at least die allele interacting features ⁇ 3 ⁇ 4* and in particular, based on positions of amino acids of the peptide sequence of peptide p K .
• the dependency score for the MHC allele h may have a high value if the MHC allele h is likely to present the peptide / , and may have a low value if presentation is not likely.
• the transformation function ⁇ -) transforms the input, and more specifically, transforms the dependency score generated by g;, ⁇ x ;i3 ⁇ 4) in this case, to an appropriate value to indicate the likelihood that the peptide p k will be presented by an MHC allele.
• J(-) is a function having the range within [0, 1] for an appropriate domain range.
• ( ⁇ ) is the expit function given by:
• ( ⁇ ) can also be the hyperbolic tangent function given by:
• f(z) tanh(z) (5) when the values for the domain z is equal to or greater than 0.
• f(z) tanh(z) (5) when the values for the domain z is equal to or greater than 0.
• predictions are made for the mass spectrometry ion current that have values outside the range [0, 1 ],./( ⁇ ) can be any function such as the identity function, the exponential function, the log function, and the like.
• the per-allele likelihood that a peptide sequence p k will be presented by a MHC allele h can be generated by applying the dependency function g ( ⁇ ) for the MHC allele h to the encoded version of the peptide sequence p k to generate the corresponding dependency score .
• the dependency score may be transformed by the transformation function ⁇ ) to generate a per-allele like/ihood that the peptide sequence p k will be presented by the 2 allele h.
• the dependency function # / ,( ⁇ ) is an affine function given by:
• the dependency function is a network function given by:
• ⁇ 3 ⁇ 4( ⁇ ) having a series of nodes arranged in one or more layers.
• a node may be connected to other nodes through connections each having an associated parameter in the set of parameters ⁇ 3 ⁇ 4.
• a value at one particular node may be represented as a sum of the values of nodes connected to the particular node weighted by the associated parameter mapped by an activation function associated with the particular node.
• network models are advantageous because the presentation model can incorporate non-linearity and process data having different lengths of amino acid sequences. Specifically, through non-linear modeling, network models can capture interaction between amino acids at different positions in a peptide sequence and how this interaction affects peptide presentation.
• netw ork models ⁇ 3 ⁇ 4( ⁇ ) may be structured as feed-forward networks, such as artificial neural networks (ANN), convolutional neural networks (CNN), deep neural networks (DNN), and/or recurrent networks, such as long short-term memory networks (LSTM), bi-directional recurrent networks, deep bi-directional recurrent networks, and the like.
• ANN artificial neural networks
• CNN convolutional neural networks
• DNN deep neural networks
• recurrent networks such as long short-term memory networks (LSTM), bi-directional recurrent networks, deep bi-directional recurrent networks, and the like.
• each MHC allele in / ⁇ ..?. ... , m is associated with a separate network model
• Nh ⁇ -) denotes the output(s) from a network model associated with MHC allele h.
• FIG. 5 illustrates an example network model ⁇ 3 ( ⁇ ) in association with an arbitrary MHC allele /? 3.
• the network model ⁇ 3 ⁇ ) for MHC allele h - 3 includes three input nodes at layer 1---1, four nodes at layer h- ⁇ - ⁇ 2, two nodes at layer ⁇ -- 3, and one output node at layer / ⁇ /
• the network model NNj(-) is associated with a set of ten parameters 03( 1), 3 ⁇ 4(2), ... . ⁇ 3 ⁇ 4(10).
• the identified MHC alleles h --- l, 2 m are associated with a single network model ⁇ ' ⁇ 3 ⁇ 4( ⁇ ), and NN,(-) denotes one or more outputs of the single network model associated with MHC allele h.
• the set of parameters O h may correspond to a set of parameters for the single ne twork model, and thus, the set of parameters ⁇ 3 ⁇ 4 may be shared by all MHC alleles.
• the single network model ⁇ ⁇ ⁇ ⁇ ( ⁇ ) may be a network model that outputs a dependency score given the allele interacting variables ⁇ 3 ⁇ 4* and the encoded protein sequence i3 ⁇ 4 of an MHC allele h.
• the set of parameters ⁇ 3 ⁇ 4 may again correspond to a set of parameters for the single network model, and thus, the set of parameters 3 ⁇ 4 may be shared by all MHC alleles.
• ⁇ , ⁇ ) may denote the output of the single network model ⁇ ⁇ ( ⁇ ) given inputs jx i/ 3 ⁇ 4 ] to the single network model.
• Such a network model is advantageous because peptide presentation probabilities for MHC alleles that were unknown in tlie training data can be predicted just by identification of their protein sequence.
• FIG. 6B illustrates an example network model ⁇ ( ⁇ ) shared by MHC alleles.
• g ' / ,(.*3 ⁇ 4*;0' / ,) is the affine function with a set of parameters ⁇ , tlie network function, or the like, with a bias parameter ⁇ / , 0 in the set of parameters for allele interacting variables for the MHC allele that represents a baseline probability of presentation for the MHC allele h.
• the bias parameter ⁇ 3 ⁇ 4° may be shared according to the gene family of the MHC allele h. That is, the bias parameter 3 ⁇ 4° for MHC allele h may be equal to where gene(h) is the gene family of MHC allele h.
• MHC alleles HLA-A*02:01 , HLA-A* 02:02, and HLA-A*02:03 may be assigned to the gene family of "HLA-A,” and the bias parameter O h for each of these MHC alleles may be shared.
• the network model NNj(-) receives the ailele-interacting variables x for MHC allele h ---3 and generates the output NNj(j: ).
• the output is mapped by function ( ⁇ ) to generate the estimated presentation likelihood xik.
• the training module 316 incorporates allele-nomnteracting variables and models the estimated presentation likelihood iik for peptide p k by:
• w* denotes the encoded allele-noninteracting variables for peptide p k
• g w (-) is a function for the allele-noninteracting variables w k based on a set of parameters 0 W determined for the allele-noninteracting variables.
• the values for the set of parameters ⁇ ⁇ > for each MHC allele h and the set of parameters ⁇ ⁇ . for allele-noninteracting variables can be determined by minimizing the loss function with respect to i3 ⁇ 4 and 0 W , where / ' is each instance in the subset S of training data 170 generated from cells expressing single MHC alleles.
• the output of the dependency function g w (w k ;fJ w ) represents a dependency score for the allele noninteractmg variables indicating whether the peptide p k will be presented by one or more MHC alleles based on the impact of allele noninteracting variables.
• the dependency score for the allele noninteracting variables may have a high value if the peptide p k is associated with a C-terminal flanking sequence that is known to positively impact presentation of the peptide / , and may have a low value if the peptide p k is associated with a C-terminal flanking sequence that is known to negatively impact presentation of the peptide p k .
• the per-aliele likelihood that a peptide sequence p k will be presented by a MHC allele h can be generated by applying the function g h (-) for the MHC allele h to the encoded version of the peptide sequence p k to generate the corresponding dependency score for allele interacting variables.
• the function g w ⁇ -) for the allele noninteracting variables are also applied to the encoded version of the allele noninteracting variables to generate the dependency score for the allele noninteracting variables. Both scores are combined, and the combined score is transformed by the transformation function /( ⁇ ) to generate a per-allele likelihood that the peptide sequence p k will be presented by the MHC allele h.
• the training module 316 may include allele-noninteracting variables w k in the prediction by adding the allele-noninteracting variables w K to the allele-interacting variables in equation (2).
• the dependency function g w 0) for allele noninteracting variables may be an affme function or a network function in which a separate network model is associated with allele-noninteracting variables w k .
• the dependency function g w (-) is an affine function given by: that linearly combines the allele-noninteracting variables in w k with a corresponding parameter in the set of parameters 0 W .
• the dependency function g w (-) may also be a network function given by:
• g k (w k ; O w ) NN w (w k ; e w ).
• NN W (-) having an associated parameter in the set of parameters 0 W .
• g w k e w g ! w ⁇ w k 0 ! w ) + h(m k ; ⁇ ), (10)
• g is the affine function, the network function with the set of allele noninteracting parameters 0 v , or the like
• w is the mRMA quantification measurement for peptide p k
• h(-) is a flinction transforming the quantification measurement
• ⁇ - " is a parameter in the set of parameters for allele noninteracting variables that is combined with the mRNA quantification measurement to generate a dependency score for tlie mRMA quantification measurement.
• /?( ⁇ ) is the log function, however in practice /?( ⁇ ) may be any one of a variety of different functions.
• the dependency function the dependency function g n ⁇ -) for the allele-noninteracting variables can be given by:
• a parameter regulars zation term such as ⁇ ⁇ ⁇ ⁇ ⁇ ⁇ ⁇ ⁇ ⁇ , where j j ⁇ j j represents LI norm, L2 norm, a combination, or the like, can be added to the loss function when determining the value of the parameters.
• the optimal value of the hyperparameter ⁇ can be determined through appropriate methods.
• the likelihood that peptide p k will be presented by MHC allele h 3. among m---4 different identified MHC alleles using the affine transformation functions g h (; ⁇ , gw('), can be generated by:
• w K are the identified allele-noninteracting variables for peptide p k
• 0 W are the set of parameters determined for the allele-noninteracting variables.
• the network model NN;(-) receives the allele-interacting variables x for MHC allele /? 3 and generates the
• the network model NN W (-) receives the allele- noninteracting variables w k for peptide p k and generates the output NN w (w k ).
• the outputs are combined and mapped by function _/( ⁇ ) to generate the estimated presentation likelihood U' .
• the training module 316 may also construct the presentation models to predict presentation likelihoods of peptides in a multiple-allele setting where two or more MHC alleles are present. In this case, the training module 316 may train the presentation models based on data instances S in the training data 170 generated from, cells expressing single MHC alleles, cells expressing multiple MHC alleles, or a combination thereof.
• the training module 316 models the estimated presentation likelihood Uk for peptide p k in association with a set of multiple MHC alleles H as a function of the presentation likelihoods u ⁇ ' eM determined for each of the MHC alleles h in the set H determined based on cells expressing singie-alleles, as described above in conjunction with equations (2)-(l 1).
• the presentation likelihood iiu can be any function of Uk eH .
• the function is the maximum function, and the presentation likelihood iik can be determined as the maximum of the presentation likelihoods for each MHC allele h in the set H.
• the training module 316 models the estimated presentation likelihood ii for peptide p k by; u k - Pr(p fc presented) - (13)
• the values for the set of parameters for each MHC allele h can be determined by minimizing the loss function with respect to 3 ⁇ 4, where i is each instance in the subset S of training data 170 generated from cells expressing single MHC alleles and/or cells expressing multiple MHC alleles.
• the dependency function gi may be in the form of any of the dependency functions gu introduced above in sections ⁇ 111 B ! .
• the presentation likelihood that a peptide sequence p k will be presented by one or more MHC alleles h can be generated by applying the dependency function gh ⁇ -) to the encoded version of the peptide sequence p k for each of the MHC alleles Hto generate the corresponding score for the allele interacting variables.
• the scores for each MHC allele h are combined, and transformed by the transformation function /( ⁇ ) to generate the presentation likelihood that peptide sequence p k will be presented by the set of MHC alleles H.
• the presentation model of equation (13) is different from the per-allele model of equation (2), in that the number of associated alleles for each peptide p k can be greater than 1. In other words, more than one element in a h k can have values of 1 for the multiple MHC alleles H associated with peptide sequence p K .
• u k f(NN 2 0 2 ) + NN 3 (x k 3 ; ⁇ 3 )),
• the network model NN ? (-) receives the allele-interacting variables x for MHC allele /: ⁇ 2 and generates the output
• the network model N V3 ⁇ 4(-) receives the allele- interacting variables x for MHC allele h 3 and generates the output NA'j(x ).
• the outputs are combined and mapped by function /( ⁇ ) to generate the estimated presentation likelihood
• the training module 316 incorporates allele-noninteracting variables and models the estimated presentation likelihood tik for peptide p k by: u k - Pr(p k presented) - 4 ⁇ g h ⁇ x k k ;. 0 h ) V (14)
• the presentation likelihood that a peptide sequence p k will be presented by one or more MHC alleles H can be generated by applying the function g3 ⁇ 4(-) to the encoded version of the peptide sequence p k for each of the MHC alleles H to generate the corresponding dependency score for allele interacting variables for each MHC allele h.
• the function g w (-) for the allele noninteracting variables is also applied to the encoded version of the allele noninteracting variables to generate the dependency score for the allele noninteracting variables.
• the scores are combined, and the combined score is transformed by the transformation function /( ⁇ ) to generate the presentation likelihood that peptide sequence p k will be presented by the MHC alleles H.
• the number of associated alleles for each peptide p k can be greater than 1.
• more than one element in a, h can have values of 1 for the multiple MHC alleles / associated with peptide sequence / .
• w k are the identified allele-noninteracting variables for peptide p k
• 0 W are the set of parameters determined for the allele-noninteracting variables.
• u k f(NN w ⁇ w k ; 0 W ) + NN 2 (x k ; 0 2 ) + NN 3 (x k 3 ; 0 3 ))
• w k are the identified ailele-interacting variables for peptide p k
• 0 W are the set of parameters determined for allele-noninteracting variables.
• the network model ⁇ ⁇ ( ⁇ ) receives the allele-noninteracting variables w k for peptide p k and generates the output NN w (w*). The outputs are combined and mapped by function ⁇ ⁇ ) to generate the estimated presentation likelihood 3 ⁇ 43 ⁇ 4.
• the training module 316 may include allele-noninteracting variables w k in the prediction by adding the allele-noninteracting variables w k to the ailele-interacting variables ⁇ 3 ⁇ 4* in equation (15).
• the presentation likelihood can be given by: u k - Pr(p k presented) - ⁇ 9h ([x w k ]; e h )j . (15) VIII.C.4.
• Example 3.1 Models Using Implicit Per- Allele
• the training module 316 models the estimated presentation likelihood 3 ⁇ 4 for peptide p K by:
• s ⁇ -) may be the summation function or the second-order function, but it is appreciated that in other embodiments, .?( ⁇ ) can be any function such as the maximum function.
• Tire values for the set of parameters 0 for the implicit per-allele likelihoods can be determined by minimizing the loss function with respect to ⁇ , where is each instance in the subset S of training data 170 generated from, cells expressing single MHC alleles and/or cells expressing multiple MHC alleles,
• the presentation likelihood in the presentation model of equation (17) is modeled as a function of implicit per-allele presentation likelihoods u V that each correspond to the likelihood peptide p" will be presented by an individual MHC allele h.
• the implicit per-allele likelihood is distinct from the per-allele presentation likelihood of section VTII.B in that the parameters for implicit per-allele likelihoods can be learned from multiple allele settings, in which direct association between a presented peptide and the corresponding MHC allele is unknown, in addition to single-allele settings.
• the presentation model can estimate not only whether peptide p k will be presented by a set of MHC alleles H as a whole, but can also provide individual likelihoods u ' ⁇ ' ⁇ ' ⁇ that indicate which MHC allele h most likely presented peptide p k .
• An advantage of this is that the presentation model can generate the implicit likelihoods without training data for cells expressing single MHC alleles.
• r(-) is a function having the range [0, 1 ] .
• r(-) may be the clip function:
• s(-) is a summation function, and the presentation likelihood is given by summing the implicit per-allele presentation likelihoods:
• the implicit per-allele presentation likelihood for MHC allele h is generated by:
• the presentation likelihood that a peptide sequence p k will be presented by one or more MHC alleles H can be generated by applying the function gh(-) to the encoded version of the peptide sequence p k for each of the MHC alleles Hto generate the corresponding dependency score for allele interacting variables.
• Each dependency score is first transformed by the function ( ⁇ ) to generate implicit per-allele presentation likelihoods u
• the per-allele likelihoods u are combined, and the clipping function may be applied to the combined likelihoods to clip the values into a range [0, 1 ] to generate the presentation likelihood that peptide sequence p k will be presented by the set of MHC alleles H.
• the dependency function gu may be in the form of any of the dependency functions gu introduced above in sections VIII. B, 1.
• the likelihood that peptide p k will be presented by MHC alleles h 2. !i 3. among -l different identified MHC alleles using the affine transformation functions gk ⁇ '), can be generated by:
• the likelihood that peptide p k will be presented by MHC alleles h ---2, h ⁇ 3, among m ---4 different identified MHC alleles using the network transformation functions £ ⁇ 4( ⁇ , gw(:), can be generated by:
• Each output is mapped by function and combined to generate the estimated presentation likelihood 3 ⁇ 4,
• r(-) is the log function and /( ⁇ ) is the exponential function.
• the implicit per-ailele presentation likelihood for MHC allele h is generated by:
• uk ' h f (an ( x h ®k) + 9w ( ⁇ fc ; o w ) , (20 ) such that the presentation likelihood is generated by:
• the presentation likelihood that a peptide sequence p k will be presented by one or more MHC alleles H can be generated by applying the function gh ⁇ ) to the encoded version of the peptide sequence p k for each of the MHC alleles Hto generate the corresponding dependency score for allele interacting variables for each MHC allele h.
• the function g w (-) for the allele noninteracting variables is also applied to the encoded version of the allele noninteracting variables to generate the dependency score for the allele noninteracting variables.
• the score for the allele noninteracting variables are combined to each of the dependency scores for the allele interacting variables.
• Each of the combined scores are transformed by the function ( ⁇ ) to generate the implicit per-allele presentation likelihoods.
• the implicit likelihoods are combined, and the clipping function may be applied to the combined outputs to clip the values into a range [0,1] to generate the presentation likelihood that peptide sequence p k will be presented by the MHC alleles H.
• the dependency function g w may be in the form of any of the dependency functions # w introduced above in sections V1II.B.3.
• the likelihood that peptide p k will be presented by MHC alleles h 2. !i 3. among -I different identified MHC alleles using the affine transformation functions gh ⁇ '), g ), can be generated by:
• w k are the identified allele-noninteracting variables for peptide p k
• ⁇ are the set of parameters determ ined for the allele-noninteracting variables.
• the likelihood that peptide p k will be presented by MHC alleles h 2. 3. among -i different identified MHC alleles using the network transformation functions gh( ), g-A;), can be generated by:
• u k r (f (NN w (w k ; 0 abuse,) + NN 2 (x k 2 ; 0 2 )) + f ⁇ NN w (w k ; 0 W ) + NN 3 (x k ; 0 3 )))
• w* are the identified allele-interacting variables for peptide p k
• 0 W are the set of parameters determined for allele-noninteracting variables.
• FIG. 12 illustrates generating a presentation likelihood for peptide p k in association with MHC alleles h ----2. h -- 3 using example network models NN£), AWj(-), and NN W (- ⁇ .
• the network model NNw(-) receives the allele-noninteracting variables w k for peptide p k and generates the output NN w (w*).
• the outputs are combined and mapped by function /( ⁇ ).
• the implicit per-allele presentation likelihood for MHC allele h is generated by:
• .?( ⁇ ) is a second-order function
• the estimated presentation likelihood iik for peptide p k is given by:
• elements u c h are the implicit per-allele presentation likelihood for MHC allele h.
• the values for the set of parameters 0 for the implicit per-allele likelihoods can be determined by- minimizing the loss function with respect to ⁇ , where is each instance in the subset S of training data 170 generated from cells expressing single MHC alleles and/or cells expressing multiple MHC alleles.
• the implicit per-allele presentation likelihoods may be in any form shown in equations (18), (20), and (22) described above.
• the model of equation (23) may imply that there exists a possibility peptide p K will be presented by two MHC alleles simultaneously, in which the presentation by two HLA alleles is statistically independent.
• the presentation likelihood that a peptide sequence p k will be presented by one or more MHC alleles H can be generated by combining the implicit per-allele presentation likelihoods and subtracting the likelihood that each pair of MHC alleles will simultaneously present the peptide from the summation to generate the presentation likelihood that peptide sequence p k will be presented by the MHC alleles H.
• u k (** ⁇ ⁇ ,) + ( ⁇ e 3 ) - / ' ( ! ⁇ ⁇ 2 ) ⁇ / ⁇ (*! ⁇ ⁇ 3 ),
• the prediction module 320 receives sequence data and selects candidate neoantigens in the sequence data using the presentation models. Specifscally, the sequence data may be DNA sequences, RNA sequences, and/or protein sequences extracted from tumor tissue cells of patients. The prediction module 320 processes the sequence data into a plurality of peptide sequences p k having 8-15 amino acids.
• the prediction module 320 may process the given sequence "IEFROEIFJEF into three peptide sequences having 9 ammo acids "IEFROEIFJ,” “EFROEIFJE,” and "FROEIFJEF.” In one embodiment, the prediction module 320 may identify candidate neoantigens that are mutated peptide sequences by comparing sequence data extracted from normal tissue cells of a patient with the sequence data extracted from tumor tissue cells of the patient to identify portions containing one or more mutations.
• the presentation module 320 applies one or more of the presentation models to the processed peptide sequences to estimate presentation likelihoods of the peptide sequences.
• the prediction module 320 may select one or more candidate neoantigen peptide sequences that are likely to be presented on tumor HLA molecules by applying the presentation models to the candidate neoantigens.
• the presentation module 320 selects candidate neoantigen sequences that have estimated presentation likelihoods above a predetermined threshold.
• the presentation model selects the N candidate neoantigen sequences that have the highest estimated presentation likelihoods (where Nis generally the maximum number of epitopes that can be delivered in a vaccine).
• a vaccine including the selected candidate neoantigens for a given patient can be injected into the patient to induce immune responses.
• AUC area-under-curve
• FIG. 13A compares performance results of an example presentation model, as presented herein, and state-of-the-art models for predicting peptide presentation on multiple- allele mass spectrometiy data. Results showed that the example presentation model performed significantly better at predicting peptide presentation than state-of-the-art models based on affinity and stability predictions.
• the example presentation model shown in FIG. 13A as "MS” was the maximum of per-aileles presentation model shown in equation (12), using the affine dependency function g,3 ⁇ 4(-) and the expit function /( ⁇ ).
• the example presentation model was trained based on a subset of the single-allele HLA-A*02:01 mass spectrometiy data from the 1EDB data set (data set "Dl”) (data can be found at
• the example presentation model trained on mass spectrometry data had a significantly higher PPV value at 0% recall rate relative to the state-of-the-art models that predict peptide presentation based on MHC binding affinity predictions or MHC binding stability predictions.
• the example presentation model had approximately 14% higher PPV than the model based on affinity predictions, and had approximately 12% higher PPV than the model based on stability predictions.
• FIG. 13B compares performance results of another example presentation model, as presented herein, and state-of-the-art models for predicting peptide presentation on T-cell epitope data.
• T-cell epitope data contains peptide sequences that were presented by MHC alleles on the cell surface, and recognized by T-cells. Results showed that even though the example presentation model is trained based on mass spectrometry data, the example presentation model performed significantly better at predicting T-cell epitopes than state-of- the-art models based on affinity and stability predictions. In other words, the results of FIG .
• the example presentation model shown in FIG. 13B as "MS" was the per-allele presentation model shown in equation (2), using the affine transformation function £ / ,( ⁇ ) and the expit function /( ⁇ ) that was trained based on a subset of data set Dl . All peptides from source protein that contain presented peptides in the test set were eliminated from the training data such that the presentation model could not simply memorize the sequences of presented antigens.
• the per-allele presentation model trained on mass spectrometiy data had a significantly higher PPV value at 10% recall rate than the state- of-the-art models that predict peptide presentation based on MHC binding affinity or MHC binding stability predictions even though the presentation model was not trained based on protein sequences that contained presented peptides.
• the per-allele presentation model had approximately 9% higher PPV than the model based on affinity predictions, and had approximately 8% higher PPV than the model based on stability predictions.
• FIG. 13C compares performance results for an example function-of-sums model (equation (13)), an example sum-of-functions model (equation (1 )), and a example second order model (equation (23)) for predicting peptide presentation on muitiple-allele mass spectrometry data. Results showed that the sum-of-functions model and second order model performed better than the function-of-sums model. This is because the function-of-sums model implies that alleles in a muitiple-allele setting can interfere with each other for peptide presentation, when in reality, the presentation of peptides are effectively independent.
• the example presentation model labeled as "sigmoid-of-sums” in FIG. -13C was the function-of-sums model using a network dependency function gh(- ⁇ , the identity function /( ⁇ ), and the expit function r(-).
• the example model labeled as "sum-of- sigmoids" was the sum-of-functions model in equation ( 19) with a network dependency function g / ,(-), the expit function /( ⁇ ), and the identity function / ⁇ ( ⁇ ).
• the example model labeled as "hyperbolic tangent” was the sum-of-functions model in equation (19) with a network dependency function g3 ⁇ 4Q, the expit function ft-), and the hyperbolic tangent function r(-).
• the example model labeled as "second order " ' was the second order model in equation (23) using an implicit per-allele presentation likelihood form shown in equation (18) with a network dependency function g h ⁇ -) and the expit function /( ⁇ ).
• Each model was trained based on a subset of data set D l, D2, and D3.
• the example presentation models were applied to a test data that is a random subset of data set D3 that did not overlap with the training data.
• the first colum refers to the AUC of the ROC when each presentation model was applied to the test set
• the second column refers to the value of the negative log likelihood loss
• the third column refers to the PPV at 10% recall rate.
• the performance of presentation models "sum-of-sigmoids,” “hyperbolic tangent,” and “second order” were approximately tied at approximately 15- 16% PPV at 10% recall, while the performance of the model “sigmoid-of-sums" was slightly lower at approximately 1 1%.
• FIG. 13D compares performance results for two example presentation models that are trained with and without singie-allele mass spectrometry data on predicting peptide presentation for multiple-allele mass spectrometry data. The results indicated that example presentation models that are trained without singie-allele data achieve comparable performance to that of example presentation models trained with singie-allele data,
• the example model "with A2 B7 singie-allele data” was the "sum-of-sigmoids" presentation model in equation (19) with a network dependency function g ('), the expit function ( ⁇ ), and the identity function r(-).
• the model was trained based on a subset of data set D3 and singie-allele mass spectrometr ⁇ ' data for a variety of MHC alleles from the IEDB database (data can be found at: http://www.iedb.org/doc/mhc_ligand_full.zip).
• the example model "without A2/B7 singie-allele data" was the same model, but trained based on a subset of the multiple-allele D3 data set without singie-allele mass spectrometry data for alleles HLA-A*02:01 and HLA-B*07:02, but with singie-allele mass spectrometr ⁇ ' data for other alleles.
• cell line HCC1937 expressed HLA-B*07:02 but not HLA-A* 02:01
• cell line HCT116 expressed HLA-A*02:01 but not HLA- B*07:02.
• the example presentation models were applied to a test data that was a random subset of data set D3 and did not overlap with the training data.
• the column “Correlation” refers to the correlation between the actual labels that indicate whether the peptide was presented on the corresponding allele in the test data, and the label for prediction.
• the predictions based on the implicit per- allele presentation likelihoods for MHC allele HLA-A*02:01 performed significantly better on singie-allele test data for MHC allele HLA-A*02:01 rather than for MHC allele HLA- B*07:02, Similar results are shown for MHC allele HLA-B*07:02.
• FIG. I3E shows performance for the "without A2/B7 singie-allele data" and "with A2/B7 singie-allele data" example models shown in FIG. 13D on singie-allele mass spectrometry data for alleles HLA-A*02:01 and HLA-B*07:02 that were held out in the analysis shown in FIG. 13D.
• Results indicate that even through the example presentation model is trained without single-allele mass spectrometry data for these two alleles, the model is able to learn binding motifs for each MHC allele,
• A2 model predicting B7 indicates the performance of the model when peptide presentation is predicted for single-allele HLA-B*07:02 data based on the implicit per-allele presentation likelihood estimate for MHC allele HLA-A*02:01.
• A2 model predicting A2 indicates the performance of the model when peptide presentation is predicted for single-allele HLA-A*02:01 based on the implicit per-allele presentation likelihood estimate for MHC allele HLA-A*02:01.
• B7 model predicting B7 indicates the performance of the model when peptide presentation is predicted for single- allele HLA-B*07:02 data based on the implicit per-allele presentation likelihood estimate for MHC allele HLA-B* 07:02.
• B7 model predicting A2 indicates the performance of the model when peptide presentation is predicted for single-allele HLA ⁇ A*02:01 based on the implicit per-allele presentation likelihood estimate for MHC allele HLA-B*07:02.
• FIG. I3F shows the common anchor residues at positions 2 and 9 among nonamers predicted by the "without A2/B7 single-allele data" example model shown in FIG. 13D.
• the peptides were predicted to be presented if the estimated likelihood was above 5%.
• Results show that most common anchor residues in the peptides identified for presentation on the MHC alleles HLA-A*02:01 and HLA-B* 07: 02 matched previously known anchor motifs for these MHC alleles. This indicates that the example presentation models correctly learned peptide binding based on particular positions of amino acids of the peptide sequences, as expected.
• amino acids L/M at position 2 and amino acids V/L at position 9 were known to be canonical anchor residue motifs (as shown in Table 4 of https://link.springer.com/article/10.1 186/1745-7580-4-2) for HLA-A* 02:01, and amino acid P at position 2 and amino acids L/V at position 9 were known to be canonical anchor residue motifs for HLA-B*07:02.
• the most common anchor residue motifs at positions 2 and 9 for peptides identified the model matched the known canonical anchor residue motifs for both HLA alleles.
• FIG. 13G compares performance results between an example presentation model that incorporated C- and -terminal flanking sequences as aliele-interacting variables, and an example presentation model that incorporated C- and N-terminal flanking sequences as allele-noninteracting variables. Results showed that incorporating C- and N-terminal flanking sequences as allele noninteracting variables significantly improved model performance. More specifically, it is valuable to identify appropriate features for peptide presentation that are common across different MHC alleles, and model them such that statistical strength for these allele-noninteracting variables are shared across MHC alleles to improve presentation model performance.
• the example '"aliele-interacting” model was the sum-of-functions model using the form of implicit per-allele presentation likelihoods in equation (22) that incorporated C- and N-terminal flanking sequences as aliele-interacting variables, with a network dependency function » / ,( ⁇ ) and the expit function /( ⁇ ).
• the example "allele-noninteracting" model was the sum-of-functions model shown in equation (21) that incorporated C ⁇ and N-terminal flanking sequences as allele-noninteracting variables, with a network dependency function g3 ⁇ 4(-) and the expit function /( ⁇ ).
• the allele-noninteracting variables were modeled through a separate network dependency function g w (-).
• Both models were trained on a subset of data set D3 and single-allele mass spectrometry data for a variety of MHC alleles from the IEDB database (data can be found at: http://www.iedb.org/doc/mhc_ligand_full.zip).
• Each of the presentation models was applied to a test data set that is a random subset of data set D3 that did not overlap with the training data.
• FIG. 13H illustrates the dependency between fraction of presented peptides for genes based on mRNA quantification for mass spectrometry data on tumor cells. Results show that there is a strong dependency between mRNA expression and peptide presentation.
• the horizontal axis in FIG.13G indicates mRNA expression in terms of transcripts per million (TPM) quartiles.
• the vertical axis in FIG. 13G indicates fraction of presented epitopes from genes in corresponding mRNA expression quartiles.
• Each solid line is a plot relating the two measurements from a tumor sample that is associated with corresponding mass spectrometry data and mRNA expression measurements.
• FIG. 13G there is a strong positive correlation between mRNA expression, and the fraction of peptides in the corresponding gene.
• peptides from genes in the top quartile of RNA expression are more than 20 times likely to be presented than the bottom quartile.
• FIG. 131 shows performance of two example presentation models, one of which is trained based on mass spectrometry tumor cell data, another of which incorporates mRNA quantification data and mass spectrometry tumor cell data.
• MHC flurry + RNA filter was a model similar to the current state-of-the-art model that predicts peptide presentation based on affinity predictions. It was implemented using MHCflurry along with a standard gene expression filter that removed all peptides from proteins with mRNA quantification measurements that were less than 3.2 FPKM.
• Example Model, no RNA model was the "sum-of-sigmoids" example presentation model shown in equation (21) with the network dependency function ⁇ 3 ⁇ 4( ⁇ ), the network dependency function gv(-), and the expit function ( ⁇ ).
• the '"Example Model, no RNA” model incorporated C-terminal flanking sequences as allele-noninteracting variables through a network dependency function g w (- ⁇ ,
• the ''Example Model, with RNA” model was the "sum-of-sigmoids" presentation model shown in equation (19) with network dependency function gh( ), the network dependency function g w (-) in equation (10) incorporating mRNA quantification data through a log function, and the expit function ( ⁇ ).
• the "Example Model, with RNA” model incorporated C-terminal flanking sequences as allele-noninteracting variables through the network dependency functions g w (-) and incorporated mRNA quantification measurements through the log function.
• Each model was trained on a combination of the single-allele mass spectrometry data from the 1EDB data set, 7 ceil lines from the multiple-alleie mass spectrometry data from the Bassani-Sternberg data set, and 20 mass spectrometry tumor samples. Each model was applied to a test set including 5,000 held-out proteins from 7 tumor samples that constituted 9,830 presented peptides from a total of 52, 156,840 peptides.
• the "Example Model, no RNA” model has a PPV value at 20% Recall of 21%, while that of the state-of-the-art model is approximately 3%, This indicates an initial performance improvement of 18% in PPV value, even without the incorporation of mRNA quantification measurements.
• the "Example Model, with RNA " model that incorporates mRN A quantification data into the presentation model shows a PPV value of approximately 30%, which is almost a 10% increase in performance compared to the example presentation model without mRNA quantification measurements.
• FIG. 131 compares probability of peptide presentation for different peptide lengths between results generated by the "Example Model, with RNA” presentation model described in reference to FIG. 131, and predicted results by state-of-the-art models that do not account for peptide length when predicting peptide presentation. Results indicated that the "Example Model, with RNA” example presentation model from FIG. 131 captured variation in likelihoods across peptides of differing lengths.
• Tl e vertical axis denoted the probability of peptide presentation conditioned on the lengths of the peptide.
• the plot "Actual Test Data Probability" showed the proportion of presented peptides according to the length of the peptide in a sample test data set.
• the presentation likelihood varied with the length of the peptide. For example, as shown in FIG. 13 , a lOmer peptide with canonical HLA-A2 L/V anchor motifs was approximately 3 times less likely to be presented than a 9mer with the same anchor residues.
• Models Ignoring Length' 1 indicated predicted measurements if state-of-the-art models that ignore peptide length were to be applied to the same test data set for presentation prediction. These models may be NetMHC versions before version 4.0, NetMHCpan versions before version 3.0, and
• relu(-) is the rectified linear unit (RELU) function
• W h ! b h , W/, 2 , and b h are the set of parameters ⁇ determined for tlie model .
• the allele interacting variables x consist of peptide sequences.
• the dimensions of W h are (231 x 256), tlie dimensions of b h (1 x 256), the dimensions of W)i are (256 x 1), and b h is a scalar.
• values for b h , b h W h , and W h 2 are listed below.
• FIG. 14 illustrates an example computer 1400 for implementing the entities shown in FIGS. 1 and 3.
• the computer 1400 includes at least one processor 1402 coupled to a chipset 1404.
• the chipset 1404 includes a memory controller hub 1420 and an input/output (I/O) controller hub 1422.
• a memory 1406 and a graphics adapter 14 2 are coupled to the memory controller hub 1420, and a display 1418 is coupled to the graphics adapter 1412.
• a storage device 1408, an input device 1414, and network adapter 1416 are coupled to the I/O controller hub 1422.
• Other embodiments of the computer 1400 have different architectures.
• the storage device 1408 is a non-transitory computer-readable storage medium such as a hard drive, compact disk read-only memory (CD-ROM), DVD, or a solid-state memory device.
• the memory 1406 holds instructions and data used by the processor 1402.
• the input interface 1414 is a touch-screen interface, a mouse, track ball, or other type of pointing device, a keyboard, or some combination thereof, and is used to input data into the computer 1400.
• the computer 1400 may be configured to receive input (e.g., commands) from the input interface 1414 via gestures from the user.
• the graphics adapter 1412 displays images and other information on the display 1418.
• the network adapter 1416 couples the computer 1400 to one or more computer networks.
• the computer 1400 is adapted to execute computer program modules for providing functionality described herein.
• module refers to computer program logic used to provide the specified functionality.
• a module can be implemented in hardware, firmware, and/or software.
• program modules are stored on the storage device 1408, loaded into the memory 1406, and executed by the processor 1402,
• the types of computers 1400 used by the entities of FIG . 1 can vary depending upon the embodiment and the processing power required by the entity.
• the presentation identification system 160 can run in a single computer 1400 or multiple computers 1400 communicating with each other through a network such as in a server farm.
• the computers 1400 can lack some of the components described above, such as graphics adapters 1412, and displays 1418.
• Bodini, M. et al. The hidden genomic landscape of acute myeloid leukemia: subclonal structure revealed by undetected mutations. Biood 125, 600-605 (2015).
• RNA CoMPASS a dual approach for pathogen and host transcriptome analysis of RNA-seq datasets. PloS One 9, e89445 (2014).
• HLA -DR monoclonal antibodies inhibit the proliferation of normal and chronic granulocytic leukaemia myeloid progenitor cells. Br J Haematol. 1982 Nov;52(3):411 -20.

Landscapes

• Health & Medical Sciences (AREA)
• Life Sciences & Earth Sciences (AREA)
• Engineering & Computer Science (AREA)
• Chemical & Material Sciences (AREA)
• Physics & Mathematics (AREA)
• General Health & Medical Sciences (AREA)
• Bioinformatics & Cheminformatics (AREA)
• Immunology (AREA)
• Molecular Biology (AREA)
• Public Health (AREA)
• Medicinal Chemistry (AREA)
• Biotechnology (AREA)
• Medical Informatics (AREA)
• Proteomics, Peptides & Aminoacids (AREA)
• Biophysics (AREA)
• Spectroscopy & Molecular Physics (AREA)
• Bioinformatics & Computational Biology (AREA)
• Analytical Chemistry (AREA)
• Organic Chemistry (AREA)
• Epidemiology (AREA)
• Veterinary Medicine (AREA)
• Animal Behavior & Ethology (AREA)
• Pharmacology & Pharmacy (AREA)
• Theoretical Computer Science (AREA)
• Microbiology (AREA)
• Evolutionary Biology (AREA)
• Hematology (AREA)
• Oncology (AREA)
• Genetics & Genomics (AREA)
• Chemical Kinetics & Catalysis (AREA)
• General Chemical & Material Sciences (AREA)
• Biomedical Technology (AREA)
• Biochemistry (AREA)
• Pathology (AREA)
• Urology & Nephrology (AREA)
• Zoology (AREA)
• Nuclear Medicine, Radiotherapy & Molecular Imaging (AREA)
• Mycology (AREA)
• Wood Science & Technology (AREA)
• Bioethics (AREA)

Priority Applications (23)

Applications Claiming Priority (10)

Publications (1)

Family

ID=59057557

Family Applications (1)

Country Status (18)

Cited By (49)

Families Citing this family (74)

Citations (2)

Family Cites Families (236)

• 2016
  • 2016-12-16 RU RU2018124997A patent/RU2729116C2/ru active
  • 2016-12-16 KR KR1020187020164A patent/KR102883915B1/ko active Active
  • 2016-12-16 CN CN201680080924.4A patent/CN108601731A/zh active Pending
  • 2016-12-16 PE PE2018001119A patent/PE20181535A1/es unknown
  • 2016-12-16 IL IL305238A patent/IL305238B2/en unknown
  • 2016-12-16 US US15/381,375 patent/US20170199961A1/en not_active Abandoned
  • 2016-12-16 KR KR1020257036981A patent/KR20250164855A/ko active Pending
  • 2016-12-16 JP JP2018550664A patent/JP7114477B2/ja active Active
  • 2016-12-16 AU AU2016369519A patent/AU2016369519B2/en active Active
  • 2016-12-16 ES ES16876766T patent/ES2970865T3/es active Active
  • 2016-12-16 MX MX2018007204A patent/MX2018007204A/es unknown
  • 2016-12-16 WO PCT/US2016/067159 patent/WO2017106638A1/en not_active Ceased
  • 2016-12-16 CA CA3008641A patent/CA3008641A1/en active Pending
  • 2016-12-16 EP EP23207311.4A patent/EP4299136A3/en active Pending
  • 2016-12-16 BR BR112018012374-9A patent/BR112018012374A2/pt not_active Application Discontinuation
  • 2016-12-16 TW TW105141968A patent/TWI765875B/zh active
  • 2016-12-16 IL IL259931A patent/IL259931B2/en unknown
  • 2016-12-16 EP EP16876766.3A patent/EP3389630B1/en active Active
  • 2016-12-16 SG SG11201804957VA patent/SG11201804957VA/en unknown
  • 2016-12-16 CN CN202510092422.4A patent/CN119905147A/zh active Pending
  • 2016-12-16 IL IL322399A patent/IL322399A/en unknown
• 2017
  • 2017-03-22 US US15/466,729 patent/US10055540B2/en active Active
• 2018
  • 2018-06-06 US US16/001,569 patent/US10847252B2/en active Active
  • 2018-06-13 MX MX2023014124A patent/MX2023014124A/es unknown
  • 2018-06-13 PH PH12018501267A patent/PH12018501267A1/en unknown
  • 2018-07-13 CO CONC2018/0007417A patent/CO2018007417A2/es unknown
  • 2018-07-19 US US16/040,409 patent/US10847253B2/en active Active
  • 2018-09-11 US US16/128,421 patent/US20190034585A1/en not_active Abandoned
• 2020
  • 2020-11-23 US US17/101,518 patent/US11183286B2/en active Active
  • 2020-11-23 US US17/101,522 patent/US20210166784A1/en active Pending
• 2022
  • 2022-06-01 JP JP2022089465A patent/JP2022133271A/ja not_active Withdrawn
• 2023
  • 2023-07-12 AU AU2023204618A patent/AU2023204618B2/en active Active
• 2025
  • 2025-01-31 JP JP2025014472A patent/JP2025065197A/ja active Pending