Methods and systems for predicting HLA class II specific epitopes and characterizing CD4 + T cells

Abstract

Methods for preparing personalized cancer vaccines and methods for training machine learning HLA-peptide presentation predictive models. Further, there is disclosed herein a method of preparing an epitope-containing HLA class II tetramer or multimer, the method comprising contacting a purified soluble HLA-DM loaded with a peptide epitope with an HLA class II tetramer or multimer, thereby forming an HLA class II tetramer or multimer loaded with a peptide epitope.

Description

[0001] Cross-references

[0002] This application claims the benefit of U.S. Provisional Application No. 63 / 343,913, filed May 19, 2022, which is incorporated herein by reference in its entirety. Background Art

[0003] The major histocompatibility complex (MHC) is a gene complex encoding human leukocyte antigen (HLA) genes. HLA genes are expressed as protein heterodimers displayed on the surface of human cells to circulating T cells. HLA genes are highly polymorphic, allowing them to fine-tune the adaptive immune system. The adaptive immune response depends in part on the ability of T cells to identify and eliminate cells that display disease-associated peptide antigens that bind to human leukocyte antigen (HLA) heterodimers.

[0004] In humans, endogenous and exogenous proteins can be processed into peptides by proteasomes and by cytoplasmic and endosomal / lysosomal proteases and peptidases and presented by two classes of cell surface proteins encoded by MHC genes. These cell surface proteins are called human leukocyte antigens (HLA class I and class II), and the set of peptides that bind to them and trigger an immune response is called HLA epitopes. HLA epitopes are key components that enable the immune system to detect danger signals such as pathogen infection and self-transformation. CD4+ T cells recognize class II MHC (HLA-DR, HLA-DQ, and HLA-DP) epitopes displayed on antigen presenting cells (APCs) such as dendritic cells and macrophages. Endogenous processing and presentation of HLA class II ligands is a complex process involving a variety of molecular chaperones and subsets of enzymes, all of which have not yet been fully characterized. HLA class II peptide presentation activates helper T cells, which subsequently promote B cell differentiation and antibody production as well as CTL responses. Activated helper T cells also secrete cytokines and chemokines that activate and induce differentiation of other T cells.

[0005] Understanding the peptide binding preferences of each HLA class II heterodimer is key to successfully predicting which cancer or tumor-specific antigens are likely to elicit cancer or tumor-specific T cell responses. There is a need for methods of identifying and isolating specific HLA class II associated peptides (e.g., neoantigenic peptides). Such methods and isolated molecules can be used, for example, to develop therapeutic agents, including but not limited to immune-based therapeutic agents. Summary of the invention

[0006] The methods and compositions described herein can be used in a wide range of applications. For example, the methods and compositions described herein can be used to identify immunogenic antigenic peptides, and can be used to develop drugs, such as personalized medicine, and the isolation and characterization of antigen-specific T cells.

[0007] CD4+ T cell responses may have anti-tumor activity. High CD4+ T cell response rates can be shown without using class II predictions (e.g., 60% of SLP epitopes in NeoVax studies (49% in NT-001, see Ott et al., Nature, July 13, 2017; 547(7662): 217-221), and 48% of mRNA epitopes in Biontech studies, see Sahin et al., Nature, July 13, 2017; 547(7662): 222-226). It may not be clear whether these epitopes are usually naturally presented (by tumors or phagocytic DCs). It may be desirable to convert high CD4+ T response rates into therapeutic effects by improving the identification of truly presented HLA class II binding epitopes.

[0008] The role of gene expression, enzyme cleavage, and pathway / localization preferences may not yet be robustly quantified. It may not be clear whether autophagy (HLA class II presentation of tumor cells) or phagocytosis (HLA class II presentation of tumor epitopes by APCs) is the more relevant pathway, although most existing MS data may be presumed to be derived from autophagy. NetMHCIIpan may be the current predictive standard, but it may not be considered accurate. Of the three HLA class II loci (DR, DP, and DQ), data may exist only for certain common alleles of HLA-DR.

[0009] There may be different data generation methods to learn the rules of HLA class II presentation, including field standards and the proposed method. Field standards may include affinity measurements, which may be the basis of the NetMHCIIpan predictor, provide low throughput and require radioactive reagents, and it omits the role of processing. The proposed method may include mass spectrometry, where data from cell lines / tissues / tumors can help determine the processing rules of autophagy, while monoallelic MS may allow for the determination of allele-specific binding rules (assuming that multi-allelic MS data are too complex for effective learning (Bassani-Sternberg. MCP. 2018)).

[0010] There may be different approaches to validate new HLA class II predictors: validation on retained MS data, which may be the default setting; retrospective of vaccine studies (e.g. NT-001), where immune monitoring data may assess vaccine peptide loading on APCs rather than tumor presentation, and the data may be sparsely extended across many different alleles; biochemical affinity measurements, which may be configured to obtain measurements for peptides that are inconsistently predicted (for only 2-3 alleles); T cell induction, which may be configured to test the rate at which Neon-preferred and NetMHCIIpan-preferred epitopes induce ex vivo T cell responses.

[0011] For validation by T cell induction, the default approach may include evaluation of predicted inconsistent neoORFs from TCGA, where the induction material may include healthy donor APCs and T cells, and induction and readout may be by SLP (~15-mer peptides). Random peptides may produce high response rates, while SLP may not adequately address processing issues. Possible solutions may include induction by mRNA.

[0012] Methods disclosed herein may include generating LC-MS / MS mono-allelic data for training allele-specific machine learning methods for epitope prediction. Such methods may include improving LC-MS / MS data quality using a set of quality metrics to rigorously remove false positives that increase prediction model performance; identifying allele-specific HLA class II binding cores from HLA-ligand panel LC-MS / MS data sets; improving HLA class II ligand and epitope prediction using machine learning algorithms; and / or identifying biological variables that affect HLA class II-ligand presentation and improve HLA class II epitope prediction, such as gene expression, cleavage, gene preference, cellular localization, and secondary structure.

[0013] Provided herein is a method comprising: (a) processing amino acid information of a plurality of candidate peptide sequences using a machine learning HLA peptide presentation prediction model to generate a plurality of presentation predictions, wherein each candidate peptide sequence in the plurality of candidate peptide sequences is encoded by a genome or exome of a subject, wherein the plurality of presentation predictions comprises an HLA presentation prediction for each of the plurality of candidate peptide sequences, wherein each HLA presentation prediction indicates a likelihood that one or more proteins encoded by a class II HLA allele of a cell of the subject is capable of presenting a given candidate peptide sequence in the plurality of candidate peptide sequences, wherein the machine learning HLA peptide presentation prediction model is trained using training data comprising sequence information of training peptide sequences, the training peptides being identified by mass spectrometry as being presented by an HLA protein expressed in a training cell; and (b) identifying a peptide sequence in the plurality of peptide sequences as presented by at least one of the one or more proteins encoded by a class II HLA allele of a cell of the subject based at least on the plurality of presentation predictions; wherein the machine learning HLA peptide presentation prediction model has a PPV of at least 0.07 according to a presentation positive predictive value (PPV) determination method.

[0014] Provided herein is a method comprising: (a) processing amino acid information of a plurality of peptide sequences encoded by a genome or exome of a subject using a machine learning HLA peptide binding prediction model to generate a plurality of binding predictions, wherein the plurality of binding predictions comprises an HLA binding prediction for each of the plurality of candidate peptide sequences, each binding prediction indicating a likelihood that one or more proteins encoded by a class II HLA allele of a cell of the subject binds to a given candidate peptide sequence in the plurality of candidate peptide sequences, wherein the machine learning HLA peptide binding prediction model is trained using training data comprising sequence information of peptide sequences identified to bind to an HLA class II protein or an HLA class II protein analog; and (b) identifying, based at least on the plurality of binding predictions, a peptide sequence in the plurality of peptide sequences having a probability of binding to at least one of the one or more proteins encoded by a class II HLA allele of a cell of the subject that is greater than a threshold binding prediction probability value; wherein the machine learning HLA peptide binding prediction model has a PPV of at least 0.1 according to a binding positive predictive value (PPV) determination method.

[0015] In some embodiments, the machine learning HLA peptide presentation prediction model is trained using training data comprising sequence information of training peptide sequences that are identified by mass spectrometry as being presented by HLA proteins expressed in training cells.

[0016] In some embodiments, the method includes ranking at least two peptides presented by at least one of the one or more proteins identified as encoded by a class II HLA allele of cells of the subject based on the presentation prediction.

[0017] In some embodiments, the method includes selecting one or more peptides from the two or more ranked peptides.

[0018] In some embodiments, the method includes selecting one or more peptides from the plurality of peptides that are identified as being presented by at least one of the one or more proteins encoded by a class II HLA allele in cells of the subject.

[0019] In some embodiments, the method includes selecting one or more peptides from two or more peptides ranked based on the presentation prediction.

[0020] In some embodiments, the machine learning HLA peptide presentation prediction model has a positive predictive value (PPV) of at least 0.07 when processing amino acid information of a plurality of test peptide sequences to generate a plurality of test presentation predictions, each test presentation prediction indicating a likelihood that one or more proteins encoded by a class II HLA allele of a cell of the subject is capable of presenting a given test peptide sequence in the plurality of test peptide sequences, wherein the plurality of test peptide sequences comprises at least 500 test peptide sequences, the sequences comprising (i) at least one hit peptide sequence identified by mass spectrometry as to be presented by an HLA protein expressed in a cell, and (ii) at least 499 decoy peptide sequences contained within a protein encoded by a genome of an organism, wherein the organism and the subject are of the same species, wherein the ratio of the at least one hit peptide sequence in the plurality of test peptide sequences to the at least 499 decoy peptide sequences is 1:499, and according to the machine learning HLA peptide presentation prediction model, a top percentage of the plurality of test peptide sequences are predicted to be presented by an HLA protein expressed in a cell.

[0021] In some embodiments, the machine learning HLA peptide presentation prediction model has a positive predictive value (PPV) of at least 0.1 when processing amino acid information of a plurality of test peptide sequences to generate a plurality of test binding predictions, each test binding prediction indicating a likelihood that one or more proteins encoded by a class II HLA allele of a cell of the subject will bind to a given test peptide sequence in the plurality of test peptide sequences, wherein the plurality of test peptide sequences comprises at least 20 test peptide sequences comprising (i) at least one test peptide sequence identified by mass spectrometry as being presented by an HLA protein expressed in a cell; The method comprises the following steps of: (i) detecting a hit peptide sequence presented by a human leukemia cell, and (ii) at least 19 decoy peptide sequences contained in a protein comprising at least one peptide sequence identified by mass spectrometry as being presented by an HLA protein expressed in a cell, such as a single HLA protein expressed in a cell (e.g., a monoallelic cell), wherein the ratio of the at least one hit peptide sequence in the plurality of test peptide sequences to the at least 19 decoy peptide sequences is 1:19, and according to the machine learning HLA peptide presentation prediction model, a top percentage of the plurality of test peptide sequences are predicted to bind to an HLA protein expressed in a cell.

[0022] In some embodiments, there is no amino acid sequence overlap between the at least one hit peptide sequence and the bait peptide sequence.

[0023] In some embodiments, the machine learning HLA peptide presentation prediction model has a positive predictive value (PPV) of at least 0.08, 0.09, 0.1, 0.11, 0.12, 0.13, 0.14, 0.15, 0.16, 0.17, 0.18, 0.19, 0.2, 0.21, 0.22, 0.23, 0.24, 0.25, 0.26, 0.27, 0.28, 0.29, 0.3, 0.31, 0.32, 0.33, 0.34, 0.35, 0.36, 0.37, 0.38, 0.39, 0.4, 0.41, 0.42, 0.43, 0.44, 0.45, 0.46, 0.47, 0.48, 0.49, 0.5、0.51、0.52、0.53、0.54、0.55、0.56、0.57、0.58、0.59、0.6、0.61、0.62、0.63、0.64、0.65、0.66、0.67、0.68、0.69、0.7、0.71、0.72、0.73、0.74、 0.75, 0.76, 0.77, 0.78, 0.79, 0.8, 0.81, 0.82, 0.83, 0.84, 0.85, 0.86, 0.87, 0.88, 0.89, 0.9, 0.91, 0.92, 0.93, 0.94, 0.95, 0.96, 0.97, 0.98 or 0.99.

[0024] In some embodiments, the at least one hit peptide sequence comprises at least 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, 94, 95, 96, 97, 98, 99, 100, 101, 102, 103, 104, 105, 106, 107, 108, 109, 110, 111, 112, 113, 114, 115, 116, 117, 118, 119, 120, 121, 122, 123, 124, 125, 126, 127, 128, 129, 130, 131, 132, 133, 134, 135, 136, 137, 138, 139, 200, 201, 202, 203, 204, 205, 206, 207, 208, 209,

[0025] In some embodiments, the at least 499 decoy peptide sequences comprise at least 500, 600, 700, 800, 900, 1000, 1100, 1200, 1300, 1400, 1500, 1600, 1700, 1800, 1900, 2000, 2100, 2200, 2300, 2400, 2500, 2600, 2700, 2800, 2900, 3000, 3100, 3200, 3300, 3400, 3500, 3600, 3700, 3800, 3900, 4000, 4100, 4200, 4300, 4400, 4500, 4600, 4700, 4800, 4900, 500 0, 5100, 5200, 5300, 5400, 5500, 5600, 5700, 5800, 5900, 6000, 6100, 6200, 6300, 6400, 6500, 6600, 6700, 6800, 6900, 7000, 7100, 7200, 7300, 7400, 75 00, 7600, 7700, 7800, 7900, 8000, 8100, 8200, 8300, 8400, 8500, 8600, 8700, 8800, 8900, 9000, 9100, 9200, 9300, 9400, 9500, 9600, 9700, 9800, 9900, 10 000、11000、12000、13000、14000、15000、16000、17000、18000、19000、20000、21000、22000、23000、24000、25000、26000、27000、28000、29000、30000、 31000、32000、33000、34000、35000、36000、37000、38000、39000、40000、41000、42000、43000、44000、45000、46000、47000、48000、49000、50000、52500 , 55000, 57500, 60000, 62500, 65000, 67500, 70000, 72500, 75000, 77500, 80000, 82500, 85000, 87500, 90000, 92500, 95000, 97500, 100000, 125000, 1 50000、175000、200000、225000、250000、275000、300000、325000、350000、375000、400000、425000、450000、475000、500000、600000、700000、800000、900,000 or 1,000,000 decoy peptide sequences. One skilled in the art will recognize that changing the hit: decoy ratio changes the PPV.

[0026] In some embodiments, the at least 500 test peptide sequences comprise at least 600, 700, 800, 900, 1000, 1100, 1200, 1300, 1400, 1500, 1600, 1700, 1800, 1900, 2000, 2100, 2200, 2300, 2400, 2500, 2600, 2700, 2800, 2900, 3000, 3100, 3200, 3300, 3400, 3500, 3600, 3700, 3800, 3900, 4000, 4100, 4200, 4300, 4400, 4500, 4600, 4700, 4800, 4900, 5000, 5100, 5200, 5300 100、5200、5300、5400、5500、5600、5700、5800、5900、6000、6100、6200、6300、6400、6500、6600、6700、6800、6900、7000、7100、7200、7300、7400、7500、 7600、7700、7800、7900、8000、8100、8200、8300、8400、8500、8600、8700、8800、8900、9000、9100、9200、9300、9400、9500、9600、9700、9800、9900、1000 0, 11000, 12000, 13000, 14000, 15000, 16000, 17000, 18000, 19000, 20000, 21000, 22000, 23000, 24000, 25000, 26000, 27000, 28000, 29000, 30000, 31 000、32000、33000、34000、35000、36000、37000、38000、39000、40000、41000、42000、43000、44000、45000、46000、47000、48000、49000、50000、52500、 55000、57500、60000、62500、65000、67500、70000、72500、75000、77500、80000、82500、85000、87500、90000、92500、95000、97500、100000、125000、15 0000、175000、200000、225000、250000、275000、300000、325000、350000、375000、400000、425000、450000、475000、500000、600000、700000、800000、900,000 or 1,000,000 test peptide sequences.

[0027] In some embodiments, the top percentage is the top 0.20%, 0.30%, 0.40%, 0.50%, 0.60%, 0.70%, 0.80%, 0.90%, 1.00%, 1.10%, 1.20%, 1.30%, 1.40%, 1.50%, 1.60%, 1.70%, 1.80%, 1.90%, 2.00%, 2.10%, 2.20%, 2.30%, 2.40%, 2.50%, %, 2.60%, 2.70%, 2.80%, 2.90%, 3.00%, 3.10%, 3.20%, 3.30%, 3.40%, 3.50%, 3.60%, 3.70%, 3.80%, 3.90%, 4.00%, 4.10%, 4.20%, 4.30%, 4.40%, 4.50%, 4.60%, 4.70%, 4.80%, 4.90%, 5.00%, 5.10%, 5.20% , 5.30%, 5.40%, 5.50%, 5.60%, 5.70%, 5.80%, 5.90%, 6.00%, 6.10%, 6.20%, 6.30%, 6.40%, 6.50%, 6.60%, 6.70%, 6.80%, 6.90%, 7.00%, 7.10%, 7.20%, 7.30%, 7.40%, 7.50%, 7.60%, 7.70%, 7.80%, 7.90%, 8.00%, 8.10%, 8.20%, 8.30%, 8.40%, 8.50%, 8.60%, 8.70%, 8.80%, 8.90%, 9.00%, 9.10%, 9.20%, 9.30%, 9.40%, 9.50%, 9.60%, 9.70%, 9.80%, 9.90%, 10%, 11%, 12%, 13%, 14%, 15%, 16%, 17%, 18%, 19% or 20%.

[0028] In some embodiments, the at least one hit peptide sequence comprises at least 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, 94, 95, 96, 97, 98, 99, 100, 101, 102, 103, 104, 105, 106, 107, 108, 109, 110, 111, 112, 113, 114, 115, 116, 117, 118, 119, 120, 121, 122, 123, 124, 125, 126, 127, 128, 129, 130, 131, 132, 133, 134, 135, 136, 137, 138, 139, 200, 201, 202, 203, 204, 205, 206, 207, 208, 209,

[0029] In some embodiments, the at least 19 decoy peptide sequences comprise at least 30, 40, 50, 60, 70, 80, 90, 100, 110, 120, 130, 140, 150, 160, 170, 180, 190, 200, 210, 220, 230, 240, 250, 260, 270, 280, 290, 300, 310, 320, 330, 340, 350, 360, 370, 380, 390, 400, 410, 420, 430, 440, 450, 460, 470, 480, 490, 500, 600, 700, 800, 900, 1000, 1100, 1200, 1300, 1400 , 1500, 1600, 1700, 1800, 1900, 2000, 2100, 2200, 2300, 2400, 2500, 2600, 2700, 2800, 2900, 3000, 3100, 3200, 3300, 3400, 3500, 3600, 3700, 3800, 3900 ,4000,4100,4200,4300,4400,4500,4600,4700,4800,4900,5000,5100,5200,5300,5400,5500,5600,5700,5800,5900,6000,6100,6200,6300,6400 , 6500, 6600, 6700, 6800, 6900, 7000, 7100, 7200, 7300, 7400, 7500, 7600, 7700, 7800, 7900, 8000, 8100, 8200, 8300, 8400, 8500, 8600, 8700, 8800, 8900 , 9000, 9100, 9200, 9300, 9400, 9500, 9600, 9700, 9800, 9900, 10000, 11000, 12000, 13000, 14000, 15000, 16000, 17000, 18000, 19000, 20000, 21000, 22 000, 23000, 24000, 25000, 26000, 27000, 28000, 29000, 30000, 31000, 32000, 33000, 34000, 35000, 36000, 37000, 38000, 39000, 40000, 41000, 42000, 4 3000、44000、45000、46000、47000、48000、49000、50000、52500、55000、57500、60000、62500、65000、67500、70000、72500、75000、77500、80000、82500、85,000, 87,500, 90,000, 92,500, 95,000, 97,500, 100,000, 125,000, 150,000, 175,000, 200,000, 225,000, 250,000, 275,000, 300,000, 325,000, 350,000, 375,000, 400,000, 425,000, 450,000, 475,000, 500,000, 600,000, 700,000, 800,000, 900,000, or 1,000,000 decoy peptide sequences.

[0030] In some embodiments, the at least 20 test peptide sequences comprise at least wherein the at least 500 test peptide sequences comprise at least 30, 40, 50, 60, 70, 80, 90, 100, 110, 120, 130, 140, 150, 160, 170, 180, 190, 200, 210, 220, 230, 240, 250, 260, 270, 280, 290, 300, 310, 320, 330, 340, 350, 360, 370, 380, 390, 400, 410, 420, 430, 440, 450, 460, 470, 480, 490, 500, 600, 700, 800, 900, 1000, 1000, 1000, 1000, 1000, 1000, 1000, 1000, 1000 1100、1200、1300、1400、1500、1600、1700、1800、1900、2000、2100、2200、2300、2400、2500、2600、2700、2800、2900、3000、3100、3200、3300、3400、3500、 3600、3700、3800、3900、4000、4100、4200、4300、4400、4500、4600、4700、4800、4900、5000、5100、5200、5300、5400、5500、5600、5700、5800、5900、6000、 6100、6200、6300、6400、6500、6600、6700、6800、6900、7000、7100、7200、7300、7400、7500、7600、7700、7800、7900、8000、8100、8200、8300、8400、8500、 8600、8700、8800、8900、9000、9100、9200、9300、9400、9500、9600、9700、9800、9900、10000、11000、12000、13000、14000、15000、16000、17000、18000、1 9000、20000、21000、22000、23000、24000、25000、26000、27000、28000、29000、30000、31000、32000、33000、34000、35000、36000、37000、38000、39000、 40000、41000、42000、43000、44000、45000、46000、47000、48000、49000、50000、52500、55000、57500、60000、62500、65000、67500、70000、72500、75000、77,500, 80,000, 82,500, 85,000, 87,500, 90,000, 92,500, 95,000, 97,500, 100,000, 125,000, 150,000, 175,000, 200,000, 225,000, 250,000, 275,000, 300,000, 325,000, 350,000, 375,000, 400,000, 425,000, 450,000, 475,000, 500,000, 600,000, 700,000, 800,000, 900,000, or 1,000,000 test peptide sequences.

[0031] In some embodiments, the top percentage is the top 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%, or 40%.

[0032] In some embodiments, for proteins encoded by corresponding HLA alleles of Table 11, the PPV is greater than the corresponding PPV of Column 2 of Table 11. In some embodiments, for proteins encoded by corresponding HLA alleles of Table 11, the PPV is at least equal to the corresponding PPV of Column 3 of Table 11.

[0033] In some embodiments, for a protein encoded by an HLA class II allele, the PPV is equal to or greater than the corresponding PPV of column 2 of Table 12.

[0034] In some embodiments, for a protein encoded by an HLA class II allele, the PPV is greater than the corresponding PPV in column 2 of Table 16.

[0035] In some embodiments, the subject is a single subject.

[0036] In some embodiments, the subject is a mammal.

[0037] In some embodiments, the subject is a human.

[0038] In some embodiments, the training cell is a cell that expresses a single protein encoded by an HLA class II allele of the subject cell.

[0039] In some embodiments, the training cell is a monoallelic HLA cell, or a cell expressing an HLA allele with an affinity tag.

[0040] In some embodiments, the subject's cells include cancer cells.

[0041] In some embodiments, the methods are used to identify peptide sequences.

[0042] In some embodiments, the methods are used to select peptide sequences.

[0043] In some embodiments, the methods are used to prepare for cancer treatment.

[0044] In some embodiments, the methods are used to prepare a subject for a specific cancer treatment.

[0045] In some embodiments, the methods are used to prepare cancer cell specific cancer treatments.

[0046] In some embodiments, each peptide sequence in the plurality of peptide sequences is associated with cancer.

[0047] In some embodiments, at least one peptide sequence in the plurality of peptide sequences is overexpressed by a cancer cell in the subject.

[0048] In some embodiments, each peptide sequence in the plurality of peptide sequences is overexpressed by a cancer cell in the subject.

[0049] In some embodiments, at least one peptide sequence in the plurality of peptide sequences is a cancer cell specific peptide.

[0050] In some embodiments, each peptide sequence in the plurality of peptide sequences is a cancer cell specific peptide.

[0051] In some embodiments, each peptide sequence in the plurality of peptide sequences is expressed by a cancer cell in the subject.

[0052] In some embodiments, at least one peptide sequence in the plurality of peptide sequences is not encoded by a non-cancer cell of the subject.

[0053] In some embodiments, each peptide sequence in the plurality of peptide sequences is not encoded by a non-cancer cell of the subject.

[0054] In some embodiments, at least one peptide sequence in the plurality of peptide sequences is not expressed by non-cancer cells of the subject.

[0055] In some embodiments, each peptide sequence in the plurality of peptide sequences is not expressed by non-cancer cells of the subject.

[0056] In some embodiments, the method includes obtaining the plurality of peptide sequences from a subject.

[0057] In some embodiments, the method includes obtaining a plurality of polynucleotide sequences from a subject.

[0058] In some embodiments, the method includes obtaining a plurality of polynucleotide sequences from a subject, the polynucleotide sequences encoding a plurality of peptide sequences encoded by the genome or exome of the subject or by a pathogen or virus in the subject.

[0059] In some embodiments, the method includes obtaining, by a computer processor, a plurality of polynucleotide sequences from a subject, the polynucleotide sequences encoding a plurality of peptide sequences encoded by the genome or exome of the subject.

[0060] In some embodiments, the method comprises obtaining a plurality of polynucleotide sequences of the subject by genome or exome sequencing.

[0061] In some embodiments, the method comprises obtaining a plurality of polynucleotide sequences of the subject by whole genome sequencing or whole exome sequencing.

[0062] In some embodiments, processing comprises processing with a computer processor.

[0063] In some embodiments, processing includes generating a plurality of predictor variables based at least on amino acid information of the plurality of peptide sequences.

[0064] In some embodiments, the plurality of predictor variables is processed using a machine learning HLA-peptide presentation prediction model.

[0065] In some embodiments, the one or more proteins encoded by the class II HLA alleles of cells of the subject are one or more proteins encoded by the class II HLA alleles expressed by the subject.

[0066] In some embodiments, the one or more proteins encoded by the HLA class II alleles of the subject's cells are one or more proteins encoded by the HLA class II alleles expressed by cancer cells of the subject.

[0067] In some embodiments, the one or more proteins encoded by the HLA class II alleles of the subject's cells is a single protein encoded by the HLA class II alleles of the subject's cells.

[0068] In some embodiments, the one or more proteins encoded by the HLA class II alleles of the subject's cells are two, three, four, five, or six or more proteins encoded by the HLA class II alleles of the subject's cells.

[0069] In some embodiments, the one or more proteins encoded by the HLA class II alleles of the subject's cells are every protein encoded by the HLA class II alleles of the subject's cells.

[0070] In some embodiments, the method further comprises administering to the subject a composition comprising one or more peptide sequences from the selected subset of peptide sequences.

[0071] In some embodiments, identifying the plurality of peptide sequences comprises comparing a DNA, RNA, or protein sequence from a cancer cell of the subject to a DNA, RNA, or protein sequence from a normal cell of the subject, wherein each of the plurality of peptides comprises at least one mutation that is present in the cancer cell of the subject but not in the normal cell of the subject.

[0072] In some embodiments, the machine learning HLA-peptide presentation prediction model includes a plurality of predictor variables identified based at least on training data, wherein the training data includes training peptide sequence information comprising amino acid position information, wherein the training peptide sequence information is associated with HLA proteins expressed in cells; and a function representing a relationship between the amino acid position information and a presentation likelihood generated as an output based on the amino acid position information and the plurality of predictor variables.

[0073] In some embodiments, identifying includes identifying, based at least on the multiple presentation predictions, a peptide sequence from the multiple peptide sequences whose probability of being presented by at least one of the one or more proteins encoded by the class II HLA alleles of the subject's cells is greater than a threshold presentation prediction probability value.

[0074] In some embodiments, the probability that one or more of 0.2% of the plurality of test peptide sequences predicted by the machine learning HLA peptide presentation prediction model to be presented is presented by at least one of the one or more proteins encoded by a class II HLA allele of a cell of the subject is greater than a threshold presentation prediction probability value.

[0075] In some embodiments, each of 0.2% of the plurality of test peptide sequences predicted by the machine learning HLA peptide presentation prediction model to be presented has a probability of being presented by at least one of the one or more proteins encoded by a class II HLA allele of a cell of the subject that is greater than a threshold presentation prediction probability value.

[0076] In some embodiments, the number of positives is limited to equal the number of hits.

[0077] In some embodiments, the mass spectrometry is monoallelic mass spectrometry.

[0078] In some embodiments, the peptide is presented by an HLA protein expressed in the cell via autophagy.

[0079] In some embodiments, the peptide is presented by an HLA protein expressed in a cell via phagocytosis.

[0080] In some embodiments, the plurality of predictor variables comprises predicted values ​​for expression levels of source proteins comprising the peptides.

[0081] In some embodiments, the plurality of predictor variables comprises a stability prediction value for a source protein comprising the peptide.

[0082] In some embodiments, the plurality of predictor variables comprises a predicted value for a degradation rate of a source protein comprising the peptide.

[0083] In some embodiments, the plurality of predictor variables comprises a predicted value for protein cleavage of a source protein comprising the peptide.

[0084] In some embodiments, the plurality of predictor variables comprises a prediction of a cellular or tissue localization of a source protein comprising the peptide.

[0085] In some embodiments, the plurality of predictor variables comprises a predictor of an intracellular processing pattern of a source protein comprising the peptide, wherein the processing pattern of the source protein comprises, among other things, a predictor of whether the source protein undergoes autophagy, phagocytosis, and intracellular transport.

[0086] In some embodiments, the quality of the training data is improved by using multiple quality metrics.

[0087] In some embodiments, the plurality of quality metrics include common contaminant peptide removal, high scoring peak intensity, high score, and high mass accuracy.

[0088] In some embodiments, the scored peak intensity is at least 50%.

[0089] In some embodiments, the scored peak intensity is at least 60%.

[0090] In some embodiments, the score is at least 7.

[0091] In some embodiments, the mass accuracy is at most 5 ppm.

[0092] In some embodiments, the peptide presented by an HLA protein expressed in a cell is a peptide presented by a single immunoprecipitated HLA protein expressed in the cell.

[0093] In some embodiments, the peptide presented by the HLA protein expressed in the cell is a peptide presented by a single exogenous HLA protein expressed in the cell.

[0094] In some embodiments, the peptide presented by the HLA protein expressed in the cell is a peptide presented by a single recombinant HLA protein expressed in the cell.

[0095] In some embodiments, the plurality of predictor variables includes a peptide-HLA affinity predictor variable.

[0096] In some embodiments, peptides presented by HLA proteins include peptides identified by searching a database of non-enzyme-specific non-modified peptides.

[0097] In some embodiments, peptides presented by HLA proteins include peptides identified by searching a peptide database using an inverse database searching strategy.

[0098] In some embodiments, the HLA protein comprises an HLA-DR, HLA-DQ, or HLA-DP protein.

[0099] In some embodiments, the HLA protein comprises an HLA class II protein selected from the group consisting of: HLA-DPB1*01:01 / HLA-DPA1*01:03, HLA-DPB1*02:01 / HLA-DPA1*01:03, HLA-DPB1*03:01 / HLA-DPA1*01:03, HLA-DPB1*04:01 / HLA-DPA1*01:03, HLA-DPB1*04:02 / HLA-DPA1*01:03, HLA-DPB1*06:01 / HLA-DPA1*01:03, HLA-DQB1*02:01 / HLA-DQA1*05:01, HLA-DQB1*02:02 / HLA-DQA1*02:01, HLA-DQB1*06:02 / HLA-DQA1*01:02, HLA-DQB1*06:04 / HLA-DQA1*01:02, HLA-DRB1*01:01, HLA-DRB1*01:02, HLA-DRB1*03:01, HLA-DRB1*03:02, HLA-DRB1*04:01, HLA-DRB1*04:02, HLA-DRB1*04:03, HLA-DRB1*04:04, HLA-DRB1*04:05, HLA-DRB1*04:07, HLA-DRB1*07:01, HLA-DRB1*08:01, HLA-DRB1*08:02, HLA-DRB1*08:03, HLA-DRB1*08:04, HLA-DRB1*09:01, HLA-DRB1*10:01, HLA-DRB1*11:01, HLA-DRB1*11:02, HLA-DRB1*11:04, HLA-DRB1*12:01, HLA-DRB1*12:02, HLA-DRB1*13:01, HLA-DRB1*13:02, HLA-DRB1*13:03, HLA-DRB1*14:01, HLA-DRB1*15:01, HLA-DRB1*15:02, HLA-DRB1*15:03, HLA-DRB1*16:01, HLA-DRB3*01:01, HLA-DRB3*02:02, HLA-DRB3*03:01, HLA-DRB4*01:01, HLA-DRB5*01:01.

[0100] In some embodiments, the HLA-DR pairs with DRA*01:01.

[0101] In some embodiments, the HLA protein is an HLA class II protein selected from the group consisting of DPA*01:03 / DPB*04:01, DRB1*01:01, DRB1*01:02, DRB1*03:01, DRB1*04:01, DRB1*04:02, DRB1*04:04, DRB1*04:05, DRB1*07:01, DRB1*08:01, DRB1*08:02, DRB1*08:03, DRB1*09:01, DRB1*11:01, DRB1*11:02, DRB 1*11:04, DRB1*12:01, DRB1*13:01, DRB1*13:02, DRB1*13:03, DRB1*14:01, DRB1*15:01, DRB1*15:02, DRB1*15 :03, DRB1*16:02, DRB3*01:01, DRB3*02:01, DRB3*02:02, DRB3*03:01, DRB4*01:01, DRB4*01:03 and DRB5*01:01.

[0102] In some embodiments, the HLA-DR protein comprises DRA*01:01 in dimeric form.

[0103] In some embodiments, the HLA protein includes an HLA-DP protein selected from the group consisting of DPB1*01:01, DPB1*02:01, DPB1*02:02, DPB1*03:01, DPB1*04:01, DPB1*04:02, DPB1*05:01, DPB1*06:01, DPB1*11:01, DPB1*13:01, and DPB1*17:01.

[0104] In some embodiments, the HLA-DP protein pair comprises DPA1*01:03.

[0105] In some embodiments, the HLA protein includes an HLA-DQ protein complex selected from the group consisting of A1*01:01+B1*05:01, A1*01:02+B1*06:02, A1*01:02+B1*06:04, A1*01:03+B1*06:03, A1*02:01+B1*02:02, A1*02:01+B1*03:03, A1*03:01+B1*03:02, A1*03:03+B1*03:01, A1*05:01+B1*02:01, and A1*05:05+B1*03:01.

[0106] In some embodiments, peptides presented by HLA proteins include peptides identified by comparing the MS / MS spectrum of the HLA-peptide to the MS / MS spectra of one or more peptides or proteins in a peptide or protein database.

[0107] In some embodiments, the mutation is selected from a point mutation, a splice site mutation, a frameshift mutation, a read-through mutation, and a gene fusion mutation.

[0108] In some embodiments, the peptide presented by the HLA protein has a length of 15-40 amino acids.

[0109] In some embodiments, peptides presented by HLA proteins include peptides identified by comparing the MS / MS spectrum of the HLA-peptide with the MS / MS spectra of one or more peptides or proteins in a peptide or protein database to identify peptides presented by HLA proteins.

[0110] In some embodiments, the personalized cancer therapy further comprises an adjuvant.

[0111] In some embodiments, the personalized cancer therapy further comprises an immune checkpoint inhibitor.

[0112] In some embodiments, the training data includes structured data, time series data, unstructured data, relational data, or any combination thereof.

[0113] In some embodiments, the unstructured data includes image data.

[0114] In some embodiments, the relationship data includes data from a client system, an enterprise system, an operating system, a website, a web-accessible application programming interface (API), or any combination thereof.

[0115] In some embodiments, the training data is uploaded to a cloud-based database.

[0116] In some embodiments, the training is performed using a convolutional neural network.

[0117] In some embodiments, the convolutional neural network comprises at least two convolutional layers.

[0118] In some embodiments, the convolutional neural network includes at least one batch normalization step.

[0119] In some embodiments, the convolutional neural network includes at least one spatial dropout step.

[0120] In some embodiments, the convolutional neural network includes at least one global maximum pooling step.

[0121] In some embodiments, the convolutional neural network comprises at least one dense layer.

[0122] In some embodiments, identifying a peptide sequence comprises identifying a peptide sequence having a mutation that is expressed in a cancer cell of the subject.

[0123] In some embodiments, identifying a peptide sequence comprises identifying a peptide sequence that is not expressed in normal cells of the subject.

[0124] In some embodiments, identifying a peptide sequence comprises identifying a viral peptide sequence.

[0125] In some embodiments, identifying a peptide sequence comprises identifying an overexpressed peptide sequence.

[0126] Provided herein is a method for identifying HLA class II specific peptides for immunotherapy of a subject, comprising: obtaining, by a computer processor, a candidate peptide comprising an epitope and a plurality of peptide sequences, each of which comprises the epitope; using a machine learning HLA-peptide presentation prediction model, processing, by a computer processor, amino acid information of the plurality of peptide sequences to generate a presentation prediction for each of the plurality of peptide sequences to immune cells, each presentation prediction indicating the likelihood that one or more proteins encoded by an HLA class II allele can present a given peptide sequence in the plurality of peptide sequences, wherein the machine learning HLA-peptide presentation prediction model is trained using training data, the training data comprising sequence information of peptides presented by HLA proteins expressed in cells and identified by mass spectrometry; obtaining a candidate peptide comprising an epitope and a plurality of peptide sequences, each of which comprises the epitope; and obtaining a candidate peptide comprising an epitope and a plurality of peptide sequences, each of which comprises the epitope; and obtaining a candidate peptide comprising an epitope and a plurality of peptide sequences, each of which comprises the epitope; and obtaining a candidate peptide comprising an epitope and a plurality of peptide sequences, each of which comprises the epitope; Selecting a protein from one or more proteins encoded by a class II allele, the protein being predicted to bind to the candidate peptide by a machine learning HLA-peptide presentation prediction model, wherein the probability of the protein presenting the candidate peptide to immune cells is greater than a threshold presentation prediction probability value; contacting the candidate peptide with the selected protein such that the candidate peptide competes for a placeholder peptide associated with the selected protein; and identifying the candidate peptide as a peptide for immunotherapy that is specific to the selected protein based on whether the candidate peptide replaces the placeholder peptide.

[0127] In some embodiments, obtaining includes identifying a candidate peptide, wherein identifying the candidate peptide includes comparing a DNA, RNA, or protein sequence from a cancer cell of the subject to a DNA, RNA, or protein sequence from a normal cell of the subject.

[0128] In some embodiments, processing includes identifying a plurality of predictor variables based at least on amino acid information of the plurality of peptide sequences, and processing the plurality of predictor variables using a machine learning HLA-peptide presentation prediction model.

[0129] In some embodiments, the machine learning HLA-peptide presentation prediction model includes a plurality of predictor variables identified based at least on training data, wherein the training data includes training peptide sequence information comprising amino acid position information, wherein the training peptide sequence information is associated with HLA proteins expressed in cells; and a function representing a relationship between the amino acid position information and a presentation likelihood generated as an output based on the amino acid position information and the plurality of predictor variables.

[0130] In some embodiments, the number of positives is limited to equal the number of hits.

[0131] In some embodiments, the mass spectrometry is monoallelic mass spectrometry.

[0132] In some embodiments, the plurality of predictive variables include any one or more of: a predicted value for expression level of a source protein comprising the peptide, a predicted value for stability, a predicted value for degradation rate, a predicted value for cleavage, a predicted value for cell or tissue localization, and a predicted value for intracellular processing patterns including autophagy, phagocytosis, and intracellular transport.

[0133] In some embodiments, the quality of the training data is improved by using multiple quality metrics.

[0134] In some embodiments, the plurality of quality metrics include common contaminant peptide removal, high scoring peak intensity, high score, and high mass accuracy.

[0135] In some embodiments, the scored peak intensity is at least 50%.

[0136] In some embodiments, the scored peak intensity is at least 60%.

[0137] In some embodiments, the placeholder peptide is a CLIP peptide.

[0138] In some embodiments, the placeholder peptide is a CMV peptide.

[0139] In some embodiments, the method further comprises measuring the IC50 of replacement of the placeholder peptide by the peptide of interest.

[0140] In some embodiments, the IC50 of the placeholder peptide being replaced by the target peptide is less than 500 nM.

[0141] In some embodiments, at least one protein from one or more proteins encoded by an HLA class II allele of a cell of the subject is an HLA class II tetramer or multimer.

[0142] In some embodiments, the target peptide is further identified by mass spectrometry.

[0143] In some embodiments, the at least one protein encoded by an HLA class II allele of a cell of the subject is a recombinant protein.

[0144] In some embodiments, the at least one protein encoded by the HLA class II allele of the subject's cells is expressed in a eukaryotic cell.

[0145] In some embodiments, the peptide is presented by an HLA protein expressed in the cell via autophagy.

[0146] In some embodiments, the peptide is presented by an HLA protein expressed in a cell via phagocytosis.

[0147] In some embodiments, the peptide presented by an HLA protein expressed in a cell is a peptide presented by a single immunoprecipitated HLA protein expressed in the cell.

[0148] In some embodiments, the peptide presented by the HLA protein expressed in the cell is a peptide presented by a single exogenous HLA protein expressed in the cell.

[0149] In some embodiments, the peptide presented by the HLA protein expressed in the cell is a peptide presented by a single recombinant HLA protein expressed in the cell.

[0150] In some embodiments, the plurality of predictor variables includes a peptide-HLA affinity predictor variable.

[0151] In some embodiments, peptides presented by HLA proteins include peptides identified by searching a database of non-enzyme-specific non-modified peptides.

[0152] In some embodiments, peptides presented by HLA proteins include peptides identified by searching a peptide database using an inverse database searching strategy.

[0153] In some embodiments, the HLA protein comprises an HLA-DR, HLA-DQ, or HLA-DP protein.

[0154] In some embodiments, the immunotherapy is cancer immunotherapy.

[0155] In some embodiments, the epitope is a cancer-specific epitope.

[0156] In some embodiments, the at least one protein encoded by the HLA class II allele comprises at least an α1 subunit and a β1 subunit of an HLA protein that exist in the form of a dimer.

[0157] In some embodiments, the identity of the peptide is known.

[0158] In some embodiments, the identity of the peptide is unknown.

[0159] In some embodiments, the identity of the peptide is determined by mass spectrometry.

[0160] In some embodiments, the peptide exchange assay comprises detecting a peptide fluorescent probe or tag.

[0161] In some embodiments, the placeholder peptide is a CLIP peptide.In some embodiments, the placeholder peptide has the amino acid sequence PVSKMRMATPLLMQA (SEQ ID NO: 1).

[0162] In some embodiments, the polynucleic acid construct comprises an expression vector, which further comprises one or more of the following: a promoter, a secretion signal, a dimerization factor, a ribosomal skipping sequence, and one or more tags for purification and / or detection.

[0163] In some embodiments, the placeholder peptide sequence is encoded by a nucleic acid sequence within a vector.

[0164] In some embodiments, the sequence encoding the cleavable domain is placed between the sequence encoding the placeholder peptide and the sequence encoding the HLAβ1 peptide.

[0165] Provided herein is a method for determining the immunogenicity of an MHC class II binding peptide, comprising: selecting a protein encoded by an HLA class II allele that is predicted to bind to an MHC class II binding peptide by a machine learning HLA-peptide presentation prediction model, wherein the machine learning HLA-peptide presentation prediction model is configured to generate a presentation prediction for a given peptide sequence, the presentation prediction indicating the likelihood that one or more proteins encoded by the HLA class II allele can present the given peptide sequence, and wherein the probability that the protein presents the MHC class II binding peptide is greater than a threshold presentation prediction probability value; contacting the peptide with the selected protein such that the peptide competes for a placeholder peptide associated with the selected protein and displaces the placeholder peptide, thereby forming a complex comprising the HLA class II protein and the MHC class II binding peptide; contacting the complex with a CD4+ T cell, and determining one or more activation parameters of the CD4+ T cell, the parameters being selected from: induction of cytokines, induction of chemokines, and expression of cell surface markers.

[0166] In some embodiments, the HLA class II allele is a tetramer or a multimer.

[0167] In some embodiments, the cytokine is IL-2.

[0168] The present invention provides a method for inducing CD4+ T cell activation in a subject for cancer immunotherapy, the method comprising: identifying a peptide sequence associated with cancer and comprising a cancer mutation, wherein identifying the peptide sequence comprises comparing a DNA, RNA or protein sequence from a cancer cell of the subject with a DNA, RNA or protein sequence from a normal cell of the subject; selecting a protein encoded by an HLA class II allele, the protein being normally expressed by the cells of the subject and predicted to bind to the peptide by a machine learning HLA-peptide presentation prediction model; wherein the prediction model has a positive predictive value of at least 0.1 at a recall rate of at least 0.1%, 0.1%-50%, or at most 50%, and wherein the probability of the protein presenting the identified peptide sequence is greater than a threshold presentation prediction probability value; contacting the identified peptide with the selected protein encoded by the HLA class II allele to verify whether the identified peptide competes with the selected protein encoded by the HLA class II allele. The invention relates to a method for treating a subject with a placeholder peptide associated with a selected protein encoded by a class II allele, thereby replacing the placeholder peptide with an IC50 value of less than 500 nM; optionally, purifying the identified peptide; and administering to the subject an effective amount of a polypeptide comprising the sequence of the identified peptide or a polynucleotide encoding the polypeptide.

[0169] Provided herein is a method for screening a drug comprising a polypeptide sequence for immunogenicity in a subject, comprising: obtaining, by a computer processor, a plurality of peptide sequences of the polypeptide sequence; processing, by a computer processor, the amino acid information of the plurality of peptide sequences using a machine learning HLA-peptide presentation prediction model to generate a presentation prediction for each of the plurality of peptide sequences, each presentation prediction indicating a likelihood that one or more proteins encoded by a class I or class II MHC allele of a cell of the subject can present an epitope sequence of a given peptide sequence in the plurality of peptide sequences, wherein the machine learning HLA-peptide presentation prediction model is trained using training data comprising sequence information associated with HLA proteins expressed in cells; determining or predicting, based on the plurality of presentation predictions, that each of the plurality of peptide sequences of the polypeptide sequence is not immunogenic to the subject; and administering to the subject a composition comprising the drug.

[0170] Provided herein is a method for preparing HLA class II tetramers or multimers by conjugating four individual HLA protein α1 and β1 heterodimers, the method comprising: expressing in eukaryotic cells a vector comprising nucleic acid sequences encoding the α and β chains of the HLA proteins, a secretion signal, a biotinylation motif, and at least one tag for identification or purification, such that each HLA protein α1 and β1 heterodimer is secreted in a dimerized state, wherein the heterodimer is associated with a placeholder peptide, purifying the secreted heterodimers from the cell culture medium, verifying the peptide binding activity using a peptide exchange assay, adding streptavidin, thereby conjugating the heterodimers into tetramers, purifying the tetramers, and having a yield greater than 1 mg / L. Multimers, such as pentamers, hexamers, or octamers, may also be produced similarly, and they are also considered herein.

[0171] In some embodiments, the vector comprises a CMV promoter.

[0172] In some embodiments, the vector comprises a sequence encoding a placeholder peptide linked to the β1 strand via a cleavable site.

[0173] In some embodiments, the peptide exchange assay involves pre-cleavage of the placeholder peptide from the beta strand.

[0174] In some embodiments, the cleavable site is a thrombin cleavage site.

[0175] In some embodiments, the peptide exchange assay is a FRET assay.

[0176] In some embodiments, the purification is performed by any of the following: column chromatography, ion exchange chromatography, size exclusion chromatography, affinity chromatography, or LC-MS.

[0177] Provided herein is an HLA class II tetramer or multimer comprising HLA-DR or HLA-DP or HLA-DQ heterodimers, each heterodimer comprising an α and a β chain, wherein the heterodimers are purified and present at a concentration greater than 1 mg / L.

[0178] In some embodiments, the HLA class II tetramer is selected from Tables 8A-8C.

[0179] In some embodiments, the HLA class II tetramer comprises a heterodimeric pair selected from the group consisting of HLA-DR, HLA-DP, and HLA-DQ proteins.

[0180] In some embodiments, the HLA protein is an HLA class II protein selected from the group consisting of HLA-DPB1*01:01 / HLA-DPA1*01:03, HLA-DPB1*02:01 / HLA-DPA1*01:03, HLA-DPB1*03:01 / HLA-DPA1*01:03, HLA-DPB1*04:01 / HLA-DPA1*01:03, HLA-DPB1*04:02 / HLA-DPA1*01:03, HLA-DPB1*06:01 / HLA-DPA1*01:03, HLA-DQB1*02:01 / HLA-DQ A1*05:01, HLA-DQB1*02:02 / HLA-DQA1*02:01, HLA-DQB1*06:02 / HLA-DQA1*01:02, HLA-DQB1*06:04 / HLA-DQA1*01:02, HLA-D RB1*01:01, HLA-DRB1*01:02, HLA-DRB1*03:01, HLA-DRB1*03:02, HLA-DRB1*04:01, HLA-DRB1*04:02, HLA-DRB1*04:03, HLA- DRB1*04:04, HLA-DRB1*04:05, HLA-DRB1*04:07, HLA-DRB1*07:01, HLA-DRB1*08:01, HLA-DRB1*08:02, HLA-DRB1*08:03, HLA -DRB1*08:04, HLA-DRB1*09:01, HLA-DRB1*10:01, HLA-DRB1*11:01, HLA-DRB1*11:02, HLA-DRB1*11:04, HLA-DRB1*12:01, HL A-DRB1*12:02, HLA-DRB1*13:01, HLA-DRB1*13:02, HLA-DRB1*13:03, HLA-DRB1*14:01, HLA-DRB1*15:01, HLA-DRB1*15:02, H LA-DRB1*15:03, HLA-DRB1*16:01, HLA-DRB3*01:01, HLA-DRB3*02:02, HLA-DRB3*03:01, HLA-DRB4*01:01 and HLA-DRB5*01:01.

[0181] In some embodiments, the heterodimeric pair is expressed in a eukaryotic cell.

[0182] In some embodiments, the heterodimer pair is encoded by a vector.

[0183] Provided herein is a vector, wherein the vector comprises a nucleic acid sequence encoding the α chain and β chain of the HLA protein described herein, a secretion signal, a biotinylation motif and at least one tag for identification or purification, such that each HLA protein α1 and β1 heterodimer is secreted in a dimerized state, wherein the secreted heterodimer is optionally associated with a placeholder peptide.

[0184] Provided herein is a cell comprising the vector described herein.

[0185] In some embodiments, the HLA class II heterodimer is secreted from eukaryotic cells into the cell culture medium, which is further purified by any of the following methods: column chromatography, ion exchange chromatography, size exclusion chromatography, affinity chromatography, or LC-MS.

[0186] Provided herein is a method for screening a drug comprising a polypeptide sequence for immunogenicity in a subject, comprising: obtaining, by a computer processor, a plurality of peptide sequences of the polypeptide sequence; processing, by a computer processor, amino acid information of the plurality of peptide sequences using a machine learning HLA-peptide presentation prediction model to generate a presentation prediction for each of the plurality of peptide sequences, each presentation prediction indicating a likelihood that one or more proteins encoded by a class I or class II MHC allele of a cell of the subject can present an epitope sequence of a given peptide sequence in the plurality of peptide sequences, wherein the machine learning HLA-peptide presentation prediction model is trained using training data, the training data comprising sequence information of sequences of peptides presented by HLA proteins expressed in cells and identified by mass spectrometry; and determining or predicting, based on the plurality of presentation predictions, that at least one of the plurality of peptide sequences of the polypeptide sequence is immunogenic to the subject.

[0187] Provided herein is a method for screening a drug comprising a polypeptide sequence for immunogenicity in a subject, the method comprising: using a computer processor to input amino acid information of a peptide sequence of the polypeptide sequence into a machine learning HLA-peptide presentation prediction model to generate a set of presentation predictions for the peptide sequence, each presentation prediction representing a probability that an epitope sequence of a given peptide sequence is presented by one or more proteins encoded by a class I or class II MHC allele of a cell of the subject; wherein the machine learning HLA-peptide presentation prediction model comprises: a plurality of predictor variables identified at least based on training data; wherein the training data comprises: sequence information of sequences of peptides presented by HLA proteins expressed in cells and identified by mass spectrometry; training peptide sequence information comprising amino acid position information, wherein the training peptide sequence information is associated with HLA proteins expressed in cells; and a function representing a relationship between the amino acid position information received as input and the presentation likelihood generated as output based on the amino acid position information and the predictor variables; based on the set of presentation predictions, determining or predicting that each of the peptide sequences of the polypeptide sequence is not immunogenic to the subject; and administering a composition comprising the drug to the subject.

[0188] Provided herein is a method for screening a drug comprising a polypeptide sequence for immunogenicity in a subject, the method comprising: using a computer processor to input amino acid information of a peptide sequence of the polypeptide sequence into a machine learning HLA-peptide presentation prediction model to generate a set of presentation predictions for the peptide sequence, each presentation prediction representing the probability that an epitope sequence of a given peptide sequence is presented by one or more proteins encoded by a class I or class II MHC allele of a cell of the subject; wherein the machine learning HLA-peptide presentation prediction model comprises: a plurality of predictor variables identified at least based on training data; wherein the training data comprises: sequence information of sequences of peptides presented by HLA proteins expressed in cells and identified by mass spectrometry; training peptide sequence information comprising amino acid position information, wherein the training peptide sequence information is associated with HLA proteins expressed in cells; and a function representing a relationship between the amino acid position information received as input and the presentation likelihood generated as output based on the amino acid position information and the predictor variables; based on the set of presentation predictions, determining or predicting that at least one of the peptide sequences of the polypeptide sequence is immunogenic to the subject.

[0189] Provided herein is a method for screening a drug comprising a polypeptide sequence for immunogenicity in a subject, comprising: obtaining, by a computer processor, a plurality of peptide sequences of the polypeptide sequence; processing, by a computer processor, the amino acid information of the plurality of peptide sequences using a machine learning HLA-peptide presentation prediction model to generate a presentation prediction for each of the plurality of peptide sequences, each presentation prediction indicating a likelihood that one or more proteins encoded by a class I or class II MHC allele of a cell of the subject can present an epitope sequence of a given peptide sequence in the plurality of peptide sequences, wherein the machine learning HLA-peptide presentation prediction model is trained using training data comprising sequence information associated with HLA proteins expressed in cells; determining or predicting, based on the plurality of presentation predictions, that each of the plurality of peptide sequences of the polypeptide sequence is not immunogenic to the subject; and administering to the subject a composition comprising the drug.

[0190] In some embodiments, the method further comprises deciding not to administer the drug to the subject.

[0191] In some embodiments, the medicament comprises an antibody or a binding fragment thereof.

[0192] In some embodiments, the peptide sequence of the polypeptide sequence has a length of 8, 9, 10, 11 or 12 amino acids, and wherein the protein encoded by a class I or class II MHC allele of a cell of the subject is a protein encoded by a class I MHC allele of a cell of the subject.

[0193] In some embodiments, the peptide sequence of the polypeptide sequence has a length of 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, or 25 amino acids, and wherein the protein encoded by a class I or class II MHC allele of a cell of the subject is a protein encoded by a class II MHC allele of a cell of the subject.

[0194] Provided herein is a method of treating a subject having an autoimmune disease or condition, comprising: (a) identifying or predicting an epitope of an expressed protein presented by class I or class II MHC of a cell of the subject, wherein a complex comprising the identified or predicted epitope and class I or class II MHC is targeted by a CD8 or CD4 T cell of the subject; (b) identifying a T cell receptor (TCR) that binds to the complex; (c) expressing the TCR in a regulatory T cell or an allogeneic regulatory T cell from the subject; and (d) administering to the subject a regulatory T cell expressing the TCR.

[0195] In some embodiments, the autoimmune disease or condition is diabetes.

[0196] In some embodiments, the cells are pancreatic islet cells.

[0197] Provided herein is a method of treating a subject having an autoimmune disease or condition, comprising administering to the subject a regulatory T cell expressing a T cell receptor (TCR) that binds to a complex comprising: (i) an epitope of an expressed protein identified or predicted to be presented by class I or class II MHC of a cell of the subject, and (ii) class I or class II MHC, wherein the complex is targeted by a CD8 or CD4 T cell of the subject.

[0198] Provided herein is a computer system for identifying peptide sequences for personalized cancer treatment of a subject, comprising: a database configured to store a plurality of peptide sequences of a subject; and one or more computer processors operably coupled to the database, wherein the one or more computer processors are individually and collectively programmed to: process amino acid information of the plurality of peptide sequences using a machine learning HLA-peptide presentation prediction model to generate a presentation prediction for each of the plurality of peptide sequences, each presentation prediction indicating a likelihood that one or more proteins encoded by class II MHC alleles of a cell of the subject are capable of presenting a given peptide sequence in the plurality of peptide sequences, wherein the machine learning HLA-peptide presentation prediction model is trained using training data, the training data comprising sequence information of sequences of peptides presented by HLA proteins expressed in cells and identified by mass spectrometry; and select a subset of the plurality of peptide sequences for personalized cancer treatment of the subject based at least on the plurality of presentation predictions.

[0199] Provided herein is a computer system for identifying HLA class II specific peptides for immunotherapy of a subject, comprising: a database configured to store candidate peptides comprising an epitope and a plurality of peptide sequences, each peptide sequence comprising the epitope; and one or more computer processors operably coupled to the database, wherein the one or more computer processors are individually and collectively programmed to: process amino acid information of the plurality of peptide sequences using a machine learning HLA-peptide presentation prediction model to generate a presentation prediction for each of the plurality of peptide sequences to immune cells, each presentation prediction indicating a likelihood that one or more proteins encoded by an HLA class II allele are capable of presenting a given peptide sequence in the plurality of peptide sequences, wherein the machine learning HLA-peptide presentation prediction model is trained using training data, the training data comprising sequence information of peptides presented by HLA proteins expressed in cells and identified by mass spectrometry; and Selecting a protein from one or more proteins encoded by class II alleles, wherein the protein is predicted to bind to the candidate peptide by a machine learning HLA-peptide presentation prediction model, wherein the probability of the protein presenting the candidate peptide to immune cells is greater than a threshold presentation prediction probability value; and identifying the candidate peptide as a peptide for immunotherapy specific to the selected protein based on whether the candidate peptide replaces the placeholder peptide when the candidate peptide contacts the selected protein so that the candidate peptide competes for the placeholder peptide associated with the selected protein.

[0200] Provided herein is a computer system for screening a drug comprising a polypeptide sequence for immunogenicity in a subject, comprising: a database configured to store a plurality of peptide sequences of the polypeptide sequence; and one or more computer processors operably coupled to the database, wherein the one or more computer processors are individually and collectively programmed to: process the amino acid information of the plurality of peptide sequences using a machine learning HLA-peptide presentation prediction model to generate a presentation prediction for each of the plurality of peptide sequences, each presentation prediction indicating a likelihood that one or more proteins encoded by a class I or class II MHC allele of a cell of the subject is capable of presenting an epitope sequence of a given peptide sequence in the plurality of peptide sequences, wherein the machine learning HLA-peptide presentation prediction model is trained using training data, the training data comprising sequence information associated with HLA proteins expressed in cells; and based on the plurality of presentation predictions, determine or predict that each of the plurality of peptide sequences of the polypeptide sequence is not immunogenic to the subject, wherein a composition comprising the drug is administered to the subject.

[0201] Provided herein is a computer system for screening drugs comprising polypeptide sequences for immunogenicity in a subject, comprising: a database configured to store a plurality of peptide sequences of the polypeptide sequences; and one or more computer processors operably coupled to the database, wherein the one or more computer processors are individually and collectively programmed to: process amino acid information of the plurality of peptide sequences using a machine learning HLA-peptide presentation prediction model to generate a presentation prediction for each of the plurality of peptide sequences, each presentation prediction indicating a likelihood that one or more proteins encoded by a class I or class II MHC allele of a cell of the subject is capable of presenting an epitope sequence of a given peptide sequence in the plurality of peptide sequences, wherein the machine learning HLA-peptide presentation prediction model is trained using training data, the training data comprising sequence information of sequences of peptides presented by HLA proteins expressed in cells and identified by mass spectrometry; and based on the plurality of presentation predictions, determine or predict that at least one of the plurality of peptide sequences of the polypeptide sequence is immunogenic to the subject.

[0202] Provided herein is a non-transitory computer-readable medium comprising machine-executable code that, when executed by one or more computer processors, implements a method for identifying peptide sequences for personalized cancer treatment of a subject, the method comprising: obtaining a plurality of peptide sequences for the subject; processing amino acid information of the plurality of peptide sequences using a machine learning HLA-peptide presentation prediction model to generate a presentation prediction for each of the plurality of peptide sequences, each presentation prediction indicating a likelihood that one or more proteins encoded by class II MHC alleles of a cell of the subject are capable of presenting a given peptide sequence in the plurality of peptide sequences, wherein the machine learning HLA-peptide presentation prediction model is trained using training data comprising sequence information of sequences of peptides presented by HLA proteins expressed in cells and identified by mass spectrometry; and selecting a subset of the plurality of peptide sequences for personalized cancer treatment of the subject based at least on the plurality of presentation predictions.

[0203] Provided herein is a non-transitory computer-readable medium comprising machine-executable code, which when executed by one or more computer processors implements a method for identifying HLA class II-specific peptides for immunotherapy of a subject, the method comprising: obtaining a candidate peptide comprising an epitope and a plurality of peptide sequences, each peptide sequence comprising the epitope; processing amino acid information of the plurality of peptide sequences using a machine learning HLA-peptide presentation prediction model to generate a presentation prediction for each of the plurality of peptide sequences to immune cells, each presentation prediction indicating a likelihood that one or more proteins encoded by an HLA class II allele can present a given peptide sequence in the plurality of peptide sequences, wherein the machine learning HLA-peptide presentation prediction model is trained using training data, the training data comprising sequence information of peptides presented by HLA proteins expressed in cells and identified by mass spectrometry; obtaining a candidate peptide comprising an epitope and a plurality of peptide sequences, each peptide sequence comprising the epitope; processing amino acid information of the plurality of peptide sequences using a machine learning HLA-peptide presentation prediction model to generate a presentation prediction for each of the plurality of peptide sequences to immune cells, each presentation prediction indicating a likelihood that one or more proteins encoded by an HLA class II allele can present a given peptide sequence in the plurality of peptide sequences; Selecting a protein from one or more proteins encoded by class II alleles, wherein the protein is predicted to bind to the candidate peptide by a machine learning HLA-peptide presentation prediction model, wherein the probability of the protein presenting the candidate peptide to immune cells is greater than a threshold presentation prediction probability value; and identifying the candidate peptide as a peptide for immunotherapy specific to the selected protein based on whether the candidate peptide replaces the placeholder peptide when the candidate peptide contacts the selected protein so that the candidate peptide competes with the placeholder peptide.

[0204] Provided herein is a non-transitory computer-readable medium comprising machine-executable code that, when executed by one or more computer processors, implements a method for screening a drug comprising a polypeptide sequence for immunogenicity in a subject, the method comprising: obtaining a plurality of peptide sequences of the polypeptide sequence; processing amino acid information of the plurality of peptide sequences using a machine learning HLA-peptide presentation prediction model to generate a presentation prediction for each of the plurality of peptide sequences, each presentation prediction indicating a likelihood that one or more proteins encoded by a class I or class II MHC allele of a cell of the subject is capable of presenting an epitope sequence of a given peptide sequence in the plurality of peptide sequences, wherein the machine learning HLA-peptide presentation prediction model is trained using training data comprising sequence information associated with HLA proteins expressed in cells; and determining or predicting, based on the plurality of presentation predictions, that each of the plurality of peptide sequences of the polypeptide sequence is not immunogenic to the subject, wherein a composition comprising the drug is administered to the subject.

[0205] Provided herein is a non-transitory computer-readable medium comprising machine-executable code, which when executed by one or more computer processors implements a method for screening a drug comprising a polypeptide sequence for immunogenicity in a subject, the method comprising: obtaining a plurality of peptide sequences of the polypeptide sequence; processing amino acid information of the plurality of peptide sequences using a machine learning HLA-peptide presentation prediction model to generate a presentation prediction for each of the plurality of peptide sequences, each presentation prediction indicating a likelihood that one or more proteins encoded by a class I or class II MHC allele of a cell of the subject can present an epitope sequence of a given peptide sequence in the plurality of peptide sequences, wherein the machine learning HLA-peptide presentation prediction model is trained using training data, the training data comprising sequence information of sequences of peptides presented by HLA proteins expressed in cells and identified by mass spectrometry; and determining or predicting, based on the plurality of presentation predictions, that at least one of the plurality of peptide sequences of the polypeptide sequence is immunogenic to the subject.

[0206] Provided herein is a method comprising: processing amino acid information of a plurality of candidate peptide sequences using a machine learning HLA peptide presentation prediction model to generate a plurality of presentation predictions, wherein each candidate peptide sequence in the plurality of candidate peptide sequences is encoded by a genome or exome of a subject, wherein the plurality of presentation predictions comprises an HLA presentation prediction for each of the plurality of candidate peptide sequences, wherein each presentation prediction indicates a likelihood that one or more proteins encoded by a class II HLA allele of a cell of the subject is capable of presenting a given candidate peptide sequence in the plurality of candidate peptide sequences, wherein the machine learning HLA peptide presentation prediction model is trained using training data comprising sequence information of training peptide sequences, the training peptides being identified by mass spectrometry as to be presented by an HLA protein expressed in the training cells; and identifying, based at least on the plurality of presentation predictions, a peptide sequence in the plurality of peptide sequences whose probability of being presented by at least one of the one or more proteins encoded by a class II HLA allele of a cell of the subject is greater than a threshold presentation prediction probability value; wherein when the amino acid information of a plurality of test peptide sequences is processed to generate a plurality of test presentation predictions, each test presentation prediction indicating the likelihood that one or more proteins encoded by a class II HLA allele of a cell of the subject is able to present a given test peptide sequence in the plurality of test peptide sequences, the machine learning HLA peptide presentation prediction model has a positive predictive value (PPV) of at least 0.07, wherein the plurality of test peptide sequences comprises at least 500 test peptide sequences, the sequences comprising (i) at least one hit peptide sequence identified by mass spectrometry as to be presented by an HLA protein expressed in a cell, and (ii) at least 499 decoy peptide sequences contained in a protein encoded by a genome of an organism, wherein the organism and the subject are of the same species, wherein the ratio of the at least one hit peptide sequence in the plurality of test peptide sequences to the at least 499 decoy peptide sequences is 1:499, and according to the machine learning HLA peptide presentation prediction model, 0.2% of the plurality of test peptide sequences are predicted to be presented by an HLA protein expressed in a cell.

[0207] Provided herein is a method comprising: processing amino acid information of a plurality of peptide sequences encoded by a genome or exome of a subject using a machine learning HLA peptide binding prediction model to generate a plurality of binding predictions, wherein the plurality of binding predictions comprises an HLA binding prediction for each of the plurality of candidate peptide sequences, each binding prediction indicating a likelihood that one or more proteins encoded by a class II HLA allele of a cell of the subject will bind to a given candidate peptide sequence of the plurality of candidate peptide sequences, wherein the machine learning HLA peptide binding prediction model is trained using training data, the training data comprising proteins identified to bind to HLA class II proteins or HLA The machine learning HLA peptide binding prediction model comprises: a method for determining a peptide sequence that binds to a class II protein analog; and identifying, based at least on the plurality of binding predictions, a peptide sequence from the plurality of peptide sequences that has a probability of binding to at least one of the one or more proteins encoded by the class II HLA alleles of the subject's cells that is greater than a threshold binding prediction probability value; wherein the machine learning HLA peptide binding prediction model has a positive predictive value (PPV) of at least 0.1 when the amino acid information of the plurality of test peptide sequences is processed to generate a plurality of test binding predictions, each test binding prediction indicating a likelihood that one or more proteins encoded by the class II HLA alleles of the subject's cells bind to a given test peptide sequence from the plurality of test peptide sequences. , wherein the plurality of test peptide sequences comprises at least 50 test peptide sequences, the sequences comprising (i) at least one hit peptide sequence identified by mass spectrometry as being presented by an HLA protein expressed in a cell, and (ii) at least 19 bait peptide sequences contained in a protein, the protein comprising a peptide sequence identified by mass spectrometry as being presented by an HLA protein expressed in a cell, wherein the organism and the subject are of the same species, wherein the ratio of the at least one hit peptide sequence in the plurality of test peptide sequences to the at least 19 bait peptide sequences is 1:19, and according to a machine learning HLA peptide presentation prediction model, 5% of the plurality of test peptide sequences are predicted to bind to an HLA protein expressed in a cell.

[0208] In some embodiments, the machine learning HLA peptide presentation prediction model is trained using training data comprising sequence information of training peptide sequences that are identified by mass spectrometry as being presented by HLA proteins expressed in training cells.

[0209] In some embodiments, the probability that one or more of 0.2% of the plurality of test peptide sequences predicted by the machine learning HLA peptide presentation prediction model to be presented is presented by at least one of the one or more proteins encoded by the class II HLA alleles of the subject's cells is greater than a threshold presentation prediction probability value.

[0210] In some embodiments, each of 0.2% of the plurality of test peptide sequences predicted by the machine learning HLA peptide presentation prediction model to be presented has a probability of being presented by at least one of the one or more proteins encoded by the class II HLA alleles of the subject's cells that is greater than a threshold presentation prediction probability value.

[0211] In some embodiments, for proteins encoded by corresponding HLA alleles of Table 13, the PPV is greater than the corresponding PPV of column 2 of Table 11. In some embodiments, for proteins encoded by corresponding HLA alleles of Table 11, the PPV is at least equal to the corresponding PPV of column 3 of Table 11.

[0212] In some embodiments, for a protein encoded by an HLA class II allele, the PPV is greater than the corresponding PPV in column 2 of Table 12.

[0213] In some embodiments, for the protein encoded by the corresponding HLA allele of Table 16, the PPV is at least equal to the corresponding PPV of column 2 of Table 16.

[0214] Provided herein is a method for preparing a personalized cancer treatment, the method comprising: identifying a peptide sequence, wherein the peptide sequence is associated with cancer, wherein the identifying comprises comparing a DNA, RNA, or protein sequence from a cancer cell of a subject with a DNA, RNA, or protein sequence from a normal cell of the subject; inputting amino acid position information of the identified peptide sequence into a machine learning HLA-peptide presentation prediction model using a computer processor to generate a set of presentation predictions for the identified peptide sequence, each presentation prediction representing a peptide sequence presented by an HLA of a cell of the subject; The method comprises: obtaining a prediction model for HLA-peptide presentation based on a set of prediction variables, wherein the prediction model comprises: a plurality of prediction variables identified at least based on training data; wherein the training data comprises: sequence information of peptide sequences presented by HLA proteins expressed in cells and identified by mass spectrometry; training peptide sequence information comprising amino acid position information, wherein the training peptide sequence information is associated with HLA proteins expressed in cells; and a function representing a relationship between amino acid position information received as input and a presentation probability generated as output based on the amino acid position information and the prediction variables; and selecting a subset of the identified peptide sequences for preparing personalized cancer treatment based on the set of presentation predictions; wherein the prediction model has a positive predictive value of at least 0.1 at a decision rate of at least 0.1%, 0.1%-50%, or at most 50%.

[0215] Provided herein is a method comprising training a machine learning HLA-peptide presentation prediction model, wherein the training comprises inputting into the HLA-peptide presentation prediction model a sequence of amino acid position information of HLA-peptides separated from one or more HLA-peptide complexes from cells expressing HLA class II alleles using a computer processor; the machine learning HLA-peptide presentation prediction model comprises: a plurality of predictor variables identified based at least on training data, the training data comprising: sequence information of sequences of peptides presented by HLA proteins expressed in cells and identified by mass spectrometry; training peptide sequence information comprising amino acid position information of training peptides, wherein the training peptide sequence information is associated with HLA proteins expressed in cells; and a function representing a relationship between amino acid position information received as input and presentation likelihood generated as output based on the amino acid position information and the predictor variables.

[0216] In some embodiments, the presentation model has a positive predictive value of at least 0.25 at a decision rate of at least 0.1%, 0.1%-50%, or at most 50%.

[0217] In some embodiments, the presentation model has a positive predictive value of at least 0.4 at a decision rate of at least 0.1%, 0.1%-50%, or at most 50%.

[0218] In some embodiments, the presentation model has a positive predictive value of at least 0.6 at a decision rate of at least 0.1%, 0.1%-50%, or at most 50%.

[0219] In some embodiments, the mass spectrometry is monoallelic mass spectrometry.

[0220] In some embodiments, the peptide is presented by an HLA protein expressed in the cell via autophagy.

[0221] In some embodiments, the peptide is presented by an HLA protein expressed in a cell via phagocytosis.

[0222] In some embodiments, the quality of the training data is improved by using multiple quality metrics.

[0223] In some embodiments, the plurality of quality metrics include common contaminant peptide removal, high scoring peak intensity, high score, and high mass accuracy.

[0224] In some embodiments, the scored peak intensity is at least 50%.

[0225] In some embodiments, the scored peak intensity is at least 60%.

[0226] In some embodiments, the score is at least 7.

[0227] In some embodiments, the mass accuracy is at most 5 ppm.

[0228] In some embodiments, the mass accuracy is at most 2 ppm.

[0229] In some embodiments, the backbone cleavage score is at least 5.

[0230] In some embodiments, the backbone cleavage score is at least 8.

[0231] In some embodiments, the peptide presented by an HLA protein expressed in a cell is a peptide presented by a single immunoprecipitated HLA protein expressed in the cell.

[0232] In some embodiments, the peptide presented by the HLA protein expressed in the cell is a peptide presented by a single exogenous HLA protein expressed in the cell.

[0233] In some embodiments, the peptide presented by the HLA protein expressed in the cell is a peptide presented by a single recombinant HLA protein expressed in the cell.

[0234] In some embodiments, the plurality of predictor variables includes a peptide-HLA affinity predictor variable.

[0235] In some embodiments, the plurality of predictor variables includes a source protein expression level predictor variable.

[0236] In some embodiments, the plurality of predictor variables includes a peptide cleavage predictor variable.

[0237] In some embodiments, the training peptide sequence information includes sequences of peptides presented by HLA proteins, including peptides identified by searching a non-enzyme-specific non-modified peptide database. In some embodiments, peptides presented by HLA proteins include peptides identified by searching a de novo peptide sequencing tool.

[0238] In some embodiments, peptides presented by HLA proteins include peptides identified by searching a peptide database using an inverse database searching strategy.

[0239] In some embodiments, the HLA protein includes HLA-DR and HLA-DP or HLA-DQ protein. In some embodiments, the HLA protein includes an HLA-DR protein selected from HLA-DR and HLA-DP or HLA-DQ protein. In some embodiments, the HLA protein includes an HLA-DR protein selected from the group consisting of HLA-DPB1*01:01 / HLA-DPA1*01:03, HLA-DPB1*02:01 / HLA-DPA1*01:03, HLA-DPB1*03:01 / HLA-DPA1*01:03, HLA-DPB1*04:01 / HLA-DPA1*01:03, HLA-DPB1*04:02 / HLA-DPA1*01:03, HLA-DPB1*06:01 / HLA-DPA1*01:03, HLA- A-DQB1*02:01 / HLA-DQA1*05:01, HLA-DQB1*02:02 / HLA-DQA1*02:01, HLA-DQB1*06:02 / HLA-DQA1*01:02, HLA-DQB1*06:04 / HLA- DQA1*01:02, HLA-DRB1*01:01, HLA-DRB1*01:02, HLA-DRB1*03:01, HLA-DRB1*03:02, HLA-DRB1*04:01, HLA-DRB1*04:02, HLA-DR B1*04:03, HLA-DRB1*04:04, HLA-DRB1*04:05, HLA-DRB1*04:07, HLA-DRB1*07:01, HLA-DRB1*08:01, HLA-DRB1*08:02, HLA-DRB1 *08:03, HLA-DRB1*08:04, HLA-DRB1*09:01, HLA-DRB1*10:01, HLA-DRB1*11:01, HLA-DRB1*11:02, HLA-DRB1*11:04, HLA-DRB1*1 2:01, HLA-DRB1*12:02, HLA-DRB1*13:01, HLA-DRB1*13:02, HLA-DRB1*13:03, HLA-DRB1*14:01, HLA-DRB1*15:01, HLA-DRB1*15: 02, HLA-DRB1*15:03, HLA-DRB1*16:01, HLA-DRB3*01:01, HLA-DRB3*02:02, HLA-DRB3*03:01, HLA-DRB4*01:01 and HLA-DRB5*01:01.

[0240] In some embodiments, peptides presented by HLA proteins include peptides identified by comparing the MS / MS spectrum of the HLA-peptide to the MS / MS spectra of one or more HLA-peptides in a peptide database.

[0241] In some embodiments, the mutation is selected from a point mutation, a splice site mutation, a frameshift mutation, a read-through mutation, and a gene fusion mutation.

[0242] In some embodiments, the peptide presented by the HLA protein has a length of 15-40 amino acids.

[0243] In some embodiments, peptides presented by HLA proteins include peptides identified by: (a) isolating one or more HLA complexes from a cell line expressing a single HLA class II allele; (b) isolating one or more HLA-peptides from the one or more isolated HLA complexes; (c) obtaining MS / MS spectra of the one or more isolated HLA-peptides; and (d) obtaining peptide sequences corresponding to the MS / MS spectra of the one or more isolated HLA-peptides from a peptide database; wherein the sequences of the one or more isolated HLA-peptides are identified from the one or more sequences obtained in step (d).

[0244] In some embodiments, the personalized cancer therapy further comprises an adjuvant.

[0245] In some embodiments, the personalized cancer therapy further comprises an immune checkpoint inhibitor.

[0246] In some embodiments, the training data includes structured data, time series data, unstructured data, relational data, or any combination thereof.

[0247] In some embodiments, the unstructured data includes image data.

[0248] In some embodiments, the relationship data includes data from a client system, an enterprise system, an operating system, a website, a web-accessible application programming interface (API), or any combination thereof.

[0249] In some embodiments, the training data is uploaded to a cloud-based database.

[0250] In some embodiments, the training is performed using a convolutional neural network.

[0251] In some embodiments, the convolutional neural network comprises at least two convolutional layers.

[0252] In some embodiments, the convolutional neural network (CNN) includes at least one batch normalization step.

[0253] In some embodiments, the convolutional neural network includes at least one spatial dropout step.

[0254] In some embodiments, the convolutional neural network includes at least one global maximum pooling step.

[0255] In some embodiments, the convolutional neural network comprises at least one dense layer.

[0256] In some embodiments, identifying a peptide sequence comprises identifying a peptide sequence having a mutation that is expressed in a cancer cell of the subject.

[0257] In some embodiments, identifying a peptide sequence comprises identifying a peptide sequence that is not expressed in normal cells of the subject.

[0258] In some embodiments, identifying a peptide sequence comprises identifying an overexpressed peptide sequence.

[0259] In some embodiments, identifying a peptide sequence comprises identifying a viral peptide sequence. In one aspect, the present invention provides a method for identifying an HLA class II specific peptide for immunotherapy that is specific to a subject, the method comprising: identifying a candidate peptide comprising an epitope; using a computer processor to input amino acid information of a plurality of peptide sequences (each comprising an epitope) into a machine learning HLA-peptide presentation prediction model to generate a set of HLA presentation predictions about the peptide sequence to immune cells, each presentation prediction representing the probability that one or more proteins encoded by an HLA class II allele of a subject's cell present a given peptide sequence comprising the epitope; wherein the prediction model has a positive predictive value of at least 0.1 at a determination rate of at least 0.1%, 0.1%-50%, or at most 50%, selecting a protein from one or more proteins encoded by an HLA class II allele of a subject's cell, the protein being predicted by the prediction model to bind to the candidate peptide, wherein the probability that the protein presents the candidate peptide to an immune cell is greater than a threshold presentation prediction probability value; contacting the candidate peptide with a protein encoded by an HLA class II allele such that the candidate peptide competes with the HLA a placeholder peptide associated with a protein encoded by a class II allele; and identifying the candidate peptide as a peptide for immunotherapy that is specific to the protein encoded by the class II allele according to whether the candidate peptide replaces the placeholder peptide.

[0260] In some embodiments, the immunotherapy is cancer immunotherapy.

[0261] In some embodiments, identifying comprises comparing a DNA, RNA, or protein sequence from a cancer cell of the subject to a DNA, RNA, or protein sequence from a normal cell of the subject.In some embodiments, the epitope is a cancer-specific epitope.

[0262] In some embodiments, the at least one protein encoded by the HLA class II allele comprises at least an α1 subunit and a β1 subunit of an HLA protein or a fragment thereof in the form of a dimer. In some embodiments, the placeholder peptide is a CLIP peptide. In some embodiments, the placeholder peptide is a CMV peptide. In some embodiments, the method further comprises measuring the IC50 of the placeholder peptide being replaced by the target peptide. In some embodiments, the IC50 of the placeholder peptide being replaced by the target peptide is less than 500nM. In some embodiments, at least one protein from one or more proteins encoded by the HLA class II allele of the subject's cell is an HLA class II tetramer or multimer. In some embodiments, the target peptide is further identified by mass spectrometry. In some embodiments, the at least one protein encoded by the HLA class II allele of the subject's cell is a recombinant protein. In some embodiments, the at least one protein encoded by the HLA class II allele of the subject's cell is expressed in a eukaryotic cell.

[0263] In one aspect, provided herein is an assay method for validating the specificity of a candidate peptide binding to an HLA class II protein, the method comprising: expressing in a eukaryotic cell a polynucleic acid construct comprising a nucleic acid sequence encoding an HLA class II protein, the protein comprising an α chain and a β chain or a portion thereof, capable of binding to a peptide comprising an MHC-II binding epitope, and wherein the expressed HLA class II protein or the portion thereof remains associated with a placeholder peptide; isolating the HLA class II protein or the portion thereof expressed in the eukaryotic cell; performing a peptide exchange assay by: (a) adding increasing amounts of a candidate peptide to determine whether the candidate peptide replaces the placeholder peptide associated with the HLA class II protein or the portion thereof; and (b) calculating the IC50 of the substitution reaction to determine the affinity of the candidate peptide for the HLA class II protein or the portion thereof relative to the placeholder peptide, thereby validating the specificity of the candidate peptide binding to the HLA class II protein.

[0264] In some embodiments, the identity of the peptide is known. In some embodiments, the identity of the peptide is unknown. In some embodiments, the identity of the peptide is determined by mass spectrometry.

[0265] In some embodiments, the peptide exchange assay comprises detecting a peptide fluorescent probe or tag. In some embodiments, the placeholder peptide is a CLIP peptide.

[0266] In some embodiments, the polynucleic acid construct comprises an expression vector, which further comprises one or more of the following: a promoter, a linker, one or more protease cleavage sites, a secretion signal, a dimerization factor, a ribosomal skipping sequence, and one or more tags for purification and / or detection.

[0267] In one aspect, the present invention provides a method for determining the immunogenicity of an MHC class II binding peptide, the method comprising: selecting a protein encoded by an HLA class II allele that is predicted to bind to a peptide by a machine learning HLA-peptide presentation prediction model; wherein the prediction model has a positive predictive value of at least 0.1 at a call rate of at least 0.1%, 0.1%-50%, or at most 50%, and wherein the probability of the protein presenting the identified peptide sequence is greater than a threshold presentation prediction probability value; contacting the peptide with the selected protein encoded by the HLA class II allele such that the peptide competes with and replaces the placeholder peptide associated with the selected protein encoded by the HLA class II allele, thereby forming a complex comprising the HLA class II protein and the identified peptide; contacting the complex of the HLA class II protein and the identified peptide with CD4+ T cells, and determining one or more activation parameters of the CD4+ T cells, the parameters being selected from the induction of cytokines, the induction of chemokines, and the expression of cell surface markers.

[0268] In some embodiments, the HLA class II allele is a tetramer or a multimer. In some embodiments, the cytokine is IL-2. In some embodiments, the cytokine is IFN-γ.

[0269] In one aspect, the present invention provides a method for inducing CD4+ T cell activation in a subject for cancer immunotherapy, the method comprising: identifying a peptide sequence associated with cancer and comprising a cancer mutation, wherein the identification comprises comparing a DNA, RNA or protein sequence from a cancer cell of the subject with a DNA, RNA or protein sequence from a normal cell of the subject; selecting a protein encoded by an HLA class II allele, the protein being normally expressed by the cells of the subject and being predicted to bind to the peptide by a machine learning HLA-peptide presentation prediction model; wherein the prediction model has a positive predictive value of at least 0.1 at a determination rate of at least 0.1%, 0.1%-50%, or at most 50%, and wherein the probability of the protein presenting the identified peptide sequence is greater than a threshold presentation prediction probability value; contacting the identified peptide with the selected protein encoded by the HLA class II allele to verify whether the identified peptide competes with the selected protein encoded by the HLA class II allele. The invention relates to a method for identifying a placeholder peptide associated with a selected protein encoded by a class II allele, thereby replacing the placeholder peptide with an IC50 value of less than 500 nM; purifying the identified peptide; and administering an effective amount of the identified peptide to the subject.

[0270] In one aspect, the present invention provides a method for preparing HLA class II tetramers or multimers, the method comprising: expressing in eukaryotic cells a vector comprising a nucleic acid sequence encoding an α chain and a β chain of an HLA protein, a linker, one or more protease cleavage sites, a secretion signal, a biotinylation motif, and at least one tag for identification or purification, so that each HLA protein α1 and β1 heterodimer is secreted in a dimerized state, wherein the heterodimer is associated with a placeholder peptide, purifying the secreted heterodimers from the cell culture medium, verifying the peptide binding activity using a peptide exchange assay, adding streptavidin to conjugate the heterodimers into tetramers, purifying the tetramers, and having a yield greater than 1 mg / L.

[0271] In some embodiments, the vector comprises a CMV promoter. In some embodiments, the vector comprises a sequence encoding a placeholder peptide connected to the β1 chain by a cleavable site. In some embodiments, the peptide exchange assay involves pre-cutting the placeholder peptide from the β chain. In some embodiments, the cleavable site is a thrombin cleavage site. In some embodiments, the peptide exchange assay is a FRET assay. In some embodiments, the purification is performed by any of the following: column chromatography, batch chromatography, ion exchange chromatography, size exclusion chromatography, affinity chromatography, or LC-MS.

[0272] In one aspect, provided herein is a composition comprising an HLA class II tetramer comprising an HLA-DR or HLA-DP or HLA-DQ heterodimer, each heterodimer comprising an α and β chain, which is purified and present at a concentration greater than 0.25 mg / L. In some embodiments, the HLA class II tetramer comprises a heterodimer pair selected from the group consisting of: a protein may be selected from HLA-DR and HLA-DP or HLA-DQ proteins.In some embodiments, the HLA protein is selected from: HLA-DPB1*01:01 / HLA-DPA1*01:03, HLA-DPB1*02:01 / HLA-DPA1*01:03, HLA-DPB1*03:01 / HLA-DPA1*01:03, HLA-DPB1*04:01 / HLA-DPA1*01:03, HLA-DPB1*04:02 / HLA-DPA1*01:03, HLA-DPB1*06:01 / HLA-DPA1*01:03, HLA-DQB1*02:01 / HLA-DQA1*05:01, HLA-DQB1*02:02 / HLA-DQA1*02:01, HLA-DQB1*06:02 / HLA-DQA1*01:02, HLA-DQB1*06:04 / HLA-DQA1*01:02, HLA-DRB1*01:01, HLA-DRB1*01:02, HLA-DRB1*03:01, HLA-DRB1*03:02, HLA-DRB1*04:01, HLA-DRB1*04:02, HLA-DRB1*04:03, HLA-DRB1*04:04, HLA-DRB1*04:05, HLA-DRB1*04:07, HLA-DRB1*07:01, HLA-DRB1*08:01, HLA-DRB1*08:02, HLA-DRB1*08:03, HLA-DRB1*08:04, HLA-DRB1*09:01, HLA-DRB1*10:01, HLA-DRB1*11:01, HLA-DRB1*11:02, HLA-DRB1*11:04, HLA-DRB1*12:01, HLA-DRB1*12:02, HLA-DRB1*13:01, HLA-DRB1*13:02, HLA-DRB1*13:03, HLA-DRB1*14:01, HLA-DRB1*15:01, HLA-DRB1*15:02, HLA-DRB1*15:03, HLA-DRB1*16:01, HLA-DRB3*01:01, HLA-DRB3*02:02, HLA-DRB3*03:01, HLA-DRB4*01:01, HLA-DRB5*01:01).

[0273] In some embodiments, the heterodimer pair is expressed in a eukaryotic cell. In some embodiments, the heterodimer pair is encoded by a vector. In some embodiments, the vector comprises: a nucleic acid sequence encoding the α chain and β chain of the HLA protein, a secretion signal, a biotinylation motif, and at least one tag for identification or purification, so that each HLA protein α1 and β1 heterodimer is secreted in a dimerized state, wherein the secreted heterodimer is associated with a placeholder peptide. In some embodiments, the vector comprises: a nucleic acid sequence encoding the α chain and β chain of the HLA protein, a secretion signal, a biotinylation motif, and at least one tag for identification or purification, so that each HLA protein α1 and β1 heterodimer is secreted in a dimerized state, wherein the secreted heterodimer is associated with a placeholder peptide.

[0274] In some embodiments, the HLA class II heterodimer is secreted from eukaryotic cells into the cell culture medium and purified by any of the following methods: column or batch chromatography, ion exchange chromatography, size exclusion chromatography, affinity chromatography, or LC-MS.

[0275] In one aspect, the present invention provides a method for screening a drug comprising a polypeptide sequence for immunogenicity in a subject, the method comprising: using a computer processor to input amino acid information of a peptide sequence of the polypeptide sequence into a machine learning HLA-peptide presentation prediction model to generate a set of presentation predictions for the peptide sequence, each presentation prediction representing the probability that an epitope sequence of a given peptide sequence is presented by one or more proteins encoded by an HLA class I or class II allele of a cell of the subject; wherein the machine learning HLA-peptide presentation prediction model comprises: a plurality of predictor variables identified based at least on training data; wherein the training data comprises: sequence information of sequences of peptides presented by HLA proteins expressed in cells and identified by mass spectrometry; training peptide sequence information comprising amino acid position information, wherein the training peptide sequence information is associated with HLA proteins expressed in cells; and a function representing a relationship between the amino acid position information received as input and the presentation likelihood generated as output based on the amino acid position information and the predictor variables; (b) determining or predicting, based on the set of presentation predictions, that each of the peptide sequences of the polypeptide sequence is not immunogenic to the subject; and (c) administering a composition comprising the drug to the subject.

[0276] In one aspect, the present invention provides a method for screening a drug comprising a polypeptide sequence for immunogenicity in a subject, the method comprising: (a) using a computer processor to input amino acid information of a peptide sequence of the polypeptide sequence into a machine learning HLA-peptide presentation prediction model to generate a set of presentation predictions for the peptide sequence, each presentation prediction representing the probability that an epitope sequence of a given peptide sequence is presented by one or more proteins encoded by an HLA class I or class II allele of a cell of the subject; wherein the machine learning HLA-peptide presentation prediction model comprises: a plurality of predictor variables identified at least based on training data; wherein the training data comprises: sequence information of sequences of peptides presented by HLA proteins expressed in cells and identified by mass spectrometry; training peptide sequence information comprising amino acid position information, wherein the training peptide sequence information is associated with HLA proteins expressed in cells; and a function representing a relationship between the amino acid position information received as input and the presentation likelihood generated as output based on the amino acid position information and the predictor variables; (b) based on the set of presentation predictions, determining or predicting that at least one of the peptide sequences of the polypeptide sequence is immunogenic to the subject.

[0277] In one embodiment, the method further comprises deciding not to administer the drug to the subject.

[0278] In one embodiment, the medicament comprises an antibody or a binding fragment thereof.

[0279] In one embodiment, the peptide sequence of the polypeptide sequence comprises each contiguous peptide sequence of the polypeptide sequence, which has a length of 8, 9, 10, 11 or 12 amino acids, and wherein the protein encoded by an HLA class I or class II allele of a cell of the subject is a protein encoded by an HLA class I allele of a cell of the subject.

[0280] In one embodiment, the peptide sequence of the polypeptide sequence comprises each contiguous peptide sequence of the polypeptide sequence, which has a length of 15, 16, 17, 18, 19, 20, 21, 22, 23, 24 or 25 amino acids, and wherein the protein encoded by an HLA class I or class II allele of a subject's cell is a protein encoded by an MHC class II allele of a subject's cell.

[0281] In one aspect, provided herein is a method of treating a subject having an autoimmune disease or condition, comprising: (a) identifying or predicting an epitope of an expressed protein presented by HLA class I or class II of cells of the subject, wherein a complex comprising the identified or predicted epitope and HLA class I or class II is targeted by CD8 or CD4 T cells of the subject; (b) identifying a T cell receptor (TCR) that binds to the complex; (c) expressing the TCR in regulatory T cells or allogeneic regulatory T cells from the subject; and (d) administering regulatory T cells expressing the TCR to the subject.

[0282] In one embodiment, the autoimmune disease or condition is diabetes.

[0283] In one embodiment, the cells are pancreatic islet cells.

[0284] In one aspect, provided herein is a method of treating a subject having an autoimmune disease or condition, comprising administering to the subject a regulatory T cell expressing a T cell receptor (TCR) that binds to a complex comprising: (i) an epitope of an expressed protein identified or predicted to be presented by HLA class I or class II of a cell of the subject, and (ii) HLA class I or class II, wherein the complex is targeted by a CD8 or CD4 T cell of the subject.

[0285] Other aspects and advantages of the present disclosure will become apparent to those skilled in the art based on the following detailed description, which merely shows and describes illustrative embodiments of the present disclosure. It should be appreciated that the present disclosure is capable of other different embodiments, and that its several details are capable of modification in various obvious respects, all without departing from the present disclosure. Therefore, the drawings and description are to be regarded as illustrative in nature, and not restrictive.

[0286] MAPTAC TM Can be used for high-throughput peptide binding assays where MAPTAC is used at different time points and under different conditions (e.g., heating at 37°C) TM After separation of the constructs, peptides binding to HLA class II were measured to obtain the sequences of peptide populations with different stabilities using LC-MS / MS.

[0287] In one aspect, the present invention provides a method for treating cancer in a subject, the method comprising: identifying a peptide sequence, wherein the peptide sequence is associated with cancer, wherein the identification comprises comparing a DNA, RNA or protein sequence from a cancer cell of the subject with a DNA, RNA or protein sequence from a normal cell of the subject; using a computer processor, inputting amino acid information of the identified peptide sequence into a machine learning HLA-peptide presentation prediction model to generate a set of presentation predictions for the identified peptide sequence, each presentation prediction representing a peptide sequence presented by an HLA of a cell of the subject; The method comprises the steps of: determining a probability that one or more proteins encoded by a class II allele present a given sequence of the identified peptide sequence; wherein the machine learning HLA-peptide presentation prediction model comprises: a plurality of predictor variables identified based at least on training data, wherein the training data comprises: sequence information of sequences of peptides presented by HLA proteins expressed in cells and identified by mass spectrometry; training peptide sequence information comprising amino acid position information, wherein the training peptide sequence information is associated with HLA proteins expressed in cells; and a function representing a relationship between amino acid position information received as input and a presentation probability generated as output based on the amino acid position information and the predictor variables; and selecting a subset of the identified peptide sequences for preparing personalized cancer treatment based on the set of presentation predictions; and administering a composition comprising one or more of the peptides to the subject, wherein the prediction model has a positive predictive value of at least 0.1 at a decision rate of at least 0.1%, 0.1%-50%, or at most 50%.

[0288] In some embodiments, the machine learning HLA-peptide presentation prediction model comprises sequence information of the sequences of peptides presented by HLA proteins expressed in cells and identified by mass spectrometry following reverse phase offline fractionation.

[0289] In some embodiments, the prediction model exhibits an improvement of 1.1x to 100x compared to NetMHC II pan. In some embodiments, the prediction model exhibits an improvement of 1.1, 2, 3, 4, 5, 6, 7, 7.4, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 18, 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 41, 42, 43, 44, 45, 50, 55, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 73, 74, 75, 76, 77, 78, 79, 80, 81, 8, 83, 84, 85, 86, 87, 88, 89, 90, 91, 92, 93, 94, 95, 96, 97, 98, 99, 100-fold or more improvement.

[0290] Incorporation by reference

[0291] All publications, patents, and patent applications mentioned in this specification are incorporated herein by reference to the same extent as if each individual publication, patent, or patent application was specifically and individually indicated to be incorporated by reference. If the publications and patents or patent applications incorporated by reference conflict with the disclosure contained in this specification, this specification is intended to supersede and / or take precedence over any such conflicting material. BRIEF DESCRIPTION OF THE DRAWINGS

[0292] The novel features of the present invention are particularly set forth in the appended claims. The features and advantages of the present invention will be better understood by referring to the following detailed description of illustrative embodiments utilizing the principles of the present invention and the accompanying drawings (also referred to herein as "figures"), in which:

[0293] Figure 1A is a diagram showing a peptide docked on an MHC class I protein. The figure discloses SEQ ID NO:36.

[0294] Figure 1B An exemplary diagram representing a peptide docked on an MHC class II protein is depicted. SEQ ID NO:37 is disclosed in the Figure.

[0295] Figure 2 An exemplary experimental method for generating monoallelic HLA class II binding peptide data is depicted. HLA class II peptides are introduced into any cell, including cells that do not express HLA class II, so as to express a specific HLA class II allele in the cell. Genetically engineered cell populations expressing HLAs are harvested, lysed, their HLA-peptide complexes are labeled (e.g., biotinylated), and immunopurified (e.g., using biotin-streptavidin interactions). HLA-associated peptides specific for a single HLA can be eluted from their labeled (e.g., biotinylated) complexes and evaluated (e.g., sequencing using high-resolution LC-MS / MS).

[0296] Figure 3 Depicted are exemplary sequence logo representations of HLA class II-DRB1*11:01 associated peptides in: NeonBAP, Expi293 cell line; Neon BAP, A375 cell line; IEDB, affinity <50 nM; and full HLA class II antibody, homozygous LCL. Figure 3 Examples of MS-derived motifs are shown to match known patterns and show consistency between transfected cell lines.

[0297] Figure 4is an exemplary depiction of the performance of the HLA class II binding predictor. Figure 4 is a bar graph showing the performance of the binding predictor (neonmhc2) and NetMHC IIpan applied to a validation dataset consisting of observed mass spectrometry peptides and decoy peptides generated by random shuffling of hit peptides at a ratio of 1:19 (hit:decoy). For the NEON binding predictor neonmhc2, a separate model was built for each MHC II allele shown. The height of the bar shows the positive predictive value (PPV), which is defined as the fraction of predicted binders in the validation set that are actually hit peptides. Alleles are ranked by the performance of the model in predicting that allele.

[0298] Figure 5 Depicted are exemplary effects of the scoring peak intensity (SPI) threshold on binding predictor validation. Figure 5 The performance of the HLA class II binding predictor when trained / validated on sets of peptides with different scoring peak intensity (SPI) cutoffs is shown. For each allele-specific model trained, the performance of the model in 3 settings is shown: training and evaluating the dataset using observed MS hit peptides with greater than or equal to 70 SPI, training on peptides with greater than or equal to 50 SPI and validating on peptides with greater than or equal to 70 SPI, and training and validating on peptides with greater than or equal to 50 SPI.

[0299] Figure 6 Depicted is an exemplary bar graph showing representative data from the number of observed peptides using a score peak intensity (SPI) cutoff of greater than or equal to 70 for allele profiling by LC-MS / MS. Each bar represents the total number of peptides observed for the allele. Data for 35 HLA-DR alleles were collected. The data for the 35 HLA-DR alleles collected had a HLA-DR population coverage of >95% (U.S. allele frequency).

[0300] Fig. 7A The PPV of the model when applied to the test partition of the data is shown for the indicated HLA class II alleles. The decoy peptides used were scrambled sequences of the positive (hit) peptide sequences, with a hit to decoy ratio of 1:19. The PPV was determined by identifying the top 5% scoring peptides in the test partition and determining the fraction of them that were positive for binding to the protein encoded by the corresponding HLA class II allele.

[0301] Figures 7B-7D Depicted are exemplary prediction performance as a function of training set size (curve obtained by artificially downsampling the training set). Figures 7B-7Dshowed that, in general, for the 35 HLA-DR alleles collected, the value of PPV increased when the training set size increased.

[0302] Figure 8 An exemplary illustration is depicted demonstrating that treatment-related variables can further improve predictions. Peptides observed by MS can be distinguished from random sequences selected from the protein-coding exome. On the training data partition, logistic regression may be fit to predict HLA class II presentation using binding strength (NetMHCIIpan or Neon predictor) and treatment features (RNA-Seq expression and derived gene-level bias terms). In a separate evaluation partition, exon positions that overlap with MHC II peptides observed by MS ("hits") can be scored together with random exon positions not observed in MS (1:499 ratio). The top 0.2% (1 / 500) can be called positive, and the positive predictive value can assess this threshold.

[0303] Fig. 9 An exemplary neural network architecture is depicted. Input peptides are represented as 20-mers, with shorter peptides padded with "missing" characters. Each peptide has a 31-dimensional embedding, so the input to the neural network is a 20x31 matrix. The 20x31 matrix is ​​feature normalized according to the mean and standard deviation of the feature values ​​in the training set before being processed by the neural network. The first convolutional layer has a kernel of 9 amino acids and 50 filters (also called channels), with a rectified linear unit (ReLU) activation function. This is followed by batch normalization, then spatial dropout, with a dropout rate of 20%. This is followed by another convolutional layer with a kernel of 3 amino acids and 20 filters, with a ReLU activation function, and then batch normalization and spatial dropout again, with a dropout rate of 20%. Global max pooling is then applied, taking the neuron with the maximum activation in each of the 20 filters; these 20 values ​​are then fed into a fully connected (dense) layer with a single neuron using a sigmoid activation function. The output of this layer is considered the bind / non-bind prediction. L2 regularization was applied to the weights of the first convolutional layer, the second convolutional layer, and the dense layer with weights of 0.05, 0.1, and 0.01, respectively. The other models used varied the number of convolutional layers and the kernel size of each layer.

[0304] Fig.10 An exemplary computer control system programmed or otherwise configured to implement the methods provided herein is depicted.

[0305] Fig.11A Depicting MAPTAC TM Exemplary overview of the experimental workflow. The Figure discloses SEQ ID NO:38.

[0306] Fig. 11BExemplary per-allele peptide counts pooled across replicates are depicted.

[0307] Fig. 11C Depicted by MAPTAC TM Exemplary peptide length distributions for the analyzed HLA class I and HLA class II alleles.

[0308] Fig.11D Describes the MAPTAC TM and IEDB (alleles DRB1*01:01, DRB1*03:01, DRB1*09:01, and DRB1*11:01), the human proteome, and exemplary per-residue cysteine ​​frequencies observed with multi-allelic MS data from previous publications.

[0309] Fig. 12A Depicted are the Caucasian frequencies of HLA-DR, -DP, and -DQ alleles present in >1% of individuals, as well as counts of peptides from the indicated sources measured as strong binders (<50 nM).

[0310] Fig. 12B An exemplary length distribution of IEDB peptides with associated HLA class II affinity measurements is depicted.

[0311] Fig. 12C Depicted are exemplary Western blots of (1) Expi293, (2) HeLa, and (3) A375 cell lines transfected with two HLA class I and two HLA class II alleles, HLA-A*02:01, HLA-B*45:01, HLA-DRB1*01:01, and HLA-DRB1*11:01, respectively. The membrane was blotted with an anti-biotin ligase epitope tag to display the biotin accepting peptide (BAP) and anti-β-tubulin as a loading control. Lanes correspond to the MAPTAC TM The following fractions were collected during the protocol: lane 1, input; lane 2, biotinylated input; lane 3, input after pull-down.

[0312] Fig.12D Describes the MAPTAC TM and IEDB (alleles DRB1*01:01, DRB1*03:01, DRB1*09:01 and DRB1*11:01), the human proteome and multi-allelic MS data from previous publications.

[0313] Fig.12EDepicted are the Caucasian frequencies of HLA-DR, -DP, and -DQ alleles present in >1% of individuals, and counts of peptides from the indicated sources that were measured as strong binders (<50 nM). The graph includes Fig. 12A Additional data from: tools.iedb.org / main / datasets / .

[0314] Fig.12F Describes the MAPTAC TM (reduced and alkylated), MAPTAC TM Exemplary amino acid frequencies per residue observed for (no treatment) and IEDB (alleles DRB1*01:01, DRB1*03:01, DRB1*09:01, and DRB1*11:01), the human proteome, and multi-allelic MS data from a previous publication.

[0315] Fig.13 Depicts each MAPTAC TM Figure 2. Example representation of core binding sequence logos for MHC II alleles from the IEDB. The sequence logo is a graphical representation where the height of each amino acid is proportional to its frequency of occurrence in peptides bound to the MHC protein encoded by the allele. Positions with the lowest entropy are indicated by color, where the color corresponds to the amino acid property. Peptides are from the datasets shown and aligned according to a CNN-based predictor (Methods). The logo represents all peptides, including those that do not closely match the overall motif (e.g., no peptides are isolated in a "junk" cluster).

[0316] Fig.14A Depicted are exemplary sequence logos of HLA-A*02:01 binding peptides (ligands) analyzed in two different cell lines (A375 and expi293) using different HLA-ligand profiling techniques including binding assays, stability assays, soluble HLA (sHLA) mass spectrometry, monoallelic mass spectrometry, and MAPTAC TM .

[0317] Fig. 14B Depicts MAPTACs showing 0, 1, 2, 3, and 4 heuristically defined anchors TM Exemplary fractions of peptides.

[0318] Fig. 14C Describes the MAPTAC TM Exemplary distribution of observed peptides (20 peptides per allele, each nested group with SPI>70 and size>=2) and NetMHCIIpan predicted binding affinities for length-matched baits sampled from the proteome.

[0319] Fig.15A Depicted is an exemplary architecture of a convolutional neural network (CNN) trained to discriminate monoallelic MHC peptides from scrambled length-matched decoys. The schematic shows the use of amino acid feature embeddings, 2 convolutional layers with different filter sizes, and the use of global max pooling as input to the final logical output node.

[0320] Fig. 15B is an exemplary result showing the Kendall Tau statistic for the correlation of measured IEDB affinity with binding predictions from neonmhc2 or NetMHCIIpan. Peptides evaluated included only those published to IEDB one year after NetMHCIIpan release.

[0321] Fig.16 is an exemplary depiction of the performance of neonmhc2 as a function of the training dataset size.

[0322] Fig.17A Delineation of MAPTACs incorporated into pan-DR and pan-class II MHC MS datasets TM Example cluster assignments of peptides (20 per allele). The dataset was deconvoluted using GibbsCluster. Each box represents a MAPTAC TM Peptides. The color of the box indicates which cluster it is assigned to, and the gray bar indicates which allele the peptide actually comes from. The total number of clusters in the Gibbs clustering scheme (right) is selected using the mutual information (MI) metric. The MI score also determines how the samples are ranked; samples with high MI schemes appear at the top.

[0323] Fig. 17B Depicted are exemplary core binding sequence identities for multi-allelic MS data deconvoluted by GibbsCluster. Each set of peptides corresponds to a MAPTAC TM The cluster with the best match for the admixture.

[0324] Fig. 17C Describes the use of MAPTAC TM Data or deconvoluted multi-allelic data predict retention of MAPTAC TM For each allele, the larger of the two data sources (usually MAPTAC TM ) are downsampled so that the predictors are based on the same number of training examples. NetMHCIIpan performance is shown as an additional comparison.

[0325] Fig.17D Depicted are exemplary core binding sequence signatures derived from multi-allelic MS data from the indicated sources.

[0326] Fig.18A Depicted are exemplary graphs of peptide versus source gene expression fractions (transcripts per million (TPM)) for MS observed peptides and random proteome decoys (data replotted from Schuster et al., 2017).

[0327] Fig.18B Depicted are exemplary observed and expected numbers of class II peptides per gene as determined by joint analysis of colorectal cancer, melanoma, and ovarian cancer datasets ( et al., 2018 and Schuster et al., 2017). Expected counts were derived by multiplying the gene length by the expression level. Expected and observed counts were summed for the relevant samples. Genes known to be present in plasma are labeled according to their concentration (inset).

[0328] Fig.18C Depicted is an exemplary distribution of enrichment scores (ratio of observed to expected observations, as Fig.18B ).

[0329] Fig.18D Depicted is an exemplary distribution of enrichment scores according to the localization of each source gene. Source gene localization was determined using Uniprot (uniprot_sprot.dat).

[0330] Fig.18E Depicted are exemplary data representing the comparison of expected and observed frequencies of the fraction of the total number of peptides with MHC-II binding affinity separated based on their cellular localization properties.

[0331] Fig.18F Depicts the exemplary representative data of the relative consistency of the observed peptides for two different gene expression profiles. For each sample, the peptide counts at the gene level are modeled as a linear combination of bulk tumor gene expression and professional APC (macrophage) gene expression profiles. The ratio of the coefficients determines the relative consistency of each expression profile with the peptide repertoire. The error bars correspond to the 95% confidence interval calculated by bootstrap resampling.

[0332] Fig.19A Depicted are exemplary representative data of HLA-DRB1 expression levels in five example studies. Each spot represents expression averaged across cells, for an individual cell type in an individual patient.

[0333] Fig.19BDepicted are exemplary representative data of HLA-DRB1 expression derived from tumors and stroma of RNA-Seq input from TCGA patients. Horizontal bars correspond to individual patients and are grouped by tumor type. If the patient has a mutation in an HLA class II pathway gene (CIITA, CD74, or CTSSS) as determined by DNA-based mutation determination, these patients are included. For each patient, the score of HLA-DRB1 expression attributed to the tumor is estimated as min(1,2f), where f is the score of RNA-Seq reads in CIITA, CD74, or CTSS showing mutations.

[0334] Fig.19C Depicted are exemplary representative data from additional single-cell RNA-Seq studies including biopsies before and after checkpoint blockade immunotherapy.

[0335] Fig. 20 Depicted are exemplary representative experimental data evaluating the overall performance of predictions on native donor tissue.

[0336] Fig.21A Depicted are exemplary representative data showing that the integrated presentation model predicts cellular HLA class II ligandomes. It represents the PPV of the pan-DR dataset at a hit: decoy ratio of 1:499 (also in Fig. 30B and Fig.32E The predictor uses a combination of predictions (NetMHCIIpan or neonmhc2) and optionally gene expression, gene preference (based on Fig.32A ) and overlapped with previously observed HLA-DQ peptides. For each candidate peptide, the binding score was calculated as the maximum value among the HLA-DR alleles in the sample genotype.

[0337] Fig.21B Depicted are exemplary representative data showing the predicted performance of tumor-derived peptides identified using SILAC that were presented by dendritic cells (analyzed from cell lysates) using Fig.21A The same hit:decoy ratios and performance metrics with and without treatment features.

[0338] Fig. 21C Depicted are exemplary expression and gene preference scores observed for heavily labeled peptides (red spots, plotted against K562 expression) in UV treatment experiments compared to lightly labeled peptides (grey spots, plotted against DC expression).

[0339] Fig.21D Depicted is an example illustration representing the overlap of heavily labeled peptide-derived genes from lysate and UV treatment experiments. Gene names are colored by functional category.

[0340] Fig.22A Depicted is an exemplary flow chart representing the assay protocol disclosed herein to validate HLA class II driven CD4+ T cells and T cell responses.

[0341] Fig. 22B An exemplary HLA protein dimer construct design for peptide exchange assay (upper panel) and a schematic diagram of an exemplary assay workflow (lower panel) are depicted. The figure discloses "10XHis" as SEQ ID NO:20.

[0342] Fig.23 Depicted is an exemplary representation of an exemplary vector design for MHC-II expression for screening of novel binding peptides, and representation of the expressed protein product. The figure discloses SEQ ID NO:39 and discloses "10XHis" as SEQ ID NO:20.

[0343] Fig.24 An exemplary flow chart for transfection, purification, and cleavage of the placeholder peptide from the beta chain is depicted.

[0344] Fig.25A Depicted is an exemplary diagram showing a vector encoding a CLIP peptide associated with increased secretion of expressed MHC-II peptides. SEQ ID NO:21 is disclosed in the Figure.

[0345] Fig.25B Depicted are exemplary schematic representations of shorter and longer forms of nucleic acids encoding CLIPO and CLIP1, respectively. The Figures disclose SEQ ID NOs 1 and 21, respectively, in order of appearance.

[0346] Fig.25C Depicted are exemplary representative results of Coomassie gel analysis of α and β chains with and without a longer clip.

[0347] Fig.26A An exemplary diagram of a TR-FRET assay is depicted.

[0348] Fig.26B Depicted are exemplary representative polarization data for an HLA class II peptide binding assay using a fluorescence resonance energy transfer (FRET) assay using specific peptides.

[0349] Fig.26C Depicted are exemplary representative polarization data for an HLA class II peptide binding assay using a fluorescence resonance energy transfer (FRET) assay using specific peptides.

[0350] Fig.26D Depicted are exemplary percentage substitutions of MHC-construct binding peptides calculated from fluorescence increases.

[0351] Fig.26E Depicted are exemplary percentage substitutions of MHC-construct binding peptides calculated from fluorescence increases.

[0352] Fig.26F Depicted is an exemplary peptide exchange obtained using an assay employing differential scanning fluorimetry (DSF). The diagram shows an exemplary mechanism for detecting the dissociation of peptides from MHC class II using heating, which also dissociates MHC class II heterodimers, resulting in binding and high fluorescence of the fluorophore. Also depicted is an exemplary schematic diagram of a placeholder peptide displaced by an epitope peptide. Also depicted is an exemplary melting curve plotted against temperature.

[0353] Figure 26G Depicted are exemplary soluble HLA-DM constructs and their use for MHC class II peptide exchange. The depicted constructs comprise a CMV promoter, a coding sequence for the HLA-DM β chain, and a coding sequence for the HLA-DM α chain downstream of the secretory sequence (leader), and a BAP sequence at the 3' end of the β chain coding sequence; a His tag at the 3' end of the α chain coding sequence. The two chains are separated by an intermediate ribosomal skipping sequence. The construct is expressed in Expi-CHO cells, and the protein secreted into the culture medium is purified. The figure discloses "10XHis" as SEQ ID NO: 20.

[0354] Fig.26H Exemplary size exclusion chromatography data for peptide exchange using HLA-sDM are shown.

[0355] Fig.27A An exemplary diagram of an exemplary DRB tetramer repertoire construction is depicted.

[0356] Fig.27B An exemplary diagram of an exemplary class II tetramer repertoire construction is depicted.

[0357] Fig.27C An exemplary diagram depicting a summary of DRB tetramer repertoire coverage of DRB1 alleles for peptide exchange is depicted.

[0358] Fig.27D Depicted are exemplary coverage of human MHC class II allele generation.

[0359] Fig.27E Exemplary results from tetramer staining of samples induced with either influenza epitopes (anamnestic response) or HIV epitopes (primary response) are shown.

[0360] Fig.28A Depicted is an exemplary representation of a method for assessing HLA class II restriction of peptides by a fluorescence polarization assay that enables a screening method to rapidly identify the allelic restriction of epitope peptides. Fig.28A The assay principle depicted in allows for affinity measurements and unambiguous measurement of peptide exchange.

[0361] Fig.28B An exemplary summary of multiple assay conditions explored in a fluorescence polarization assay using DRB1*01:01 is depicted (top). Also depicted are schematic representations of soluble MHC class II alleles and full-length MHC class II alleles with transmembrane domains in detergent micelles (bottom), both constructed with placeholder peptides with cleavable linkers for use in the assay.

[0362] Fig.28C Describes the previous study Fig.28B The full-length and soluble alleles are shown in the figure below as an exemplary illustration of the assay. In short, both the full-length and soluble alleles are expressed in cells. The membrane-bound full-length allele form is collected by permeabilizing the membrane, while the secreted form is collected from the cell supernatant. The harvested class II HLA allele proteins are then purified by passing through nickel (Ni 2+ ) column for purification.

[0363] Fig.28D Depicted are exemplary data showing that the purification method does not affect the potency of the peptide. Shown on the left are full-length HLA-DR1 and Ni from purification using L243. 2+ Average IC50 values ​​from experiments with purified full-length HLA-DR1.

[0364] Fig.28E Depicted are exemplary data showing that the choice of soluble form (sDR1) or full-length form (fDR1) does not affect peptide potency. Shown on the left are mean IC50 values ​​from experiments using the sDR1 form or fDR1. FP, fluorescence polarization.

[0365] Fig.28F Depicted are exemplary graphical views of exemplary evaluation of neoNMC2- and NetMHCIIpan-predicted peptides in binding assays and identification of discordant peptides.

[0366] Figure 28G Depicted are exemplary fluorescence polarization binding screening data for evaluating neonmhc2 predicted peptides; displayed as a heat map along with the percent inhibition of probe binding indicated for each concentration of peptide used. Green depicts good binding proportional to the color intensity. Yellow depicts moderate binding and red depicts poor binding, also indicated by the corresponding percent inhibition value.

[0367] Fig.28H Depicted is a summary of the evaluation of NeoNMC2-predicted peptides in an exemplary binding assay.

[0368] Fig.29Depicted are the average MAPTAC TM Exemplary average peptide counts from experimental replicates (50 million cells).

[0369] Figures 30A-30C Delineation of HLA class II MAPTAC TM Exemplary binding core analysis of allele + / - HLA-DM and multi-allelic deconvolution fidelity. Fig. 30A Depicts the MAPTAC TM , Example sequence logos for a representative HLA-DR, -DQ, and -DP allele with and without HLA-DM co-transfection (expi293 cell line) and IEDB (where the height of each amino acid is proportional to its frequency). Amino acids with frequencies greater than 10% are colored according to chemical properties; all others are shown in gray. Peptides were aligned according to the GibbsCluster tool (Supplementary Methods), and the logos represent all peptides, including those that do not closely match the overall motif (e.g., no peptides are isolated in a "junk" cluster). Fig. 30B Depicting MAPTAC incorporation into the pan-DR MS data set TM Example depiction of cluster assignments of peptides (20 per allele). The dataset was deconvoluted using GibbsCluster. Each colored box represents a MAPTAC TM Peptides. The color of the box indicates which cluster it is assigned to, and the gray bar indicates which allele the peptide is from. Fig. 30C Depicted is an exemplary diagram showing Fig. 30B The alleles shown in , show the share of peptides with 0, 1, 2, 3, or 4 expected residues at the anchor position. The anchor positions were defined as the four positions with the lowest entropy, and the "expected" residues were defined as those with ≥ 10% frequency at these positions.

[0370] Figures 31A-31F Depicted is an exemplary architecture and benchmarking of the neoNMHC2 binding prediction algorithm. Fig.31A Depicted is an exemplary architecture of a convolutional neural network (CNN) trained to discriminate monoallelic HLA class II peptides from scrambled length-matched decoys. The schematic illustrates the use of an amino acid feature embedding layer, 2 convolutional layers of width 6, the presence of skip-to-end connections, and a combination of average and max pooling operations as input to the final logic output node. Fig.31B Depicts MAPTAC without using it for training or hyperparameter optimization TMExample positive predictive values ​​(PPVs) of NetMHCIIpan and neonmhc2 evaluated on data partitions. For each allele, n MS-observed peptides are scored together with 19n length-matched decoys sampled from the same set of source genes, and the n top-ranked peptides (e.g., top 5%) for each predictor are called positive. According to this evaluation scheme, PPV is the same as recall, since the number of false positives and false negatives is necessarily equal. Fig.31C Depicted are exemplary PPVs of NetMHCIIpan and NeoMHC2 on the TGEM dataset. For each allele, the n top-ranked peptides were called positive, where n is the number of confirmed immunogenic epitopes in the evaluated set. Fig.31D Depicted are exemplary ex vivo T cell induction results for neoantigenic peptides. Peptides were selected based on a high neonmhc2 score and a weak NetMHCIIpan score for HLA-DRB1*11:01. The figure discloses SEQ ID NOs 87-89, 91, 90, 2, 92-94, 3, and 95-96, respectively, in order of appearance. Fig.31E Depicted is a comparison of a model trained on the monoallelic MAPTAC data and the deconvolved multiallelic data evaluated on the held-out monoallelic data. Values ​​are as for neonmhc2, where the training dataset was downsampled to match the size of the deconvolved training set. Fig.31F PPVs on the TGEM dataset for NetMHCIIpan-v3.1, the deconvolution-trained predictor, and neonmhc2 (with and without downsampling) are shown. For each allele, the n top-ranked peptides were called positive, where n is the number of confirmed immunogenic epitopes in the evaluated set.

[0371] Figures 32A-32E Exemplary gene representation and protein processing in the HLA class II tumor peptidomes are depicted. Fig.32A Depicted are exemplary results of the observed and expected number of HLA class II peptides for each gene determined by joint analysis of colorectal cancer, melanoma, and ovarian cancer data sets. The expected counts are derived by multiplying the gene length by the expression level. The expected and observed counts for the relevant samples are summed. Genes known to be present in plasma are labeled according to their concentration. Fig.32B Exemplary results of expected versus observed frequencies of localized peptides per cell are depicted. Fig.32C The enrichment scores (ratio of observed to expected observations, partly as Fig.32B ) distribution. The gene set includes those with known ubiquitination sites and those whose abundance increases after application of a proteasome inhibitor. Fig.32DDepicted is a diagram presenting three exemplary working models on how HLA class II peptides are processed, according to which i) cathepsins and other enzymes cleave proteins into peptide fragments that are subsequently bound by HLA, ii) proteins or unfolded polypeptides bind HLA and are subsequently cleaved into peptide lengths, iii) proteins are partially digested before binding and further trimmed after binding. Each model corresponds to a different prediction method. Fig.32E Depicted are the absolute increases in PPV observed for logistic regression models that included treatment-related variables and the neoNHC2 binding prediction compared with models using binding prediction alone. Eleven samples analyzed by HLA-DR antibody were evaluated ( Fig. 30B Same samples analyzed in ); each point corresponds to one sample. Asterisks indicate significant improvements according to a two-tailed paired t-test (*: p < 0.01, **: p < 0.001, ***: p < 0.0001). The same analysis is shown in Fig.40B , but using NetMHCIIpan as the base predictor. The methods used for decoy selection and PPV calculation are the same as Fig.31B The same method is used in .

[0372] Figures 33A-33G Depicted are exemplary results for identifying and predicting tumor antigens presented by dendritic cells. Fig.33A An exemplary illustration of the experimental workflow for identifying HLA-II ligands presented by DCs derived from cancer cells (K562) is depicted. Cancer cells were grown in SILAC medium until fully incorporated, lysed or irradiated, and then seeded with monocyte-derived dendritic cells. Presented peptides were isolated with a pan-DR antibody and sequenced by LC-MS / MS. Fig.33B Depicts representative use and Fig.21A Exemplary data of the predictive performance of the same hit:decoy ratios and performance metrics for tumor-derived peptides presented by dendritic cells. The performance of NetMHCIIpan and neoNeonMHC2-based models with and without treatment features is shown. Fig.33C Depicted are exemplary gene expression profiles observed in UV treatment experiments for source genes of heavily labeled peptides (red curve, plotted against K562 expression) compared to source genes of lightly labeled peptides (grey curve, plotted against DC expression). Fig.33DAn exemplary illustration of the PPV of the tumor antigen presented at a hit:decoy ratio of 1:499 using NetMHCIIpan and neonmhc2-based models with and without treatment features is shown. Data points from left to right represent the following samples: Donor 1 HOCl-treated cells: NetMHCIIpan continuous expression; NetMHCIIpan continuous expression + gene preference; NetMHCIIpan continuous expression + gene preference + DQ overlap, full treatment mode; Donor 1, UV treated: neonmhc2; neonmhc2 + threshold expression; neonmhc2 + continuous expression; neonmhc2 + continuous expression + gene preference; neonmhc2 + continuous expression + gene preference + DQ overlap. Fig.33E Depicted are the significance of various gene locations and functional categories in predicting heavy (K562-derived) and light (DC-derived) peptides, respectively. P values ​​were calculated from logistic regression controlling for NeoNMC2 binding score and source gene expression. Bar color indicates the sign associated with the regression coefficient. Fig.33F An exemplary graph of the results is depicted showing the overlap of tumor cell-derived peptide-derived genes (colored by functional category) in UV and HOCl-treated experiments. Figure 33G Depicted are exemplary data showing the PPV for predicting tumor antigens presented in a second donor using a logistic model fit on heavy labeled peptides observed in a first donor. Models were fit using neonmhc2 binding alone; using binding and expression; or using binding, expression, and a binary variable indicating whether the peptide was from a mitochondrial gene.

[0373] Figures 34A-34B Depicted with Fig.29 Related MAPTAC TM Exemplary representation of data. Fig.34A Depicted is the MAPTAC of HLA-A*02:01-BAP encoded with an affinity tag TM Exemplary HLA cell surface analysis by FACS of Expi293 cell line transfected with constructs. Fig.34B Depicted is the MAPTAC of HLA-DRB1*11:01-BAP encoded by affinity tag TMExemplary HLA cell surface analysis by FACS of Expi293 cell lines transfected with constructs (bottom). HLA cell surface expression of transfected Expi293 cells (orange) was compared to stained untransfected Expi293 (blue), unstained untransfected Expi293 (red), stained PBMC (dark green), and unstained PBMC (light green). All HLA class I stains used W6 / 32 (pan-HLA class I), while HLA class II stains used REA332 (pan-HLA class II).

[0374] Fig.35 Depicted with Fig. 30A Related MAPTAC TM Example comparison of IEDB identifications. Measured affinity observed by MS and predicted affinity by NetMHCIIpan for peptides that did not show a good NetMHCIIpan score but were well supported by MS (scoring peak intensity >70 and nesting group size ≥1).

[0375] Figures 36A-36C Depicted with Figures 30A-30C HLA-DR1-associated MAPTAC TM Exemplary analysis of data fidelity. Fig.36A Depicted are HLA-DR1 MAPTAC for common alleles compared to 50,000 length-matched decoy peptides randomly sampled from the proteome (blue). TM Exemplary NetMHCIIpan3.1 scores for peptides (green) (length 12-23). Fig.36B Depicted are exemplary measured affinities and NetMHCIIpan predicted affinities for exemplary MS observed peptides that did not show good NetMHCIIpan scores but were well supported by MS (scoring peak intensity > 70 and nesting group size ≥ 1). The figure discloses SEQ ID NOs 40-86, respectively, from top to bottom and from left to right in order of appearance. Fig.36C Depicted by MAPTAC TM Exemplary HLA class II sequence signatures of HLA-DRB1 alleles determined in different cell types.

[0376] Figures 37A-37C and 37D( Fig.37C Continuation of Figures 30A-30C Related MAPTAC TM Additional exemplary analyses of motifs. Fig.37A Depicts MAPTAC used in experiments with and without HLA-DM co-transfection (expi293 cell line) TMDerived sequence identifier. Fig.37B Depicts the MAPTAC TM Sequence logos of several HLA class I alleles from and IEDB. Note that A*32:01 does not show high frequency Q at P2, and C*03:03 does not show high frequency Y at P9, which is different from previous studies using multi-allelic deconvolution; the logo of B*52:01 has not been previously published. Fig.37C and 37D ( Fig.37C MAPAC TM Exemplary alignment of observed peptides with the gene sequence of CD74.

[0377] Figure 38X , 38Y , 38B-38D depict the Figures 31A-31D Exemplary NeoNMC2 performance statistics and associated T cell flow staining. Figure 38X Plotted is an example performance of neonmc2 as a function of the training dataset size. Fig.31B PPV was evaluated in the same manner and using the same evaluation peptides; however, the training data were randomly downsampled to simulate a smaller training dataset. Figure 38Y Depicted are exemplary sequence logos of peptide clusters derived from a multi-allelic HLA-DR ligand panel using GibbsCluster (default settings; "allow junk clusters"). Fig.38B Depicted are exemplary representative flow cytometry graphs of IFN-γ expression in CD4+ cells from induced samples determined by neoantigen peptides predicted by neoMHC2. Delta values ​​are calculated by subtracting the percentage of CD4+ cells expressing IFN-γ when determined by neoantigens (+peptides) from the percentage of CD4+ expressing IFN-γ when determined in the absence of neoantigens (no peptides). The two flow cytometry graphs on the left represent neoantigens that induce CD4+ T cell T cell responses (PEASLYGALSKGSGG (SEQ ID NO: 2)) and neoantigens that do not induce T cell responses (PATYILILKEFCLVG (SEQ ID NO: 3)). Fig.38C Depicted are exemplary delta values ​​from wells called with a single neonmch2 neoantigenic peptide. If a peptide had a positive response (delta response above 3%, highlighted), it was considered an induced hit. The figure discloses SEQ ID NOs 87-91, 2, 92-94, 3, and 95-96, respectively, in order of appearance. Fig.38D Shown are exemplary sequence logos of peptide clusters derived for a multi-allelic HLA-DR ligand panel using GibbsCluster (default settings; allowing "junk" clusters).

[0378] Figures 39A-39C Depicted with Figures 32A-32E Additional exemplary cell-of-origin analysis of relevant HLA class II. Fig.39A Describes the presence or absence of the peptide-derived gene in human plasma in the presence or absence of the pan-DR antibody (from Fig. 30B Example percentile ranking neonmhc2 scores for HLA class II peptides observed in 4 PBMC samples analyzed (RG1248, RG1104, RG1095, and HDSC from donors). For each peptide, the best (lowest) percentile ranking was used among the alleles present in the donor. Scores for randomly length-matched proteome baits are shown for comparison. The box plots mark the 5th, 25th, 50th, 75th, and 95th percentiles. Fig.39B Describes the use and Fig.32A Example counts of observed versus expected peptides for each gene in HLA class I using the same method as in Figure 2. The data correspond to the same tumor types (colorectal cancer, ovarian cancer, and melanoma). Genes present in human plasma are highlighted in blue and sized according to their concentration. Fig.39C Depicted are exemplary relative concordances of peptide observations for two different gene expression profiles. For each sample, gene-level peptide counts were modeled as a linear combination of the bulk tumor gene expression and the professional APC gene expression profile. The ratio of the coefficients determines the relative concordance of each expression profile with the peptide repertoire. Error bars correspond to 95% confidence intervals calculated by bootstrap resampling.

[0379] Figures 40A-40B Depicted with Figures 32A-32E Additional exemplary analyses of relevant processing motifs. Fig.40A Figure 3 shows the expression of cytokines in donor PBMC, monocyte-derived dendritic cells, colorectal cancer, melanoma, ovarian cancer, and expi293 cell lines (used in most MAPTACs). TM Exemplary amino acid frequencies near the N-terminal and C-terminal peptide cleavage sites relative to the average proteome frequency (for upstream positions U3-U1 and downstream positions D1-D3) or relative to the average peptide frequency (for internal positions N1-C1) as observed in the data generation (data generation). Fig.40B Depicts the use of NetMHCIIpan as a basic predictor with Fig.32E The same analysis was performed for the 8 samples analyzed by HLA-DR antibody ( Fig.31B Figure 2. Absolute increase in PPV observed for logistic regression models that included treatment-related variables in addition to NetMHCIIpan predictions (compared to models using NetMHCIIpan alone) for the same samples analyzed in . Asterisks indicate significant improvements according to a two-tailed paired t-test (*: p<0.01, **: p<0.001, ***: p<0.0001).

[0380] Fig.41 An exemplary nomenclature system for referring to positions upstream of a peptide, within a peptide, and downstream of a peptide is depicted.

[0381] Fig.42A Depicted is a diagram representing an exemplary workflow for analysis of endogenously processed and HLA-1 and HLA class II presented peptides by nLC-MS / MS.

[0382] Fig.42B Depicted are diagrams showing exemplary experimental results from nLC-MS / MS analysis of tryptic peptides with or without FAIMS. Also depicted are representative overlaps of HLA-1 and HLA class II peptides detected by nLC-MS / MS analysis with or without FAIMS at the indicated analytical scale.

[0383] Fig.43A Depicted are exemplary HLA class I acidic and basic reverse phase fractionation peptide detection with and without FAIMS.

[0384] Fig.43B Depicted are exemplary experimental results showing detection of unique peptides for HLA class I binding plotted against retention time.

[0385] Fig.44A Depicted are exemplary HLA class II acidic and basic reverse phase fractionation peptide detection with and without FAIMS.

[0386] Fig.44B Depicted are exemplary experimental results showing detection of unique peptides for HLA class II binding plotted against retention time.

[0387] Fig.45A and 45B Depicted are an exemplary graphical representation of the cross-size of HLA class I binding peptides detected using the indicated methods (left) and a Venn diagram of an exemplary standard workflow and optimized workflow for LC-MS / MS detection of HLA class I binding peptides (right).

[0388] Fig.46A and 46B Depicted are an exemplary graphical representation of the cross-size of HLA class II binding peptides detected using the indicated methods (left) and a Venn diagram of an exemplary standard workflow and optimized workflow for LC-MS / MS detection of HLA class II binding peptides (right).

[0389] Fig.47ADescribe research, in which MHC class II alleles covering a wide range of populations are produced as soluble heterodimers, with cleavable peptide placeholders from transiently transfected human cells. Upper figure, soluble MHC class II construct design (see method and table 19). Lower left, schematic outline or protein expression and purification strategy, to generate the MHCII protein ready for epitope loading, multimerization and flow cytometry staining. Protein purification example of HLA-DRB4*01:03 / DRA*01:01 heterodimers bound to the biotinylation (via BirA) of CLIPO placeholder (PVSKMRMATPLLMQA) and thrombin digestion. Lower right, gel filtration chromatogram and SDS-PAGE gel illustrate the purified fractions collected for epitope loading and flow cytometry staining. Lower figure illustrates gel filtration chromatogram and SDS-PAGE gel, and purified fractions are subsequently collected for epitope loading and flow cytometry staining, with outline marking as "collected" indication.

[0390] Fig.47B The European allele frequencies of the MHC II alleles demonstrated by protein purification are depicted (Table 19).

[0391] Fig.48A Depicted are data indicating that soluble HLA-DM catalyzes rapid, on-demand and universal MHC class II peptide exchange. Schematic diagram (upper left) shows probe binding assay, MHCII alleles loaded with placeholder peptides are exchanged with high-affinity FITC-labeled peptide probes via soluble HLA-DM (catalyst). The right graph shows the peptide binding percentage; the binding of the FITC probe is measured across three (un) catalytic conditions via fluorescence polarization at four time points (see method). The peptide binding percentage is normalized to the 24h soluble HLA-DM catalytic condition. The FITC conjugation site is bold and underlined text (Table 20). The lower left shows the peptide binding characteristics of mouse MHCH2-1-A (b).

[0392] Fig.48B (Left) shows the IC of epitope peptides quantified by fluorescence polarization competition assay. 50 and schematic diagram of allele restriction. The diagram on the right shows the predicted 31 Dose-response IC of SARS-CoV-2 Spike (S)-derived epitopes 50 For each allele, predicted binders (P1-P4) and non-binders (P5-P6) were competed with the FITC probe and peptide binding was measured via fluorescence polarization (Table 21).

[0393] Fig.49ADepicts results showing in-depth characterization of neoantigen-specific CD4+ T cells from personalized peptide vaccine clinical trials, revealing clonal populations with memory and activation phenotypes. PBMCs from cancer patients were stained ex vivo for MHCII multimers by flow cytometry in personalized peptide vaccine clinical trials for non-small cell lung cancer (NSCLC), melanoma, and bladder cancer (Alspach, E. et al. MHC-II neoantigens shape tumour immunity and response to immunotherapy. Nature 574, 696–701 (2019)). Where possible, multimers were combi-coded (Tarke, A. et al. Impact of SARS-CoV-2 variants on the total CD4+andCD8+ T cell reactivity in infected or vaccinated individuals. Cell Rep. Med. 2, 100355 (2021)).

[0394] Fig.49B Depicts from Fig.50A Results from the same study in , showed persistence of the multimer-positive population over the course of treatment. PBMC samples from patients were stained ex vivo before vaccination (week 10) and after vaccination (weeks 20, 52, and 76, as applicable).

[0395] Fig.49C Depicts from Fig.50A and 50B Results from the same study in . Left panel, shows UMAP clustering of bulk CD4 T cells and 3 tetramer+ populations from NSCLC patient L7 based on CITE antibody. Right panel, phenotype of CD4 T cells of bulk and multimer-positive populations from NSCLC patient L7 based on expression levels of CITE antibody.

[0396] Fig.49D Shown are the clonal distribution and abundance of TCRs sorted from NSCLC patient L7.

[0397] Fig.50A Flow cytometry data showing SARS-CoV-2 antigen-specific CD4 T cells identified using MHC class II multimers are shown. Ex vivo identification of SARS-CoV-2 antigen-specific CD4 T cells in five convalescent COVID-19 donor PBMCs using MHC class II multimers +T cells. SARS-CoV-2 spike (S), membrane (M), and nucleocapsid (N) derived epitopes.

[0398] Fig.50B Shows depictions from Fig.49A Data on the characterization of CD4 T cells from the same study in Figure 2. The left panel shows that antigen-specific T cells are primarily effectors (EM) and central memory (CM). Naïve, effector, and memory subsets are based on the expression of CD45RA and CD62L. The right panel shows the SARS-CoV-2 antigen-specific CD4 + Expression of activation and inhibition markers in T cells, such as Lag3, TIM3, PD1, CD69, CD137, and ICOS. Expression is shown for SARS-CoV-2 antigen-specific CD4 + T cells relative to a large number of CD4 + Fold change of mean fluorescence intensity (MFI) of T cells.

[0399] Fig.51A Depicted is a representation of a pMHCII technology platform to study any potential CD4 + Schematic diagram of the strategy for the response. Step 1. MHCII alleles can be purified in parallel with epitope identification (via computational prediction and / or immunogenicity screening). Step 2. Validation of candidate epitope / allele pairs using FP assays. Step 3. Loading epitope peptides of interest onto MHCII via HLA-sDM to generate pMHCII antigens for staining. Step 4. pMHCII is multimerized via conjugation with fluorescent streptavidin and subsequently combinatorially encoded to stain CD4 + T cells (here, three different antigen-specific CD4 + T cell populations are combinatorially coded, each with a unique dual color combination). Step 5. Staining of antigen-specific CD4 + T cells can be further analyzed by flow cytometry for expression markers and / or sorted for single cell analysis.

[0400] Fig.51BDepicted is the purification of soluble HLA-DM from the ExpiCHO culture of transient transfection.Upper left, soluble HLA-DM construct design (see method).Lower left, schematic diagram represents the purification workflow and purification strategy (with timetable) for protein expression, to secrete HLA-sDM from ExpiCHO culture.Use IMAC resin to purify the protein of polyhistidine tag directly from culture medium, and load for downstream epitope.Construct as shown in the figure is transiently transfected into ExpiCHO suspension cells and cultured for a total of 14 days, during which soluble HLA-DM protein is directly secreted into the culture supernatant.Right, SDS-PAGE gel of HLA-sDM purified by IMAC.

[0401] Fig.52 The results showing the rapid MHC II class peptide exchange across many MHCII alleles of soluble HLA-DM catalysis are shown. The combination of FITC probes was measured across three (non-) catalytic conditions via fluorescence polarization at four time points (see the method in Example 17). Peptide binding percentages were normalized to 24h soluble HLA-DM catalytic conditions. The FITC conjugation site of the allele-specific probe is marked with a red underline (all verified FITC probes are referred to Table 20).

[0402] Fig.53A Results are shown, which show that peptide-loaded MHCII tetramers are sensitive to rare antigen-specific CD4 + T cell populations are sensitive and can be multiplexed to detect multiple antigens in one sample. 116-129 Stimulated healthy donor PBMCs were stained for pMHCII tetramers, where epitope-loaded DRB1*01:01 monomers were conjugated to Klickmer (at a defined streptavidin:pMHCII molar ratio) or streptavidin tetramers. CLIP / DRB1*01:01 conjugated to either multimeric scaffold was used as a negative control. Table inset summarizes the antigen-specific CD4 + Frequency and staining index.

[0403] Fig.53B pMHCII tetramer staining for the three epitopes (and a CLIP negative control) is shown, demonstrating antigen-specific CD4 + Sensitive detection of T cells. Influenza (HA1 306-318 )、CMV(pp65 116-129 ) and HIV (Gag 262-276 )-stimulated healthy donor PBMC were serially diluted with unstimulated PBMC from the same donor. The observed multimer-positive frequency was plotted against each antigen-specific CD4 +Linear regression plot between dilutions of T cells; the horizontal dashed line represents the limit of detection based on observed tetramer frequency of unrelated (CLIP) pMHCII staining.

[0404] Fig.53C Combinatorial encoding strategy, tetramer staining flow cytometry plots, and observed / expected tetramer frequencies for three pMHCII antigens (and CLIP negative control) are shown. Healthy donor PBMCs stimulated with influenza, CMV, and HIV epitopes were mixed in equal ratios and stained with the corresponding loading DRB1*01:01 tetramers. The table inset summarizes the tetramer-positive frequencies between the single-epitope and combinatorially encoded multi-epitope staining approaches. Flow cytometry plots show pMHCII tetramer-positive populations for all three antigens using combinatorial encoding (bottom). Right, CD8 + (Top) or CD4 + (Bottom) pMHCII tetramer staining of T cells gated.

[0405] Fig.54A Figure 3 is shown for characterization of pMHCII tetramer-positive CD4 from convalescent COVID-19 donors. + T cell gating scheme for Figures 54B-54E The data shown in .

[0406] Fig.54B PBMCs from COVID-19 convalescent donors M, Q, N, O, and P stained for irrelevant peptides (CLIP, IGRP, and / or proinsulin) are shown. Naïve, effector, central memory (CM), and effector memory (EM) subsets were defined by CD45RA and CD62L surface expression. + Cells were gated on the surface of the cells. Quantification of activation and depletion markers for each tetramer-positive population is shown. Histogram values ​​are based on the mean fluorescence intensity (MFI) of each surface marker.

[0407] Fig.54C Data are presented for the phenotypic characterization of bulk (grey, back) and multimer-positive (red, foreground) CD4 T cell populations from convalescent COVID19 donor #N. Naïve, effector, central memory (CM), and effector memory (EM) subsets were defined by CD45RA and CD62L surface expression. + Cells were gated on the surface of the cells. Quantification of activation and depletion markers for each tetramer-positive population is shown. Histogram values ​​are based on the mean fluorescence intensity (MFI) of each surface marker.

[0408] Fig.54DPhenotypic characterization of bulk (grey, back) and multimer-positive (red, foreground) CD4 T cell populations from convalescent COVID19 donor #P is shown. Naïve, effector, central memory (CM), and effector memory (EM) subsets were defined by CD45RA and CD62L surface expression. + Cells were gated on the surface of the cells. Quantification of activation and depletion markers for each tetramer-positive population is shown. Histogram values ​​are based on the mean fluorescence intensity (MFI) of each surface marker.

[0409] Fig.54E Phenotypic characterization of bulk (grey, back) and multimer-positive (red, foreground) CD4 T cell populations from convalescent COVID19 donor #O is shown. Naïve, effector, central memory (CM), and effector memory (EM) subsets were defined by CD45RA and CD62L surface expression. + Cells were gated on the surface of the cells. Quantification of activation and depletion markers for each tetramer-positive population is shown. Histogram values ​​are based on the mean fluorescence intensity (MFI) of each surface marker.

[0410] Fig.55A Antigen-specific CD4 from patients enrolled in a personalized peptide cancer vaccine trial is shown. + MHCII multimer analysis and sorting of T cells. The top panel shows the gating scheme for sorting multimer-positive cells for CITEseq and TCRseq analysis. The middle panel shows irrelevant peptide (CLIP) staining of PBMCs from NSCLC patient L7, bladder cancer patients B9 and B10, and melanoma patient M23. The bottom panel shows a single-presentation bulk CD4 + CD4 sorting of T cells and three pMHCII tetramers + UMAP analysis of CITE marker expression levels of T cells.

[0411] Fig.55B (Left) Shown are the expression levels and clustering of specific CITE markers overlaid on total UMAP for bulk + all multimer sorted CD4 + Right, expression levels and clustering of specific CITE markers overlaid on total UMAP for bulk plus all multimer sorted CD4 + T cells.

[0412] Fig.55C (Left) Shows the CD4 +UMAP distribution of the top TCR clones of the T cell population; (right) CD4 of the top 5 TCR clones of each tetramer-sorted population from NSCLC L7 + T cell phenotype distribution.

[0413] Fig.56 Data showing that high post-translational modifications (PTMs) of MHC class II proteins affect staining with labeled epitopes that can bind to MHC class II proteins are shown (shown here is an exemplary MHC class II protein DRB1*01:01). Comparison of row 2 from the top with row 1 shows that low PTM DRB1*01:01 confers superior staining properties compared to high PTM MHCII proteins. Similarly, low PTM DRB1*01:01 shows high fluorescent staining with exchange epitopes, comparable to the data in the middle row. DETAILED DESCRIPTION

[0414] All terms should be understood as they would be understood by those skilled in the art. Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this disclosure belongs.

[0415] The section headings used herein are for organizational purposes only and are not to be construed as limiting the subject matter described.

[0416] Although various features of the present disclosure may be described in the context of a single embodiment, these features may also be provided separately or in any suitable combination. Conversely, although for clarity, the present disclosure may be described herein in the context of separate embodiments, the present disclosure may also be implemented in a single embodiment.

[0417] The present disclosure is based on the important discovery that presentation of antigens, particularly cancer antigens, by specific HLA class II α and β chain pairs can be predicted with high confidence using a new computer-based machine learning HLA-peptide presentation prediction model that allows for improved immunotherapy using HLA class II-specific peptides.

[0418] In one aspect, the present disclosure provides methods for predicting peptides that can accurately pair or bind to specific HLA class II α and β chain heterodimers, such that high fidelity binding of the peptide to HLA class II proteins (including α and β chain heterodimers) ensures that the specific peptide is presented to T lymphocytes, thereby eliciting a specific immune response and avoiding any cross-reactions or immune promiscuity. Several recent studies have shown that CD4+ T cells can also recognize ligands presented by HLA class II and help control tumors. Ideally, cancer vaccines and other immunotherapies would utilize guided CD4+ T cell responses, but current efforts have completely abandoned HLA class II antigen predictions because of the lack of accuracy of current prediction tools.

[0419] In one aspect, the present disclosure provides methods for predicting peptides that can accurately bind to specific HLA class II proteins, so that when the peptide is therapeutically administered to a subject expressing a specific homologous HLA class II protein, the peptide can be used to activate a more sustained and powerful immune response by virtue of the ability of the HLA class II protein to activate CD4+ T cells and stimulate immune memory. In some embodiments, the methods provided herein show improvements in predictions of specific HLA class II proteins relative to currently available predictors. In some embodiments, the methods provided herein show at least an improvement of about 1.1 times in predictions of specific HLA class II proteins relative to currently available predictors. In some embodiments, the methods provided herein show at least an improvement of about 2 times in predictions of specific HLA class II proteins relative to currently available predictors. In some embodiments, the methods provided herein show at least an improvement of about 3 times in predictions of specific HLA class II proteins relative to currently available predictors. In some embodiments, the methods provided herein show at least an improvement of about 4 times in predictions of specific HLA class II proteins relative to currently available predictors. In some embodiments, the methods provided herein show at least about 5-fold improvement in prediction of a particular HLA class II protein relative to currently available predictors. In some embodiments, the methods provided herein show at least about 6-fold improvement in prediction of a particular HLA class II protein relative to currently available predictors. In some embodiments, the methods provided herein show at least about 7-fold improvement in prediction of a particular HLA class II protein relative to currently available predictors. In some embodiments, the methods provided herein show at least about 8-fold improvement in prediction of a particular HLA class II protein relative to currently available predictors. In some embodiments, the methods provided herein show at least about 9-fold improvement in prediction of a particular HLA class II protein relative to currently available predictors. In some embodiments, the methods provided herein show at least about 10-fold improvement in prediction of a particular HLA class II protein relative to currently available predictors. In some embodiments, the methods provided herein show at least about 15-fold improvement in prediction of a particular HLA class II protein relative to currently available predictors. In some embodiments, the methods provided herein show at least about 20-fold improvement in prediction of a particular HLA class II protein relative to currently available predictors. In some embodiments, the methods provided herein show at least about 30-fold improvement in prediction of a particular HLA class II protein relative to currently available predictors. In some embodiments, the methods provided herein show at least about 40-fold improvement in prediction of a particular HLA class II protein relative to currently available predictors.In some embodiments, the methods provided herein show at least about 50-fold improvement in prediction of a particular HLA class II protein relative to currently available predictors. In some embodiments, the methods provided herein show at least about 60-fold improvement in prediction of a particular HLA class II protein relative to currently available predictors.

[0420] On the one hand, the present invention proposes a customized or personalized immunotherapy method for a specific subject. Each subject or patient expresses a specific set of HLA class I and HLA class II proteins. HLA typing is a well-known technology that allows the determination of a specific repertoire of HLA proteins expressed by a subject. Once the HLA heterodimers expressed by a specific subject are known, it is possible to generate a specific immune response specifically customized for the subject with improved, complex and reliable methods for predicting peptides that can bind to specific HLA class II alpha and beta chain heterodimers with high fidelity as described herein.

[0421] In this application, unless otherwise specifically stated, the use of the singular includes the plural. It must be noted that, as used in this specification, the singular forms "a", "an" and "the" include plural referents unless the context clearly indicates otherwise. In this application, the use of "or" means "and / or" unless otherwise stated. In addition, the use of the term "includes" and other forms such as "comprises", "contains" and "has" is not restrictive. The term "one or more" or "at least one", such as one or more or at least one member of a group of members, is clear in itself, and by further illustration, the term particularly includes reference to any one of the members, or any two or more of the members, for example, any 3, 4, 5, 6 or 7 of the members, etc., up to all the members.

[0422] References in this specification to "some embodiments," "embodiments," "one embodiment," or "other embodiments" mean that a feature, structure, or characteristic described in connection with the embodiment is included in at least some embodiments of the present disclosure, but not necessarily in all embodiments.

[0423] As used in this specification and claims, the words "comprising" (and any form of comprising), "having" (and any form of having), "including" (and any form of including), or "containing" (and any form of containing) are inclusive or open-ended and do not exclude other unrecited elements or method steps. It is contemplated that any embodiment discussed in this specification can be implemented using any method or composition of the present disclosure, and vice versa. In addition, the compositions of the present disclosure can be used to implement the methods of the present disclosure.

[0424] As used herein, the terms "about" or "approximately" when referring to measurable values ​​such as parameters, amounts, durations, etc., are intended to encompass variations of + / -20% or less, + / -10% or less, + / -5% or less, or + / -1% or less of the specified value, as long as such variations are suitable for use in the present disclosure. It should be understood that the value to which the modifier "about" or "approximately" refers is itself also specifically disclosed.

[0425] The term "immune response" includes T cell-mediated and / or B cell-mediated immune responses that are affected by T cell costimulation regulation. Exemplary immune responses include T cell responses, such as the production of cytokines and the cytotoxicity of cells. In addition, the term immune response includes immune responses that are indirectly affected by T cell activation, such as antibody production (humoral response) and cytokine responsive cells such as the activation of macrophages.

[0426] "Receptor" should be understood to refer to a biological molecule or grouping of molecules that can bind to a ligand. Receptors can be used to transmit information in cells, cell formations or organisms. Receptors contain at least one receptor unit and may contain two or more receptor units, wherein each receptor unit may be composed of a protein molecule, such as a glycoprotein molecule. The receptor has a structure that is complementary to the structure of the ligand and can be complexed with the ligand as a binding partner. Signaling information can be transmitted by conformational changes after the receptor binds to the ligand on the cell surface. According to the present disclosure, a receptor may refer to MHC class I and class II proteins that can form a receptor / ligand complex with a ligand, such as a peptide or peptide fragment of appropriate length. Class I and class II MHC peptides encoded by HLA class I and class II alleles are generally referred to herein as HLA class I and HLA class II peptides, or HLA class I and HLA class II peptides, or HLA class I and class II proteins, or HLA class I and class II proteins, or HLA class I and class II molecules, or such common variants thereof, as known to those of ordinary skill in the art in the context of the discussion.

[0427] A "ligand" is a molecule that is able to form a complex with a receptor. According to the present disclosure, a ligand is understood to refer to a peptide or peptide fragment having, for example, a suitable length and a suitable binding motif in its amino acid sequence, such that the peptide or peptide fragment is able to bind to and form a complex with an MHC class I or MHC class II protein (i.e., HLA class I and HLA class II protein).

[0428] "Antigen" is a molecule that can stimulate an immune response and can be produced by cancer cells or infectious agents or autoimmune diseases. Antigens recognized by T cells (whether helper T lymphocytes (T helper (TH) cells) or cytotoxic T lymphocytes (CTL)) are not recognized as complete proteins, but as small peptides associated with HLA class I or class II proteins on the cell surface. In the process of naturally occurring immune responses, antigens recognized by association with HLA class II molecules on antigen presenting cells (APCs) are obtained from outside the cell, internalized, and processed into small peptides associated with HLA class II molecules. APCs can also cross-present peptide antigens by processing exogenous antigens and presenting the processed antigens to HLA class I molecules. Antigens that produce peptides recognized by association with HLA class I MHC molecules are usually peptides produced intracellularly, and these antigens are processed and associated with class I MHC molecules. It is now understood that peptides that associate with a given HLA class I or class II molecule are characterized as having a common binding motif, and binding motifs have been determined for a large number of different HLA class I and class II molecules. Synthetic peptides that correspond to the amino acid sequence of a given antigen and contain a binding motif for a given HLA class I or class II molecule can also be synthesized. These peptides can then be added to appropriate APCs, and the APCs can be used to stimulate a T helper cell or CTL response in vitro or in vivo. The binding motifs, methods of synthesizing peptides, and methods of stimulating a T helper cell or CTL response are all known to those of ordinary skill in the art and are readily available.

[0429] In this specification, the term "peptide" is used interchangeably with "mutant peptide" and "neoantigenic peptide". Similarly, in this specification, the term "polypeptide" is used interchangeably with "mutant polypeptide" and "neoantigenic polypeptide". "Neoantigen" or "neo-epitope" refers to a class of tumor antigens or tumor epitopes produced by tumor-specific mutations in expressed proteins. The present disclosure further includes peptides comprising tumor-specific mutations, peptides comprising known tumor-specific mutations, and mutant polypeptides or fragments thereof identified by the methods of the present disclosure. These peptides and polypeptides are referred to herein as "neoantigenic peptides" or "neoantigenic polypeptides". These polypeptides or peptides may have a variety of lengths, may be in their neutral (uncharged) form, may be in salt form, and may not contain modifications such as glycosylation, side chain oxidation, phosphorylation, or any post-translational modification, or may contain these modifications, provided that the modification does not destroy the biological activity of the polypeptide described herein. In some embodiments, the neoantigenic peptides of the present disclosure may include: for HLA class I, a length of 22 or less residues, for example, about 8 to about 22 residues, about 8 to about 15 residues, or 9 or 10 residues; for HLA class II, a length of 40 or less residues, for example, a length of about 8 to about 40 residues, a length of about 8 to about 24 residues, about 12 to about 19 residues, or about 14 to about 18 residues. In some embodiments, the neoantigenic peptide or neoantigenic polypeptide comprises a neoepitope.

[0430] The term "epitope" includes any protein determinant that can specifically bind to an antibody, antibody peptide and / or antibody-like molecule (including but not limited to T cell receptor) as defined herein. Epitope determinants are usually composed of chemically active surface groups of molecules such as amino acids or sugar side chains, and usually have specific three-dimensional structural characteristics and specific charge characteristics.

[0431] A "T cell epitope" is a peptide sequence which can be bound by an MHC molecule of class I or II in the form of a peptide-presenting MHC molecule or MHC complex and which is then recognized and bound in this form by cytotoxic T lymphocytes or T helper cells, respectively.

[0432] The term "antibody" as used herein includes IgG (including IgG1, IgG2, IgG3 and IgG4), IgA (including IgA1 and IgA2), IgD, IgE, IgM and IgY, and is intended to include complete antibodies, including single-chain complete antibodies, and antigen-binding (Fab) fragments thereof. Antigen-binding antibody fragments include, but are not limited to, Fab, Fab' and F(ab')2, Fd (consisting of VH and CH1), single-chain variable fragments (scFv), single-chain antibodies, disulfide-linked variable fragments (dsFv) and fragments comprising VL or VH domains. Antibodies can be from any animal source. Antigen-binding antibody fragments, including single-chain antibodies, can contain variable regions alone or in combination with all or part of the following: hinge region, CH1, CH2 and CH3 domains. Any combination of variable regions and hinge regions, CH1, CH2 and CH3 domains is also included. The antibodies can be, for example, monoclonal antibodies, polyclonal antibodies, chimeric antibodies, humanized antibodies, and human monoclonal and polyclonal antibodies that specifically bind to HLA-associated polypeptides or HLA-HLA binding peptide (HLA-peptide) complexes. Those skilled in the art will recognize that a variety of immunoaffinity techniques are suitable for enriching soluble proteins, such as soluble HLA-peptide complexes or membrane-bound HLA-associated polypeptides, for example, which have been cleaved from a membrane by proteolysis. This includes techniques in which (1) one or more antibodies capable of specifically binding to soluble proteins are fixed to a fixed or removable substrate (e.g., a plastic well or resin, latex or paramagnetic beads), and (2) a solution containing soluble proteins from a biological sample is passed through the antibody-coated substrate, thereby binding the soluble proteins to the antibodies. The substrate with the antibodies and bound soluble proteins is separated from the solution, and the antibodies and soluble proteins are optionally dissociated, for example, by changing the pH and / or ionic strength and / or ionic composition of the solution in which the antibodies are bathed. Alternatively, immunoprecipitation techniques can be used, in which antibodies and soluble proteins are combined and form macromolecular aggregates. The macromolecular aggregates can be separated from the solution by size exclusion techniques or by centrifugation.

[0433] The term "immunopurification (IP)" (or immunoaffinity purification or immunoprecipitation) is a method well known in the art and is widely used to separate desired antigens from samples. Typically, the method includes contacting a sample containing the desired antigen with an affinity matrix comprising antibodies to the antigen covalently attached to a solid phase. The antigen in the sample is bound to the affinity matrix via immunochemical bonds. The affinity matrix is ​​then washed to remove any unbound material. The antigen is removed from the affinity matrix by changing the chemical composition of the solution in contact with the affinity matrix. Immunopurification can be performed on a column containing an affinity matrix, in which case the solution is an eluent. Alternatively, immunopurification can be a batch process, in which case the affinity matrix is ​​maintained as a suspension in a solution. An important step in the process is to remove the antigen from the matrix. This is typically achieved by increasing the ionic strength of the solution in contact with the affinity matrix, such as by adding an inorganic salt. Changes in pH can also effectively dissociate the immunochemical bonds between the antigen and the affinity matrix.

[0434] An "agent" is any small molecule compound, antibody, nucleic acid molecule or polypeptide or fragment thereof.

[0435] A "change" or "variation" is an increase or decrease. The change may be as little as 1%, 2%, 3%, 4%, 5%, 10%, 20%, 30% or 40%, 50%, 60%, or even as much as 70%, 75%, 80%, 90% or 100%.

[0436] A "biological sample" is any tissue, cell, fluid or other substance derived from an organism. As used herein, the term "sample" includes biological samples, such as any tissue, cell, fluid or other substance derived from an organism. "Specific binding" refers to a compound (e.g., a peptide) that recognizes and binds to a molecule (e.g., a polypeptide) but does not substantially recognize and bind to other molecules in a sample (e.g., a biological sample).

[0437] "Capture reagent" refers to a reagent that specifically binds to a molecule (eg, a nucleic acid molecule or a polypeptide) in order to select or isolate the molecule (eg, a nucleic acid molecule or a polypeptide).

[0438] As used herein, the terms "determine," "assess," "determine," "measure," "detect," and their grammatical equivalents refer to both quantitative and qualitative determinations, and thus, the terms "determine" and "determine," "measure," and the like are used interchangeably herein. Where quantitative determination is intended, the phrase "determine the amount of an analyte, etc." is used. Where qualitative and / or quantitative determination is intended, the phrase "determine the level of an analyte" or "detect" an analyte is used.

[0439] A "fragment" is a portion of a protein or nucleic acid that is substantially identical to a reference protein or nucleic acid. In some embodiments, the portion retains at least 50%, 75%, or 80%, or 90%, 95%, or even 99% of the biological activity of the reference protein or nucleic acid described herein.

[0440] The terms "isolated," "purified," "biologically pure," and grammatical equivalents thereof refer to a substance that is freed to varying degrees from components that normally accompany it in its native state. "Isolated" means a degree of separation from an original source or environment. "Purity" means a degree of separation greater than isolated. A "purified" or "biologically pure" protein is sufficiently free of other substances such that any impurities do not materially affect the biological properties of the protein or cause other adverse consequences. That is, a nucleic acid or peptide of the present disclosure is purified if it is substantially free of cellular material, viral material, or culture medium when produced by recombinant DNA technology, or substantially free of chemical precursors or other chemicals when chemically synthesized. Purity and homogeneity are typically determined using analytical chemistry techniques, such as polyacrylamide gel electrophoresis or high performance liquid chromatography. The term "purified" may mean that a nucleic acid or protein produces essentially one band in an electrophoretic gel. For proteins that can be modified, such as phosphorylation or glycosylation, different modifications may produce different isolated proteins that can be purified separately.

[0441] An "isolated" polypeptide (e.g., a peptide from an HLA-peptide complex) or polypeptide complex (e.g., an HLA-peptide complex) is a polypeptide or polypeptide complex of the present disclosure that has been separated from naturally associated components. Typically, a polypeptide or polypeptide complex is isolated when it is at least 60% by weight free of proteins and naturally occurring organic molecules with which it is naturally associated. The preparation may be at least 75%, at least 90%, or at least 99% by weight of a polypeptide or polypeptide complex of the present disclosure. An isolated polypeptide or polypeptide complex of the present disclosure may be obtained, for example, by extraction from a natural source, by expression of a recombinant nucleic acid encoding one or more components of the polypeptide or polypeptide complex, or by chemical synthesis of one or more components of the polypeptide or polypeptide complex. Purity may be measured by any appropriate method, such as column chromatography, polyacrylamide gel electrophoresis, or by HPLC analysis. In some cases, the MHC class II protein (i.e., MHC class II peptide) encoded by an HLA allele is referred to interchangeably in this document as an HLA class II protein (or HLA class II peptide).

[0442] The term "vector" refers to a nucleic acid molecule capable of transporting or mediating the expression of heterologous nucleic acids. Plasmid is one of the types covered by the term "vector". A vector generally refers to a nucleic acid sequence containing a replication origin and other entities necessary for replication and / or maintenance in a host cell. A vector capable of directing the expression of a gene and / or nucleic acid sequence operably connected thereto is referred to herein as an "expression vector". Typically, a useful expression vector is typically in the form of a "plasmid", which refers to a circular double-stranded DNA molecule that is not bound to a chromosome in the form of a vector and typically contains an entity or encoded DNA for stable or transient expression. Other expression vectors that can be used in the methods disclosed herein include, but are not limited to, plasmids, episomes, bacterial artificial chromosomes, yeast artificial chromosomes, phages or viral vectors, and such vectors can be integrated into the host's genome or replicate autonomously in a cell. The vector can be a DNA or RNA vector. Other forms of expression vectors known to those skilled in the art that function equivalently, for example, self-replicating extrachromosomal vectors or vectors capable of integrating into the host genome, can also be used. An exemplary vector is a vector capable of autonomous replication and / or expression of a nucleic acid connected thereto.

[0443] The term "spacer" or "connector" used for fusion proteins refers to a peptide that connects the protein containing the fusion protein. Generally, the spacer has no specific biological activity except connecting or maintaining a certain minimum distance or other spatial relationship between the protein or RNA sequence. However, in some embodiments, the constituent amino acids of the spacer can be selected to affect certain properties of the molecule, such as the folding, net charge or hydrophobicity of the molecule. Suitable connectors for the embodiments of the present disclosure are well known to those skilled in the art, and include but are not limited to straight or branched carbon connectors, heterocyclic carbon connectors or peptide connectors. The connector is used to separate the two antigenic peptides by a certain distance, which is sufficient to ensure that each antigenic peptide is correctly folded in some embodiments. Exemplary peptide connector sequences adopt a flexible extended conformation and do not show a tendency to develop an ordered secondary structure. Typical amino acids in flexible protein regions include Gly, Asn and Ser. In fact, it is expected that any arrangement of the amino acid sequence containing Gly, Asn and Ser will meet the above-mentioned standards for the connector sequence. Other near-neutral amino acids, such as Thr and Ala, can also be used in the connector sequence. Other amino acid sequences that can be used as linkers are disclosed in Maratea et al. (1985), Gene 40:39-46, Murphy et al. (1986) Proc. Nat'l. Acad. Sci. USA 83:8258-62, US Pat. No. 4,935,233, and US Pat. No. 4,751,180.

[0444] The term "neoplasia" refers to any disease that results in or is caused by inappropriately high levels of cell division, inappropriately low levels of apoptosis, or both. Glioblastoma is a non-limiting example of a neoplasia or cancer. The term "cancer" or "tumor" or "hyperproliferative disorder" refers to the presence of cells having typical characteristics of cancer cells such as uncontrolled proliferation, unlimited proliferation, metastatic potential, rapid growth and proliferation rate, and certain unique morphological characteristics. Cancer cells are usually in the form of tumors, but such cells can exist alone in an animal or can be non-tumorigenic cancer cells, such as leukemia cells. Cancers include, but are not limited to, B-cell cancers, e.g., multiple myeloma, Waldenstrom's macroglobulinemia, heavy chain diseases (e.g., alpha chain disease, gamma chain disease, and mu chain disease), benign monoclonal gammopathy and immune cell amyloidosis, melanoma, breast cancer, lung cancer, bronchial cancer, colorectal cancer, prostate cancer (e.g., metastatic, hormone-refractory prostate cancer), pancreatic cancer, gastric cancer, ovarian cancer, bladder cancer, brain or central nervous system cancer, peripheral nervous system cancer, esophageal cancer, cervical cancer, uterine cancer or endometrial cancer, cancer of the oral cavity or pharynx, liver cancer, kidney cancer, testicular cancer, biliary tract cancer, small intestine or appendix cancer, salivary gland cancer, thyroid cancer, adrenal cancer, osteosarcoma, chondrosarcoma, cancer of blood tissue, etc. Other non-limiting examples of cancer types suitable for the methods encompassed by the present disclosure include human sarcomas and carcinomas, e.g., fibrosarcoma, myxosarcoma, liposarcoma, chondrosarcoma, osteosarcoma, chordoma, angiosarcoma, endotheliosarcoma, lymphangiosarcoma, lymphangioendotheliosarcoma, synovioma, mesothelioma, Ewing's tumor, tumor), leiomyosarcoma, rhabdomyosarcoma, colon cancer, colorectal cancer, pancreatic cancer, breast cancer, ovarian cancer, squamous cell carcinoma, basal cell carcinoma, adenocarcinoma, sweat gland cancer, sebaceous gland cancer, papillary carcinoma, papillary adenocarcinoma, cystadenocarcinoma, medullary carcinoma, bronchogenic carcinoma, renal cell carcinoma, hepatoma, bile duct cancer, liver cancer, choriocarcinoma, seminoma, embryonal carcinoma, Wilms' tumor, cervical cancer, bone cancer, brain tumor, testicular cancer, lung cancer, small cell lung cancer, bladder cancer, epithelial cancer, glioma, astrocytoma, medulloblastoma, craniopharyngioma, ependymoma, pineal leukemias, e.g., acute lymphocytic leukemia and acute myeloid leukemia (myeloblastic, promyelocytic, myelomonocytic, monocytic and erythroleukemia); chronic leukemias (chronic myeloid (granulocytic) leukemia and chronic lymphocytic leukemia); as well as polycythemia vera, lymphomas (Hodgkin's disease and non-Hodgkin's disease), multiple myeloma, Waldenstrom's macroglobulinemia and heavy chain disease.In some embodiments, the cancer is an epithelial cancer, such as but not limited to bladder cancer, breast cancer, cervical cancer, colon cancer, gynecological cancer, kidney cancer, laryngeal cancer, lung cancer, oral cancer, head and neck cancer, ovarian cancer, pancreatic cancer, prostate cancer or skin cancer. In other embodiments, the cancer is breast cancer, prostate cancer, lung cancer or colon cancer. In other embodiments, the epithelial cancer is non-small cell lung cancer, non-papillary renal cell carcinoma, cervical cancer, ovarian cancer (e.g., serous ovarian cancer) or breast cancer. Epithelial cancer can be characterized in various other ways, including but not limited to serous, endometrioid, mucinous, clear cell, Brenner type (brenner) or undifferentiated. In some embodiments, the present disclosure is used for the treatment, diagnosis and / or prognosis of lymphoma or its subtypes (including but not limited to mantle cell lymphoma). Lymphoproliferative disorders are also considered to be proliferative diseases.

[0445] The term "vaccine" should be understood to refer to a composition used to generate immunity to prevent and / or treat a disease (e.g., neoplasia / tumor / infectious agent / autoimmune disease). Therefore, a vaccine is a drug containing an antigen and is intended to be used in humans or animals to produce specific defenses and protective substances by vaccination. A "vaccine composition" may include a pharmaceutically acceptable excipient, carrier, or diluent. Aspects of the present disclosure relate to the use of this technology in the preparation of antigen-based vaccines. In these embodiments, a vaccine refers to one or more disease-specific antigenic peptides (or their corresponding nucleic acids encoding them). In some embodiments, the antigen-based vaccine contains at least two, at least three, at least four, at least five, at least six, at least seven, at least eight, at least nine, at least 10, at least 11, at least 12, at least 13, at least 14, at least 15, at least 16, at least 17, at least 18, at least 19, at least 20, at least 21, at least 22, at least 23, at least 24, at least 25, at least 26, at least 27, at least 28, at least 29, at least 30 or more antigenic peptides. In some embodiments, the antigen-based vaccine contains 2 to 100, 2 to 75, 2 to 50, 2 to 25, 2 to 20, 2 to 19, 2 to 18, 2 to 17, 2 to 16, 2 to 15, 2 to 14, 2 to 13, 2 to 12, 2 to 10, 2 to 9, 2 to 8, 2 to 7, 2 to 6, 2 to 5, 2 to 4, 3 to 100, 3 to 75, 3 to 50, 3 to 25, 3 to 20, 3 to 19, 3 to 18, 3 to 17, 3 to 16, 3 to 15, 3 to 14, 3 to 13, 3 to 12, 3 to 10, 3 to 9, 3 to 8 4 to 15, 4 to 14, 4 to 13, 4 to 12, 4 to 10, 4 to 9, 4 to 8, 4 to 7, 4 to 6, 5 to 100, 5 to 75, 5 to 50, 5 to 25, 5 to 20, 5 to 19, 5 to 18, 4 to 17, 4 to 16, 4 to 15, 4 to 14, 4 to 13, 4 to 12, 4 to 10, 4 to 9, 4 to 8, 4 to 7, 4 to 6, 5 to 100, 5 to 75, 5 to 50, 5 to 25, 5 to 20, 5 to 19, 5 to 18, 5 to 17, 5 to 16, 5 to 15, 5 to 14, 5 to 13, 5 to 12, 5 to 10, 5 to 9, 5 to 8, or 5 to 7 antigenic peptides. In some embodiments, the antigen-based vaccine contains 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19 or 20 antigenic peptides. In some cases, the antigenic peptide is a new antigenic peptide. In some cases, the antigenic peptide comprises one or more new epitopes.

[0446] The term "pharmaceutically acceptable" refers to approved or approvable by a regulatory agency of a federal or state government, or listed in the U.S. Pharmacopeia or other generally recognized pharmacopoeia for use in animals, including humans. "Pharmaceutically acceptable excipient, carrier or diluent" refers to an excipient, carrier or diluent that can be applied to a subject together with a medicament and does not destroy its pharmacological activity and is non-toxic when applied in a dose sufficient to deliver a therapeutic amount of the medicament. As described herein, a "pharmaceutically acceptable salt" of a combined disease-specific antigen can be an acid salt or basic salt that is generally considered in the art to be suitable for contact with human or animal tissues without excessive toxicity, irritation, allergic reactions or other problems or complications. Such salts include inorganic and organic acid salts of basic residues such as amines, and alkali metal or organic salts of acidic residues such as carboxylic acids. Specific pharmaceutical salts include, but are not limited to, salts of acids such as hydrochloric acid, phosphoric acid, hydrobromic acid, malic acid, glycolic acid, fumaric acid, sulfuric acid, p-sulfamic acid, aminobenzenesulfonic acid, formic acid, toluenesulfonic acid, methanesulfonic acid, benzenesulfonic acid, ethanedisulfonic acid, 2-hydroxyethylsulfonic acid, nitric acid, benzoic acid, 2-acetoxybenzoic acid, citric acid, tartaric acid, lactic acid, stearic acid, salicylic acid, glutamic acid, ascorbic acid, pamoic acid, succinic acid, fumaric acid, maleic acid, propionic acid, hydroxymaleic acid, hydroiodic acid, phenylacetic acid, alkanoic acids such as acetic acid, HOOC-(CH2)n-COOH, where n is 0-4, etc. Similarly, pharmaceutically acceptable cations include, but are not limited to, sodium, potassium, calcium, aluminum, lithium, and ammonium. Those of ordinary skill in the art will recognize from the present disclosure and the knowledge in the art that other pharmaceutically acceptable salts of the combined disease-specific antigens provided herein include those listed in Remington's Pharmaceutical Sciences, 17th edition, Mack Publishing Company, Easton, PA, p. 1418 (1985). Generally, pharmaceutically acceptable acid salts or basic salts can be synthesized from parent compounds containing basic or acidic moieties by any conventional chemical method. In short, such salts can be prepared by reacting the free acid or base form of these compounds with a stoichiometric amount of an appropriate base or acid in a suitable solvent.

[0447] Nucleic acid molecules that can be used in the methods of the present disclosure include any nucleic acid molecules encoding polypeptides of the present disclosure or fragments thereof. Such nucleic acid molecules do not have to be 100% identical to endogenous nucleic acid sequences, but generally show substantial identity. Polynucleotides having substantial identity to endogenous sequences are generally capable of hybridizing with at least one strand of a double-stranded nucleic acid molecule. "Hybridization" refers to the pairing of nucleic acid molecules under various stringent conditions between complementary polynucleotide sequences or portions thereof to form double-stranded molecules. (See, e.g., Wahl, GM and SL Berger (1987) Methods Enzymol. 152: 399; Kimmel, AR (1987) Methods Enzymol. 152: 507). For example, stringent salt concentrations may generally be less than about 750 mM NaCl and 75 mM trisodium citrate, less than about 500 mM NaCl and 50 mM trisodium citrate, or less than about 250 mM NaCl and 25 mM trisodium citrate. Low stringency hybridization can be obtained in the absence of organic solvents such as formamide, while high stringency hybridization can be obtained in the presence of at least about 35% formamide or at least about 50% formamide. Stringent temperature conditions can generally include a temperature of at least about 30°C, at least about 37°C, or at least about 42°C. It is well known to those skilled in the art to change other parameters, such as hybridization time, the concentration of detergents such as sodium dodecyl sulfate (SDS), and the inclusion or exclusion of carrier DNA. Various stringency levels are achieved by combining these various conditions as needed. In an exemplary embodiment, hybridization can occur at 30°C in 750mM NaCl, 75mM trisodium citrate, and 1% SDS. In another exemplary embodiment, hybridization can be carried out at 37°C in 500mM NaCl, 50mM trisodium citrate, 1% SDS, 35% formamide, and 100 μg / ml denatured salmon sperm DNA (ssDNA). In another exemplary embodiment, hybridization can occur at 42°C in 250mM NaCl, 25mM trisodium citrate, 1% SDS, 50% formamide, and 200 μg / ml ssDNA. Useful variations on these conditions will be apparent to those skilled in the art. For most applications, the washing steps after hybridization may also differ in stringency. Washing stringency conditions may be defined by salt concentration and temperature. As described above, washing stringency may be increased by reducing salt concentration or by increasing temperature. For example, the stringent salt concentration of the washing step may be less than about 30mM NaCl and 3mM trisodium citrate, or less than about 15mM NaCl and 1.5mM trisodium citrate. The stringent temperature conditions of the washing step may include a temperature of at least about 25°C, at least about 42°C, or at least about 68°C.In an exemplary embodiment, the washing step can be performed at 25° C. in 30 mM NaCl, 3 mM trisodium citrate, and 0.1% SDS. In other exemplary embodiments, the washing step can be performed at 42° C. in 15 mM NaCl, 1.5 mM trisodium citrate, and 0.1% SDS. In another exemplary embodiment, the washing step can be performed at 68° C. in 15 mM NaCl, 1.5 mM trisodium citrate, and 0.1% SDS. Other variations of these conditions will be apparent to those skilled in the art. Hybridization techniques are well known to those skilled in the art and are described, for example, in Benton and Davis (Science 196:180, 1977); Grunstein and Hogness (Proc. Natl. Acad. Sci., USA 72:3961, 1975); Ausubel et al. (Current Protocols in Molecular Biology, Wiley Interscience, New York, 2001); Berger and Kimmel (Guide to Molecular Cloning Techniques, 1987, Academic Press, New York); and Sambrook et al., Molecular Cloning: A Laboratory Manual, Cold Spring Harbor Laboratory Press, New York.

[0448] "Substantially identical" refers to a polypeptide or nucleic acid molecule that shows at least 50% identity with a reference amino acid sequence (e.g., any amino acid sequence described herein) or a nucleic acid sequence (e.g., any nucleic acid sequence described herein). Such a sequence can be at least 60%, 80% or 85%, 90%, 95%, 96%, 97%, 98% or even 99% or higher levels identical to the sequence used for comparison at the amino acid level or nucleic acid level. Sequence identity is usually measured using sequence analysis software (e.g., Sequence Analysis Software Package of the Genetics Computer Group, University of Wisconsin Biotechnology Center, 1710 University Avenue, Madison, Wis. 53705, BLAST, BESTFIT, GAP, or PILEUP / PRETTYBOX programs). Such software matches identical or similar sequences by assigning degrees of homology to various substitutions, deletions and / or other modifications. Conservative substitutions generally include substitutions within the following groups: glycine, alanine; valine, isoleucine, leucine; aspartic acid, glutamic acid, asparagine, glutamine; serine, threonine; lysine, arginine; and phenylalanine, tyrosine. In an exemplary method of determining the degree of identity, the BLAST program can be used, where probability scores between e-3 and em° represent closely related sequences. A "reference" is a comparison standard.

[0449] The term "subject" or "patient" refers to an animal that is the object of treatment, observation or experiment. By way of example only, a subject includes, but is not limited to, a mammal, including, but not limited to, a human or a non-human mammal, such as a non-human primate, mouse, cow, horse, dog, sheep or cat.

[0450] The terms "treatment", "treatment" and the like mean to reduce, prevent or improve a condition and / or symptoms associated therewith (e.g., neoplasia or tumor or infectious agent or autoimmune disease). "Treatment" may refer to administering treatment to a subject after an onset or suspected onset of a disease (e.g., cancer or infection with an infectious agent or autoimmune disease). "Treatment" includes the concept of "alleviation", which refers to reducing the frequency or severity of occurrence or recurrence of any symptom or other adverse effect associated with the disease and / or side effects associated with the treatment. The term "treatment" also encompasses the concept of "management", which refers to reducing the severity of a disease or condition in a patient, for example, prolonging the life span of a patient with the disease or prolonging their survival, or delaying their recurrence, for example, prolonging the remission period of a patient already suffering from the disease. It should be understood that, although not excluded, treating a disease or condition does not require the complete elimination of the condition, condition or symptoms associated therewith.

[0451] As used herein, the terms "prevent," "preventing," and grammatical equivalents thereof refer to avoiding or delaying the onset of symptoms associated with a disease or condition in a subject who has not yet developed such symptoms when administration of an agent or compound is initiated.

[0452] The term "therapeutic effect" refers to a certain degree of alleviation of one or more symptoms of a disease (e.g., neoplasia, infection of a tumor or infectious agent or autoimmune disease) or its related pathology. "Therapeutically effective amount" as used herein refers to a dosage that effectively prolongs the survival of patients with such diseases, alleviates one or more signs or symptoms of the disease, prevents or delays, etc., beyond the expected degree in the absence of such treatments after a single or multiple dose administration to a cell or subject. "Therapeutically effective amount" is intended to limit the amount required for the therapeutic effect. A physician or veterinarian with ordinary skills in the art can easily determine and prescribe the "therapeutically effective amount" (e.g., ED50) of the desired pharmaceutical composition. For example, a physician or veterinarian can start the dosage of the compound of the present invention used in a pharmaceutical composition at a level lower than the level required to obtain the desired therapeutic effect, and gradually increase the dosage until the desired effect is obtained. Disease, condition and disease are used interchangeably herein.

[0453] Those of ordinary skill in the art will recognize that the terms "peptide tag," "affinity tag," "epitope tag," or "affinity receptor tag" are used interchangeably herein. As used herein, the term "affinity receptor tag" refers to an amino acid sequence that allows for easy detection or purification of the tagged protein, for example, by affinity purification. Affinity receptor tags are typically (but not necessarily) placed at or near the N- or C-terminus of the HLA allele. Various peptide tags are well known in the art. Non-limiting examples include polyhistidine tags (e.g., 4 to 15 consecutive His residues (SEQ ID NO: 4), such as 8 consecutive His residues (SEQ ID NO: 5)); polyhistidine-glycine tags; HA tags (e.g., Field et al., Mol. Cell. Biol., 8: 2159, 1988); c-myc tags (e.g., Evans et al., Mol. Cell. Biol., 5: 3610, 1985); herpes simplex virus glycoprotein D (gD) tags (e.g., Paborsky et al., Protein Engineering, 3:547, 1990); FLAG tag (e.g., Hopp et al., BioTechnology, 6:1204, 1988; U.S. Pat. Nos. 4,703,004 and 4,851,341); KT3 epitope tag (e.g., Martine et al., Science, 255:192, 1992); tubulin epitope tag (e.g., Skinner, Biol. Chem., 266:15173, 1991); T7 gene 10 protein peptide tag (e.g., Lutz-Freyemuth et al., P Roc. Natl. Acad. Sci. USA, 87:6393, 1990); streptavidin tag (StrepTagTM or StrepTagIITM; see, e.g., Schmidt et al., J. Mol. Biol., 255(5):753-766, 1996 or U.S. Pat. No. 5,506,121; also commercially available from Sigma-Genosys); or a VSV-G epitope tag derived from the glycoprotein of vesicular stomatitis virus; or a V5 tag derived from a small epitope (Pk) found on the P and V proteins of the simian virus 5 (SV5) paramyxovirus. In some embodiments, the affinity receptor tag is an "epitope tag," a type of peptide tag that adds a recognizable epitope (antibody binding site) to an HLA protein to provide binding of a corresponding antibody, thereby allowing identification or affinity purification of the tagged protein. Non-limiting examples of epitope tags are protein A or protein G, which bind to IgG. In some embodiments, the matrix of IgG Sepharose 6 Fast Flow chromatography resin is covalently coupled to human IgG. This resin allows high flow rates, rapid and convenient purification of proteins labeled with protein A.Many other tag moieties are known and contemplated by the skilled artisan and are contemplated herein.Any peptide tag may be used so long as it is capable of being expressed as a component of an affinity receptor-tagged HLA-peptide complex.

[0454] As used herein, the term "affinity molecule" refers to a molecule or ligand that binds to an affinity receptor peptide with chemical specificity. Chemical specificity is the ability of a protein binding site to bind a specific ligand. The fewer ligands a protein can bind, the higher its specificity. Specificity describes the strength of the binding between a given protein and a ligand. This relationship can be described by the dissociation constant (KD), which characterizes the equilibrium between the bound and unbound states of a protein-ligand system.

[0455] The term "affinity receptor-tagged HLA-peptide complex" refers to a complex comprising an HLA class I or class II associated peptide or a portion thereof that specifically binds to a monoallelic recombinant HLA class I or class II peptide comprising an affinity receptor peptide.

[0456] The terms "specific binding" or "specific binding" when applied to the interaction of affinity molecules and affinity receptor tags or epitopes with HLA peptides means that the interaction is dependent on the presence of a particular structure (e.g., an antigenic determinant or epitope) on the protein; in other words, the affinity molecule recognizes and binds to a specific affinity receptor peptide structure rather than binding to the protein in general.

[0457] As used herein, the term "affinity" refers to a measure of the strength of binding between two members of a binding pair (e.g., an "affinity receptor tag" and an "affinity molecule" and an HLA binding peptide and an HLA class I or class II molecule). KD is the dissociation constant and has units of molar concentration. The affinity constant is the reciprocal of the dissociation constant. The affinity constant is sometimes used as a general term to describe the chemical entity. It is a direct measure of binding energy. Affinity can be determined experimentally by surface plasmon resonance (SPR), for example, using a commercially available Biacore SPR unit. Affinity can also be expressed as an inhibitory concentration 50 (IC50), i.e., the concentration at which 50% of the peptide is displaced. Similarly, lnIC50 refers to the natural logarithm of IC50. K off Refers to the dissociation rate constant, for example, the rate constant for the dissociation of an affinity molecule from an affinity receptor-tagged HLA-peptide complex.

[0458] In some embodiments, the affinity receptor-tagged HLA-peptide complex comprises a biotin acceptor peptide (BAP) and is immunopurified from the complex cell mixture using streptavidin / NeutrAvidin beads. Biotin-avidin / streptavidin binding is the strongest non-covalent interaction known in nature. This property is used in a wide range of applications as a biological tool, such as immunopurification of proteins covalently linked to biotin. In an exemplary embodiment, the nucleic acid sequence encoding the HLA allele uses a biotin acceptor peptide (BAP) as an affinity receptor tag for immunopurification. BAP can be specifically biotinylated at a single lysine residue within the tag in vivo or in vitro (e.g., U.S. Patents 5,723,584; 5,874,239; and 5,932,433; and British Patent GB2370039). BAP is typically 15 amino acids long and contains one lysine as a biotin acceptor residue. In some embodiments, BAP is placed at or near the N- or C-terminus of a monoallelic HLA peptide. In some embodiments, the BAP is placed between the heavy chain domain and the β2 microglobulin domain of an HLA class I peptide. In some embodiments, the BAP is placed between the β chain domain and the α chain domain of an HLA class II peptide. In some embodiments, the BAP is placed in a loop region between the α1, α2, and α3 domains of an HLA class I heavy chain, or between the α1 and α2 and β1 and β2 domains of an HLA class II α chain and β chain, respectively. Figure 2 Exemplary constructs designed for HLA class I and II expression using BAP for biotinylation and immunopurification are described in .

[0459] As used herein, the term "biotin" refers to the compound biotin itself and its analogs, derivatives and variants. Therefore, the term "biotin" includes biotin (cis-hexahydro-2-oxo-1H-thieno[3,4]imidazole-4-pentanoic acid) and any derivatives and analogs thereof, including biotin-like compounds. Such compounds include, for example, amino or sulfhydryl derivatives of biotin-eN-lysine, biocytin hydrazide, 2-iminobiotin and biotinyl-E-aminocaproic acid-N-hydroxysuccinimide ester, sulfosuccinimidyl iminobiotin, biotin bromoacetyl hydrazide, p-diazobenzoyl biocytin, 3-(N-maleimidopropionyl) biocytin, desthiobiotin, etc. The term "biotin" also includes biotin variants that can specifically bind to one or more of Rhizavidin, avidin, streptavidin, tamavidin moieties or other avidin-like peptides.

[0460] As used herein, a "PPV determination method" may refer to a presentation PPV determination method. For example, a "PPV determination method" may refer to a method comprising the following steps: (a) using an HLA peptide presentation prediction model, such as a machine learning HLA peptide presentation prediction model, to process amino acid information of a plurality of test peptide sequences to generate a plurality of test presentation predictions, each test presentation prediction indicating the likelihood that one or more proteins encoded by a class II HLA allele of a cell (such as a class II HLA allele of a subject's cell) are capable of presenting a given test peptide sequence of the plurality of test peptide sequences, wherein the plurality of test peptide sequences comprises at least 500 test peptide sequences, the test peptide sequences comprising (i) at least one hit peptide sequence identified by mass spectrometry as being presented by an HLA protein expressed in the cell, and (ii) at least 499 hit peptide sequences contained in a hit peptide sequence expressed by an organism (e.g., a subject belonging to a class II HLA protein); (a) identifying or determining the top percentage of the plurality of test peptide sequences, such as the top 0.2% of the plurality of test peptide sequences, as presented by the class II HLA alleles of the cell; (c) calculating the PPV of the HLA peptide presentation prediction model, wherein the PPV is the fraction of the plurality of test peptide sequences identified or determined as presented by the class II HLA alleles of the cell, which peptides are peptides observed to be presented by the class II HLA alleles of the cell by mass spectrometry. In some embodiments, the bait peptides have the same length, i.e., contain the same number of amino acids as the hit peptides. In some embodiments, the bait peptides may contain one more or one less amino acid than the hit peptides. In some embodiments, the bait peptides are peptides that are endogenous peptides. In some embodiments, the bait peptide is a synthetic peptide. In some embodiments, the bait peptide is an endogenous peptide that has been identified by mass spectrometry as binding to a first MHC class I or class II protein, wherein the first MHC class I or class II protein is different from a second MHC class I or class II protein that binds to the hit peptide. In some embodiments, the bait peptide can be a scrambled peptide, for example, the bait peptide can comprise an amino acid sequence in which the amino acid positions are rearranged relative to the amino acid positions of the hit peptide within the length of the peptide. In some embodiments, the PPV determination method can be a presentation PPV determination method. In some embodiments, the ratio of the number of hit peptide sequences to the number of bait peptide sequences is about 1:10, 1:20, 1:50, 1:100, 1:250, 1:500, 1:1000, 1:1500, 1:2000, 1:2500, 1:5000, 1:7500, 1:10000, 1:25000,1:50000 or 1:100000. In some embodiments, the at least one hit peptide sequence comprises at least 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, 94, 95, 96, 97, 98, 99, 100, 101, 102, 103, 104, 105, 106, 107, 108, 109, 110, 111, 112, 113, 114, 115, 116, 117, 118, 119, 120, 121, 122, 123, 124, 125, 126, 127, 128, 129, 130, 131, 132, 133, 134, 135, 136, 137, 138, 139, 200, 201, 202, 203, 204, 205, 206, 207, 208, 209, In some embodiments, the at least 499 decoy peptide sequences comprise at least 500, 600, 700, 800, 900, 1000, 1100, 1200, 1300, 1400, 1500, 1600, 1700, 1800, 1900, 2000, 2100, 2200, 2300, 2400, 2500, 2600, 2700, 2800, 2900, 3000, 3100, 3200, 3300, 3400、3500、3600、3700、3800、3900、4000、4100、4200、4300、4400、4500、4600、4700、4800、4900、5000、5100、5200、5300、5400、5500、5600、5700、5800、5900、6000、6100、6200、6300、6400、6500、6600、6 700、6800、6900、7000、7100、7200、7300、7400、7500、7600、7700、7800、7900、8000、8100、8200、8300、8400、8500、8600、8700、8800、8900、9000、9100、9200、9300、9400、9500、9600、9700、9800、9900、10 000、11000、12000、13000、14000、15000、16000、17000、18000、19000、20000、21000、22000、23000、24000、25000、26000、27000、28000、29000、30000、31000、32000、33000、34000、35000、36000、37000、38000、39000、40000、41000、42000、43000、44000、45000、46000、47000、48000、49000、50000、52500、55000、57500、60000、62500、65000、67500、70000、72500、75000、77500、80000、82500、85000、87500、90000、92 700,000, 800,000, 900,000, or 1,000,000 decoy peptide sequences. In some embodiments, the at least 500 test peptide sequences comprise at least 600, 700, 800, 900, 1000, 1100, 1200, 1300, 1400, 1500, 1600, 1700, 1800, 1900, 2000, 2100, 2200, 2300, 2400, 2500, 2600, 2700, 2800, 2900, 3000, 3100, 3200, 3300, 3400、3500、3600、3700、3800、3900、4000、4100、4200、4300、4400、4500、4600、4700、4800、4900、5000、5100、5200、5300、5400、5500、5600、5700、5800、5900、6000、6100、6200、6300、6400、6500、66 00、6700、6800、6900、7000、7100、7200、7300、7400、7500、7600、7700、7800、7900、8000、8100、8200、8300、8400、8500、8600、8700、8800、8900、9000、9100、9200、9300、9400、9500、9600、9700、9800 、9900、10000、11000、12000、13000、14000、15000、16000、17000、18000、19000、20000、21000、22000、23000、24000、25000、26000、27000、28000、29000、30000、31000、32000、33000、34000、35000、36000、37000、38000、39000、40000、41000、42000、43000、44000、45000、46000、47000、48000、49000、50000、52500、55000、57500、60000、62500、65000、67500、70000、72500、75000、77500、80000、82500、85000、87500、90 0000, 100000, 125000, 150000, 175000, 200000, 225000, 250000, 275000, 300000, 325000, 350000, 375000, 400000, 425000, 450000, 475000, 500000, 600000, 700000, 800000, 900000 or 1000000 test peptide sequences. In some embodiments, identifying or determining a top percentage of the plurality of test peptide sequences as presented by the class II HLA alleles of the cell comprises identifying or determining a top 0.20%, 0.30%, 0.40%, 0.50%, 0.60%, 0.70%, 0.80%, 0.90%, 1.00%, 1.10%, 1.20%, 1.30%, 1.40%, 1.50%, 1.60%, 1.70%, 1.80%, 1.90%, 1.01%, 1.12%, 1.13%, 1.14%, 1.15%, 1.16%, 1.17%, 1.18%, 1.19% or more of the plurality of test peptide sequences as presented by the class II HLA alleles of the cell. 0%, 2.00%, 2.10%, 2.20%, 2.30%, 2.40%, 2.50%, 2.60%, 2.70%, 2.80%, 2.90%, 3.00%, 3.10%, 3.20%, 3.30%, 3.40%, 3.50%, 3.60%, 3.70%, 3.80%, 3.90%, 4.00%, 4.10%, 4.20%, 4.30%, 4.40%, 4.50%, 4.60%, 4.70%, 4.80%, 4.90%, 5.00%, 5.10%, 5.20%, 5.30%, 5.40%, 5.50%, 5.60%, 5.70%, 5.80%, 5.90%, 6.00%, 6.10%, 6.20%, 6.30%, 6.40%, 6.50%, 6.60%, 6.70%, 6.80%, 6.90%, 7.00%, 7.10%, 7.2 0%, 7.30%, 7.40%, 7.50%, 7.60%, 7.70%, 7.80%, 7.90%, 8.00%, 8.10%, 8.20%, 8.30%, 8.40%, 8.50%, 8.60%, 8.70%, 8.80%, 8.90%, 9.00%, 9.10%, 9.20%, 9.30%, 9.40%, 9.50%, 9.60%, 9.70%, 9.80%,9. 90%, 10%, 11%, 12%, 13%, 14%, 15%, 16%, 17%, 18%, 19% or 20% are identified or determined to be presented by a class II HLA allele of the cell. In some embodiments, the cell is a monoallelic cell.

[0461] As used herein, a "PPV determination method" may refer to a binding PPV determination method. For example, a "PPV determination method" may refer to a method comprising the following steps: (a) using an HLA peptide binding prediction model, such as a machine learning HLA peptide binding prediction model, to process amino acid information of a plurality of test peptide sequences to generate a plurality of binding presentation predictions, each binding presentation prediction indicating the likelihood that one or more proteins encoded by a class II HLA allele of a cell (such as a class II HLA allele of a subject's cell) will bind to a given test peptide sequence of the plurality of test peptide sequences, wherein the plurality of test peptide sequences comprises at least 20 test peptide sequences, the test peptide sequences comprising (i) at least one hit peptide sequence identified by mass spectrometry as being presented by an HLA protein expressed in a cell, and (ii) at least 19 decoy peptide sequences contained within a protein, the protein (a) comprising at least one peptide sequence identified by mass spectrometry as being presented by an HLA protein expressed in a cell, wherein the ratio of the number of hit peptide sequences to the number of bait peptide sequences in the multiple test peptide sequences is less than 1, for example, the ratio of the at least one hit peptide sequence to the at least 19 bait peptide sequences is 1:19; (b) identifying or determining a top percentage of the multiple test peptide sequences, such as the top 5% of the multiple test peptide sequences, as binding to the HLA protein; and (c) calculating the PPV of the HLA peptide binding prediction model, wherein the PPV is the fraction of the test peptide sequences in the multiple test peptide sequences identified or determined as binding to the class II HLA alleles of the cell, which peptides are peptides observed to be presented by the class II HLA alleles of the cell by mass spectrometry. In some embodiments, the ratio of the number of hit peptide sequences to the number of decoy peptide sequences is about 1:2, 1:3, 1:4, 1:5, 1:10, 1:20, 1:25, 1:30, 1:40, 1:50, 1:75, 1:100, 1:200, 1:250, 1:500, or 1:1000. In some embodiments, the at least one hit peptide sequence comprises at least 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 ,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 or 100 hit peptide sequences. In some embodiments, the at least 19 decoy peptide sequences comprise at least 30, 40, 50, 60, 70, 80, 90, 100, 110, 120, 130, 140, 150, 160, 170, 180, 190, 200, 210, 220, 230, 240, 250, 260, 270, 280, 290, 300, 310, 320, 330, 340, 350, 360, 370, 380, 390, 400, 410, 420, 430, 440, 450, 460, 470, 480, 490, 500, 600, 700, 800, 900, 1000, 1100, 1200, 1300, 1400 400、1500、1600、1700、1800、1900、2000、2100、2200、2300、2400、2500、2600、2700、2800、2900、3000、3100、3200、3300、3400、3500、3600、3700、3800 , 3900, 4000, 4100, 4200, 4300, 4400, 4500, 4600, 4700, 4800, 4900, 5000, 5100, 5200, 5300, 5400, 5500, 5600, 5700, 5800, 5900, 6000, 6100, 6200, 630 0, 6400, 6500, 6600, 6700, 6800, 6900, 7000, 7100, 7200, 7300, 7400, 7500, 7600, 7700, 7800, 7900, 8000, 8100, 8200, 8300, 8400, 8500, 8600, 8700, 8 800、8900、9000、9100、9200、9300、9400、9500、9600、9700、9800、9900、10000、11000、12000、13000、14000、15000、16000、17000、18000、19000、20000 , 21000, 22000, 23000, 24000, 25000, 26000, 27000, 28000, 29000, 30000, 31000, 32000, 33000, 34000, 35000, 36000, 37000, 38000, 39000, 40000, 41 000、42000、43000、44000、45000、46000、47000、48000、49000、50000、52500、55000、57500、60000、62500、65000、67500、70000、72500、75000、77500、425000, 450000, 475000, 500000, 600000, 700000, 800000, 90000 or 1000000 decoy peptide sequences. In some embodiments, the at least 20 test peptide sequences comprise at least 30, 40, 50, 60, 70, 80, 90, 100, 110, 120, 130, 140, 150, 160, 170, 180, 190, 200, 210, 220, 230, 240, 250, 260, 270, 280, 290, 300, 310, 320, 330, 340, 350, 360, 370, 380, 390, 400, 410, 420, 430, 440, 450, 460, 470 , 480, 490, 500, 600, 700, 800, 900, 1000, 1100, 1200, 1300, 1400, 1500, 1600, 1700, 1800, 1900, 2000, 2100, 2200, 2300, 2400, 2500, 2600, 2700, 2800, 2900, 3000, 3100, 3200, 3300, 3400, 3500, 3600, 3700, 3800, 3900, 4000, 4100, 4200, 4300 , 4400, 4500, 4600, 4700, 4800, 4900, 5000, 5100, 5200, 5300, 5400, 5500, 5600, 5700, 5800, 5900, 6000, 6100, 6200, 6300, 6400, 6500, 6600, 6700, 6800, 6900, 7000, 7100, 7200, 7300, 7400, 7500, 7600, 7700, 7800, 7900, 8000, 8100, 8200, 8 300、8400、8500、8600、8700、8800、8900、9000、9100、9200、9300、9400、9500、9600、9700、9800、9900、10000、11000、12000、13000、14000、15000、16000、17000、18000、19000、20000、21000、22000、23000、24000、25000、26000、27000、28000、29000、30000、31000、32000、33000、34000、35000、36000、37000、38000、39000、40000、41000、42000、43000、44000、45000、46000、47000、48000、49000、50000、52500、55000、57500、60000、62500、65000、67500、70000、72500、75000、77500、80000 , 82500, 85000, 87500, 90000, 92500, 95000, 97500, 100000, 125000, 150000, 175000, 200000, 225000, 250000, 275000, 300000, 325000, 350000, 375000, 400000, 425000, 450000, 475000, 500000, 600000, 700000, 800000, 900000 or 1000000 test peptide sequences. In some embodiments, identifying or determining a top percentage of the plurality of test peptide sequences as presented by a class II HLA allele of the cell comprises identifying or determining the top 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%, or 40% as presented by a class II HLA allele of the cell. In some embodiments, the cell is a monoallelic cell.

[0462] Human Leukocyte Antigen (HLA) System

[0463] The immune system can be classified into two functional subsystems: the innate immune system and the adaptive immune system. The innate immune system is the first line of defense against infection, and most potential pathogens are rapidly neutralized by this system before they can cause, for example, an overt infection. The adaptive immune system reacts to molecular structures of invading organisms, known as antigens. Unlike the innate immune system, the adaptive immune system is highly specific to pathogens. Adaptive immunity can also provide long-lasting protection; for example, a person who has recovered from measles is now protected from measles for life. There are two types of adaptive immune responses, including humoral and cell-mediated. In humoral immune responses, antibodies secreted into body fluids by B cells bind to antigens derived from pathogens, resulting in the elimination of the pathogens through a variety of mechanisms, such as complement-mediated lysis. In cell-mediated immune responses, T cells that are capable of destroying other cells are activated. For example, if disease-associated proteins are present in cells, they are proteolytically broken into peptides within the cell. Specific cellular proteins then attach themselves to the antigens or peptides formed in this way and transport them to the cell surface, where they are presented to the molecular defense mechanisms in the body's T cells. Cytotoxic T cells recognize these antigens and kill cells bearing the antigen.

[0464] The term "major histocompatibility complex (MHC)", "MHC molecule" or "MHC protein" refers to a protein that is capable of binding peptides generated by proteolytic cleavage of a protein antigen and representing a potential T-cell epitope, transporting it to the cell surface, and presenting the peptides to specific cells, such as in cytotoxic T lymphocytes or T helper cells. The human MHC is also known as the HLA complex. Thus, the term "human leukocyte antigen (HLA) system", "HLA molecule" or "HLA protein" refers to a gene complex encoding a human MHC protein. The term MHC is referred to as the "H-2" complex in murine species. One of ordinary skill in the art will recognize that the terms "major histocompatibility complex (MHC)", "MHC molecule", "MHC protein" and "human leukocyte antigen (HLA) system", "HLA molecule", "HLA protein" are used interchangeably herein.

[0465] HLA proteins are classified into two types, known as HLA class I and HLA class II. The structures of the proteins of both HLA classes are very similar; however, they have very different functions. HLA class I proteins are found on the surface of almost all cells in the body, including most tumor cells. HLA class I proteins are loaded with antigens, which usually originate from endogenous proteins or pathogens present within the cell, and are then presented to naive or cytotoxic T lymphocytes (CTLs). HLA class II proteins are found on antigen presenting cells (APCs), including but not limited to dendritic cells, B cells, and macrophages. They primarily present peptides processed from external antigen sources, such as outside the cell, to helper T cells. Most peptides bound by HLA class I proteins originate from cytoplasmic proteins produced in the organism's own healthy host cells and generally do not stimulate an immune response.

[0466] HLA class I molecules (Figure 1) consist of two non-covalently linked polypeptide chains: an HLA-encoded α chain (heavy chain, 44 to 47 kD) and a non-HLA-encoded subunit, called β2-microglobulin (or β2m), (12 kD). The α chain has three extracellular domains, α1, α2 and α3, and a transmembrane region, wherein the α1 and α2 regions are capable of binding peptides of about 7 to 13 amino acids (e.g., about 8 to 11 amino acids, or 9 or 10 amino acids). HLA class 1 molecules bind to peptides with appropriate binding motifs and present them to cytotoxic T lymphocytes. HLA class 1 heavy chains can be the protein product of the HLA-A allele, also known as the HLA-A monomer, or the protein product of the HLA-B allele (again, the HLA-B monomer) or the protein product of the HLA-C allele (HLA-C monomer), each of which is complexed with β-2-microglobulin. α1 is dependent on the non-HLA protein β2m; β2m is encoded by the β-2-microglobulin gene located on chromosome 15 in humans. The α3 domain is linked to the transmembrane region, anchoring the HLA class I molecule to the cell membrane. The presented peptide is held in the central region of the α1 / α2 heterodimer (a molecule composed of two different subunits) by the bottom of the peptide-binding groove. HLA class IA, HLA class IB, or HLA class IC are highly polymorphic. The HLA class 1-A gene (referred to as the HLA-A gene), the HLA class 1-B gene (referred to as the HLA-B gene), and the HLA class 1-C gene (referred to as the HLA-C gene) each contain 8 exons, with exon 1 encoding the leader peptide, exons 2 and 3 encoding the α1 and α2 domains, exon 5 encoding the transmembrane region, and exons 6 and 7 encoding the cytoplasmic tail. Polymorphisms in exons 2 and 3 determine the peptide binding specificity of each class 1 molecule. There are many possible variations, expression patterns, and antigens presented by the HLA class I-B genes (HLA-B). This group is subdivided into those encoded within the HLA locus, such as HLA-E, HLA-F, HLA-G, and those not encoded therein, such as stress ligands, such as ULBP, Rae1, and H60. The antigens / ligands for many of these molecules remain unknown, but they can interact with each of CD8+ T cells, NKT cells, and NK cells.

[0467] In some embodiments, the present disclosure utilizes non-classical HLA class IE alleles. HLA-E molecules are recognized by natural killer (NK) cells and CD8+ T cells. HLA-E is expressed in almost all tissues, including lung, liver, skin and placental cells. HLA-E expression has also been detected in solid tumors (e.g., osteosarcoma and melanoma). HLA-E molecules bind to TCR expressed on CD8+ T cells, thereby causing T cell activation. HLA-E is also known to bind to CD94 / NKG2 receptors expressed on NK cells and CD8+ T cells. CD94 can be paired with several different isoforms of NKG2 to form receptors with the potential to inhibit (NKG2A, NKG2B) or promote (NKG2C) cell activation. HLA-E can bind to peptides derived from amino acid residues 3-11 of the leader sequences of most HLA-A, -B, -C and -G molecules, but cannot bind to its own leader peptide. It has also been shown that HLA-E presents peptides derived from endogenous proteins similar to HLA-A, -B and -C alleles. Under physiological conditions, CD94 / NKG2A and the engagement of HLA-E loaded with peptides from HLA class I leader sequences usually induce inhibitory signals. Cytomegalovirus (CMV) utilizes the mechanism of escaping NK cell immune surveillance by expressing UL40 glycoprotein (simulating HLA-A leader sequence). However, it has also been reported that CD8+ T cells can identify the HLA-E loaded with the UL40 peptide derived from CMV Toledo strain, and work in defense CMV. A large number of studies have revealed several important functions of HLA-E in infectious diseases and cancer.

[0468] Peptide antigens attach themselves to HLA class I molecules by competitive affinity binding within the endoplasmic reticulum before presentation on the cell surface. Here, the affinity of an individual peptide antigen is directly related to its amino acid sequence and the presence of specific binding motifs at defined positions within the amino acid sequence. If the sequence of such a peptide is known, it is possible to manipulate the immune system against diseased cells using, for example, peptide vaccines.

[0469] MHC molecules are highly polymorphic, that is, there are many MHC variants. Each variant is encoded by a variant of the gene encoding the protein, and each such variant gene is called an allele. For humans, MHC is called human leukocyte antigen (HLA), which involves three types of HLA class II molecules: DP, DQ and DR. HLA class II peptides (Figure 1) have two chains: α and β, each of which has two domains-α1 and α2 and β1 and β2-each chain has a transmembrane domain: α2 and β2, anchoring HLA class II molecules to the cell membrane. The peptide binding groove is formed by heterodimers of α1 and β1. The most widely studied HLA-DR molecules have DRA and DRB, corresponding to the α and β domains, respectively. DRB is diverse, and DRA is almost the same. Therefore, the binding specificity of the DRB allele indicates the binding specificity of the corresponding HLA-DR. Each MHC protein has its own binding specificity, which means that a set of peptides bound to an MHC molecule may be different from the peptides bound to another MHC molecule. Classical molecules present peptides to CD4+ lymphocytes. Nonclassical molecules with intracellular functions, annexins, are not exposed on the cell membrane but in the inner membrane of lysosomes and usually load antigenic peptides onto classical HLA class II molecules.

[0470] In the HLA class II system, phagocytic cells such as macrophages and immature dendritic cells take up entities by phagocytosis into phagosomes - although B cells show more common endocytosis into endosomes - which fuse with lysosomes, whose acidic enzymes cleave the ingested proteins into many different peptides. Autophagy is another source of HLA class II peptides. Through physicochemical dynamics of molecular interactions with HLA class II variants carried by the host (encoded in the host genome), specific peptides become immunodominant and loaded onto HLA class II molecules. They are transported to and externalized on the cell surface. The most studied subclasses of HLA class II genes are: HLA-DPA1, HLA-DPB1, HLA-DQA1, HLA-DQB1, HLA-DRA, and HLA-DRB1.

[0471] Presentation of peptides by HLA class II molecules to CD4+ helper T cells is required for immune responses to foreign antigens (Roche and Furuta, 2015). Once activated, CD4+ T cells promote B cell differentiation and antibody production, as well as CD8+ T cell (CTL) responses. CD4+ T cells also secrete cytokines and chemokines that activate and induce differentiation of other immune cells. HLA class II molecules are heterodimers of α and β chains that interact to form a peptide binding groove that is more open than the HLA class I peptide binding groove (Unanue et al., 2016). Peptides that bind to HLA class II molecules are thought to have a 9-amino acid binding core with flanking residues protruding from the binding groove on the N-terminal or C-terminal side (Jardetzky et al., 1996; Stern et al., 1994). These peptides are typically 12-16 amino acids in length and usually contain 3-4 anchor residues at positions P1, P4, P6 / 7, and P9 of the binding moiety (Rossjohn et al., 2015).

[0472] HLA alleles are expressed in a codominant manner, which means that the alleles (variants) inherited from both parents are expressed equally. For example, each person carries 2 alleles of each gene in 3 class I genes (HLA-A, HLA-B and HLA-C), so six different types of HLA class II can be expressed. In the HLA class II locus, each person inherits a pair of HLA-DP genes (DPA1 and DPB1, encoding α and β chains), HLA-DQ (DQA1 and DQB1 for α and β chains), an HLA-DRα gene (DRA1) and one or more HLA-DRβ genes (DRB1 and DRB3, -4 or -5). For example, HLA-DRB 1 has more than nearly 400 known alleles. This means that a heterozygous individual can inherit six or eight functional HLA class II alleles, three or more per parent. Therefore, HLA genes are highly polymorphic; there are many different alleles in different individuals within a population. The genes encoding the HLA proteins have many possible variations, allowing each person's immune system to respond to a wide variety of foreign invaders. Some HLA genes have hundreds of identified forms (alleles), each of which is given a specific number. In some embodiments, the HLA class I alleles are HLA-A*02:01, HLA-B*14:02, HLA-A*23:01, HLA-E*01:01 (non-classical). In some embodiments, the HLA class II alleles are HLA-DRB*01:01, HLA-DRB*01:02, HLA-DRB*11:01, HLA-DRB*15:01, and HLA-DRB*07:01.

[0473] The subject's subject-specific HLA allele or HLA genotype can be determined by any method known in the art. In an exemplary embodiment, the HLA genotype is determined by any method described in International Patent Application PCT / US2014 / 068746 (published as WO2015085147 on June 11, 2015), which is incorporated herein by reference in its entirety. Briefly, the method includes determining a polymorphic gene type, which may include generating an alignment of reads extracted from a sequencing data set with a gene reference set comprising allelic variants of the polymorphic gene, determining a first posterior probability or a score derived from the posterior probability for each allelic variant in the alignment, identifying the allelic variant with the largest first posterior probability or a score derived from the posterior probability as a first allelic variant, identifying one or more overlapping reads aligned to the first allelic variant and one or more other allelic variants, determining a second posterior probability or a score derived from the posterior probability for the one or more other allelic variants using a weighting factor, determining a second allelic variant by selecting the allelic variant with the largest second posterior probability or a score derived from the posterior probability, the first and second allelic variants defining the gene type of the polymorphic gene, and providing an output of the first and second allelic variants.

[0474] In some embodiments, MHC class II peptides described herein: antigen peptides combine and present prediction methods have the ability of predicting binding substances from the large group library MHC class II peptides encoded by independent HLA alleles.In some embodiments, the large database of HLA matching peptides verified by mass spectrometry of MAPTAC technology is trained.In some embodiments, the large database of HLA matching peptides verified by mass spectrometry comprises more than 1.2x10^6 such HLA matching peptides.In some embodiments, the large database of HLA matching peptides verified by mass spectrometry covers more than 150 HLA alleles, including MHC class I and class II allele hypotypes.In some embodiments, the database covers at least 95% of the U.S. population for HLA-I and HLA-II (DR hypotype).

[0475] As described in this article, there is ample evidence in animals and humans that mutated epitopes can effectively induce immune responses, that spontaneous tumor regression or long-term survival is associated with CD8+ T cell responses to mutated epitopes, and that “immunoediting” can track changes in the expression of dominant mutant antigens in mice and humans.

[0476] Sequencing techniques reveal that each tumor contains multiple patient-specific mutations that alter the protein-coding content of genes. Such mutations produce altered proteins ranging from single amino acid changes (caused by missense mutations) to long regions that add new amino acid sequences due to frameshifts, readthrough of stop codons, or translation of intronic regions (new open reading frame mutations; neoORFs). These mutant proteins are valuable targets for the host's immune response to tumors because, unlike natural proteins, they are not affected by the immunosuppressive effects of self-tolerance. Therefore, mutated proteins are more likely to be immunogenic and more specific to tumor cells than to normal cells of the patient. In essence, short peptides (8-24 amino acids long) containing cancer-associated mutations are candidates for cancer immunotherapy.

[0477] In some embodiments, the algorithm of the driver prediction method can be further used to determine the mutation of the peptide. In some embodiments, the prediction method can be used to determine the driver mutation status, and / or RNA expression status, and / or cleavage prediction within the peptide.

[0478] The term "T cell" includes CD4+ T cells and CD8+ T cells. The term T cell also includes T helper type 1 T cells and T helper type 2 T cells. As used herein, T cells are generally classified into two major categories: helper T (TH) cells and cytotoxic T lymphocytes (CTLs) by function and cell surface antigens (cluster differentiation antigens or CDs) that also help the T cell receptor bind to the antigen.

[0479] Mature helper T (TH) cells express surface protein CD4 and are called CD4+ T cells. After T cell development, mature naive T cells leave the thymus and begin to spread throughout the body, including lymph nodes. Naive T cells are T cells that have never been exposed to the antigens to which they are programmed to respond. Like all T cells, they express T cell receptor-CD3 complexes. T cell receptor (TCR) consists of constant region and variable region. The variable region determines what antigens the T cell can respond to. CD4+ T cells have TCRs with affinity for MHC class II proteins, and CD4 participates in determining MHC affinity during thymus maturation. MHC class II proteins are usually only found on the surface of specialized antigen presenting cells (APCs). Specialized antigen presenting cells (APCs) are mainly dendritic cells, macrophages, and B cells, although dendritic cells are the only group of cells that constitutively (always) express MHC class II. Some APCs also bind native (or unprocessed) antigens to their surface, such as follicular dendritic cells, but unprocessed antigens do not interact with T cells and do not participate in their activation. Peptide antigens bound to HLA class I proteins are generally shorter than those bound to HLA class II proteins.

[0480] Cytotoxic T lymphocytes (CTL), also known as cytotoxic T cells, cytolytic T cells, CD8+ T cells or killer T cells, refer to lymphocytes that induce apoptosis in targeted cells. CTL forms antigen-specific conjugates with target cells through the interaction of TCR and antigens (Ag) processed on the surface of target cells, thereby causing apoptosis of target cells. Apoptotic bodies are eliminated by macrophages. The term "CTL response" is used to refer to the primary immune response mediated by CTL cells. Cytotoxic T lymphocytes have both T cell receptors (TCR) and CD8 molecules on their surfaces. T cell receptors can recognize and bind peptides complexed with HLA class I molecules. Each cytotoxic T lymphocyte expresses a unique T cell receptor that can bind to a specific MHC / peptide complex. Most cytotoxic T cells express T cell receptors (TCRs) that can recognize specific antigens. In order for TCR to bind to HLA class I molecules, the former must be accompanied by a glycoprotein called CD8, which binds to the constant part of HLA class I molecules. Therefore, these T cells are called CD8+ T cells. The affinity between CD8 and MHC molecules brings T cells and target cells together during antigen-specific activation. Once activated, CD8+ T cells are identified as T cells and are generally classified as having a predetermined cytotoxic role in the immune system. However, CD8+ T cells also have the ability to produce certain cytokines.

[0481] "T cell receptor (TCR)" is a cell surface receptor involved in the activation of T cells in response to antigen presentation. TCR is usually composed of two chains, α and β, which are assembled to form a heterodimer and associate with the CD3 transduction subunit to form a T cell receptor complex present on the cell surface. Each α and β chain of TCR consists of immunoglobulin-like N-terminal variable (V) and constant (C) regions, hydrophobic transmembrane domains, and short cytoplasmic regions. As for immunoglobulin molecules, the variable regions of α and β chains are produced by V(D)J recombination, thereby producing a great diversity of antigen specificity in T cell populations. However, compared with immunoglobulins that recognize complete antigens, T cells are activated by processed peptide fragments associated with MHC molecules, thereby introducing an additional dimension to T cell recognition of antigens, which is called MHC restriction. Recognition of MHC differences between donors and recipients by T cell receptors can lead to T cell proliferation and potential development of GVHD. It has been shown that normal surface expression of the TCR depends on the coordinated synthesis and assembly of all seven components of the complex (Ashwell and Klusner 1990). Inactivation of TCRα or TCRβ can lead to the elimination of the TCR from the T cell surface, thereby preventing recognition of alloantigens and thus preventing GVHD. However, TCR disruption usually leads to the elimination of CD3 signaling components and alters the pattern of further T cell expansion.

[0482] The term "HLA peptide group" refers to a group of peptides that specifically interact with a particular HLA class and can contain thousands of different sequences. The HLA peptide group includes diverse peptides that are derived from normal and abnormal proteins expressed in cells. Therefore, the HLA peptide group can be studied to identify cancer-specific peptides for the development of tumor immunotherapy and as a source of information about protein synthesis and degradation programs within cancer cells. In some embodiments, the HLA peptide group is a group of soluble HLA peptides (sHLA). In some embodiments, the HLA peptide group is a group of membrane-bound HLA (mHLA).

[0483] "Antigen presenting cells" or "APCs" include professional antigen presenting cells (e.g., B lymphocytes, macrophages, monocytes, dendritic cells, Langerhans cells), as well as other antigen presenting cells (e.g., keratinocytes, endothelial cells, astrocytes, fibroblasts, oligodendrocytes, thymic epithelial cells, thyroid epithelial cells, glial cells (brain), pancreatic β cells, and vascular endothelial cells). "Antigen presenting cells" or "APCs" are cells that express major histocompatibility complex (MHC) molecules and can display foreign antigens complexed with MHC on their surface.

[0484] Monoallelic HLA cell lines

[0485] Monoallelic cell lines expressing a single HLA class I allele, a pair of HLA class II alleles, or a single HLA class I allele and a pair of HLA class II alleles can be generated by transducing or transfecting a suitable cell population with a polynucleic acid (e.g., a vector) encoding a single HLA allele. Figure 2). Suitable cell populations include, for example, HLA class I defective cell lines in which a single HLA class I allele is exogenously expressed, HLA class II defective cell lines in which a pair of HLA class II alleles are exogenously expressed, or class I and class I type II defective cell lines in which a single HLA class I and / or a pair of class II alleles are exogenously expressed. As an exemplary embodiment, the HLA class I defective B cell line is B721.221. However, it is clear to the skilled person that other cell populations with HLA class I and / or HLA class II defects can be generated. Exemplary methods for deleting / inactivating endogenous HLA class I or HLA class II genes include CRISPR-Cas9-mediated genome editing, for example in THP-1 cells. In some embodiments, the cell population is a professional antigen presenting cell, such as a macrophage, a B cell, and a dendritic cell. The cell may be a B cell or a dendritic cell. In some embodiments, the cell is a tumor cell or a cell from a tumor cell line. In some embodiments, the cell is isolated from a patient. In some embodiments, the cells contain an infectious agent or a portion thereof. In some embodiments, the cell population comprises at least 10 7 Cells. In some embodiments, the cell colony is further modified, for example, by increasing or decreasing the expression and / or activity of at least one gene. In some embodiments, the gene encodes a member of an immunoproteasome. It is known that the immunoproteasome is involved in the processing of HLA class I binding peptides and includes LMP2 (β1i), MECL-1 (β2i) and LMP7 (β5i) subunits. The immunoproteasome can also be induced by interferon-γ. Therefore, in some embodiments, the cell colony can be contacted with one or more cytokines, growth factors or other proteins. The cells can be stimulated with inflammatory cytokines such as interferon-γ, IL-10, IL-6 and / or TNF-α. The cell colony can also be subjected to various environmental conditions, such as stress (heat stress, hypoxia, glucose starvation, DNA damaging agents, etc.). In some embodiments, cells are contacted with one or more of chemotherapeutic drugs, radiotherapy, targeted therapy or immunotherapy. Therefore, the methods disclosed herein can be used to study the effects of various genes or conditions on HLA peptide processing and presentation. In some embodiments, the conditions used are selected to match the conditions of the patient whose HLA-peptide population is to be identified.

[0486] A virus-based system (e.g., an adenovirus system, an adeno-associated virus (AAV) vector, a poxvirus, or a lentivirus) can be used to encode and express a single HLA allele of the present disclosure. Plasmids that can be used for adeno-associated virus, adenovirus, and lentivirus delivery have been previously described (see, e.g., U.S. Patents 6,955,808 and 6,943,019, and U.S. Patent Application 20080254008, incorporated herein by reference). In vectors that can be used in the practice of the present disclosure, integration into the host genome of the cell can be achieved using retroviral gene transfer methods, typically resulting in long-term expression of the inserted transgene. In an exemplary embodiment, the retrovirus is a lentivirus. In addition, high transduction efficiencies have been observed in many different cell types and target tissues. The tropism of retroviruses can be changed by incorporating foreign envelope proteins and expanding the potential target population of target cells. Retroviruses can also be engineered to allow conditional expression of the inserted transgene so that only certain cell types are infected by the lentivirus. Cell type-specific promoters can be used to target expression in specific cell types. Lentiviral vectors are retroviral vectors (thus both lentiviral and retroviral vectors can be used in the practice of the present disclosure). In addition, lentiviral vectors are able to transduce or infect non-dividing cells and typically produce high viral titers.

[0487] The choice of retroviral gene transfer system can depend on the target tissue. Retroviral vectors consist of cis-acting long terminal repeats with a packaging capacity of up to 6-10 kb of foreign sequence. Minimal cis-acting LTRs are sufficient to replicate and package the vector, which is then used to integrate the desired nucleic acid into the target cell to provide permanent expression. Widely used retroviral vectors that can be used in the practice of the present disclosure include vectors based on murine leukemia virus (MuLV), gibbon ape leukemia virus (GaLV), simian immunodeficiency virus (SIV), human immunodeficiency virus (HIV), and combinations thereof (see, e.g., Buchscher et al. (1992) J. Virol. 66:2731-2739; Johann et al. (1992) J. Virol. 66:1635-1640; Sommnerfelt et al. (1990) Virol. 176:58-59; Wilson et al. (1998) J. Virol. 63:2374-2378; Miller et al. (1991) J. Virol. 65:2220-2224; PCT / US94 / 05700). Additionally, useful in the practice of the present disclosure are minimal non-primate lentiviral vectors, such as lentiviral vectors based on equine infectious anemia virus (EIAV) (see, e.g., Balagaan, (2006) J Gene Med; 8:275-285, published online on November 21, 2005 in Wiley InterScience DOI: 10.1002 / jgm.845). The vector may have a cytomegalovirus (CMV) promoter driving target gene expression. Thus, the present disclosure relates to one or more vectors that can be used to practice the present disclosure: viral vectors, including retroviral vectors and lentiviral vectors.

[0488] Any HLA allele can be expressed in a cell population. In an exemplary embodiment, the HLA allele is an HLA class I allele. In some embodiments, the HLA class I allele is an HLA-A allele or an HLA-B allele. In some embodiments, the HLA allele is an HLA class II allele. The sequences of HLA class I and class II alleles can be found in the IPD-IMGT / HLA database. Exemplary HLA alleles include, but are not limited to, HLA-A*02:01, HLA-B*14:02, HLA-A*23:01, HLA-E*01:01, HLA-DRB*01:01, HLA-DRB*01:02, HLA-DRB*11:01, HLA-DRB*15:01, and HLA-DRB*07:01.

[0489] In some embodiments, the HLA allele is selected to correspond to the target genotype. In some embodiments, the HLA allele is a mutated HLA allele, which can be a non-naturally occurring allele or a naturally occurring allele in a sick patient. The methods disclosed herein have the further advantage of identifying HLA binding peptides for HLA alleles associated with various diseases and alleles present at low frequencies. Therefore, in some embodiments, the methods provided herein can identify HLA alleles, even if they are present in a population such as a Caucasian population at a frequency of less than 1%.

[0490] In some embodiments, the nucleic acid sequence encoding the HLA allele further comprises an affinity receptor tag that can be used to immunopurify the HLA protein. Suitable tags are well known in the art. In some embodiments, the affinity receptor tag is a polyhistidine tag, a polyhistidine-glycine tag, a polyarginine tag, a polyaspartic acid tag, a polycysteine ​​tag, a polyphenylalanine, a c-myc tag, a herpes simplex virus glycoprotein D (gD) tag, a FLAG tag, a KT3 epitope tag, a tubulin epitope tag, a T7 gene 10 protein peptide tag, a streptavidin tag, a streptavidin binding peptide (SPB) tag, a Strep-tag, a Strep-tag II, an albumin binding protein (ABP) tag, an alkaline phosphatase (AP) tag, a blue tongue virus tag (B-tag), a calmodulin binding peptide (CBP) tag, a chloramphenicol acetyltransferase (CAT) tag, a choline Binding domain (CBD) tag, chitin binding domain (CBD) tag, cellulose binding domain (CBP) tag, dihydrofolate reductase (DHFR) tag, galactose binding protein (GBP) tag, maltose binding protein (MBP), glutathione-S-transferase (GST), Glu-Glu (EE) tag, human influenza hemagglutinin (HA) tag, horseradish peroxidase (HRP) tag, NE-tag, HSV tag, ketosteroid isomerase (KSI) tag, KT3 tag, LacZ tag, luciferase tag, NusA tag, PDZ domain tag, AviTag, calmodulin tag, E-tag, S-tag, SBP-tag, Softag 1, Softag 3, TC tag, VSV-tag, Xpress tag, Isopeptag, SpyTag, SnoopTag, Profinity eXact tag, Protein C tag, S1-tag, S-tag, biotin-carboxyl carrier protein (BCCP) tag, green fluorescent protein (GFP) tag, small ubiquitin-like modifier (SUMO) tag, tandem affinity purification (TAP) tag, HaloTag, Nus-tag, thioredoxin tag, Fc-tag, CYD tag, HPC tag, TrpE tag, ubiquitin tag, VSV-G epitope tag derived from vesicular stomatitis virus glycoprotein, or V5 tag derived from a small epitope (Pk) found on the P and V proteins of simian virus 5 (SV5) paramyxovirus. In some embodiments, the affinity receptor tag is an "epitope tag", which is a type of peptide tag that adds a recognizable epitope (antibody binding site) to the HLA protein to provide binding of the corresponding antibody, thereby allowing identification or affinity purification of the tagged protein. Non-limiting examples of epitope tags are protein A or protein G, which bind to IgG. In some embodiments, the affinity receptor tag comprises a biotin acceptor peptide (BAP) or a human influenza hemagglutinin (HA) peptide sequence.Many other tag moieties are known and contemplated by the skilled artisan and are contemplated herein.Any peptide tag may be used so long as it is capable of being expressed as a component of an affinity receptor-tagged HLA-peptide complex.

[0491] The methods provided herein include isolating HLA-peptide complexes from cells transfected or transduced with affinity pull-down of HLA constructs ( Figure 3 ). In some embodiments, the complex can be isolated using standard immunoprecipitation techniques known in the art with commercially available antibodies. The cells can be lysed first. HLA class I-peptide complexes can be isolated using HLA class I-specific antibodies such as the W6 / 32 antibody, while HLA class II-peptide complexes can be isolated using HLA class II-specific antibodies such as the M5 / 114.15.2 monoclonal antibody. In some embodiments, a single (or a pair of) HLA alleles are expressed as a fusion protein with a peptide tag, and the HLA-peptide complex is isolated using a binding molecule that recognizes the peptide tag.

[0492] The method further comprises isolating the peptide from the HLA-peptide complex and sequencing the peptide. The peptide is isolated from the complex by any method known to those skilled in the art, such as acid elution. Although any sequencing method can be used, in some embodiments, a method using mass spectrometry is used, such as liquid chromatography-mass spectrometry (LC-MS or LC-MS / MS, or HPLC-MS or HPLC-MS / MS). These sequencing methods are well known to those skilled in the art and are reviewed in Medzihradszky KF and Chalkley RJ. Mass Spectrom Rev. 2015 Jan-Feb; 34(1):43-63.

[0493] In some embodiments, the cell population expresses one or more endogenous HLA alleles. In some embodiments, the cell population is an engineered cell population lacking one or more endogenous HLA class I alleles. In some embodiments, the cell population is an engineered cell population lacking endogenous HLA class I alleles. In some embodiments, the cell population is an engineered cell population lacking one or more endogenous HLA class II alleles. In some embodiments, the cell population is an engineered cell population lacking endogenous HLA class II alleles or an engineered cell population lacking endogenous HLA class I alleles and endogenous HLA class II alleles. In some embodiments, the cell population comprises cells that have been enriched or sorted, for example, by fluorescence activated cell sorting (FACS). In some embodiments, fluorescence activated cell sorting (FACS) is used to sort cell populations. In some embodiments, the cell population is pre-FACS sorted for cell surface expression of HLA class I or class II or both HLA class I and class II. For example, FACS can be used to sort a population of cells for cell surface expression of HLA class I alleles, HLA class II alleles, or a combination thereof.

[0494] Methods for preparing personalized cancer vaccines

[0495] Once a specific mutation for cancer is identified, such that the mutation is present in the DNA of cancer cells but not in normal cells of the same human subject, and the mutation causes a change in one or more amino acids in the protein encoded by the DNA, the mutation can be a target for the host immune response. The natural immune response can be directed against the mutated protein, resulting in the destruction of cancer cells expressing the protein. Due to the natural tolerance response and immunocompromised environment in cancer tissues, immunotherapy is a clinical approach that attempts to enhance this immune response to exceed the body's tolerance and immunosuppression. Therefore, proteins or peptides containing the above mutations are suitable candidates for immunotherapy.

[0496] The mutated proteins are taken up by professional phagocytes acting as antigen presenting cells (APCs), chopped up, and displayed on the cell surface as antigens for T cell activation in antigen presentation complexes containing major histocompatibility complex (MHC) proteins. Human MHC proteins are called human leukocyte antigens, HLA. MHC proteins can be either MHC class I or class II proteins, and some functional differences are attributed to the presentation of peptides by class I or class II MHC proteins (HLA class I and HLA class II proteins), with one notable difference being the fact that HLA class I-peptide complexes present antigens to cytotoxic CD8+ T cells, while HLA class II peptide complexes are also able to activate CD4+ T cells, leading to a prolonged immune response. CD8+ T cells are essential in the task of cell-by-cell elimination of diseased cells, such as infected or tumor cells. CD4+ T cells have a more lasting impact after activation, the most important of which is the generation of immune memory. CD4 subsets are recruited differently depending on the type of immune threat, and multiple subsets with overlapping or different functions can be recruited together. This helps balance the immune response associated with pathogen threats. In these aspects, HLA class II peptide-mediated antigen presentation enables sustained and tailored immune responses. On the other hand, HLA class II binding to peptides can be promiscuous, so nonspecific peptide binding and presentation to the immune system can lead to abnormal immune responses, such as autoimmunity.

[0497] In one aspect, the present disclosure provides a method for predicting a peptide that can accurately pair or bind to a specific HLA class II α and β chain heterodimer, such that high-fidelity binding of the peptide to HLA class II proteins (including α and β chain heterodimers) ensures that the specific peptide is presented to T lymphocytes, thereby eliciting a specific immune response and avoiding any cross-reaction or immune promiscuity.

[0498] In one aspect, the present disclosure provides a method for predicting a peptide that can accurately bind to a specific HLA class II protein, so that when the peptide is therapeutically administered to a subject expressing a specific homologous HLA class II protein, the peptide can be used to activate a more sustained and powerful immune response by virtue of the ability of the HLA class II protein to activate CD4+ T cells and stimulate immune memory. In some embodiments, a given peptide predicted to bind to an HLA class II protein with high specificity is a peptide comprising a mutation, wherein the mutation is prevalent in a subject's cancer or tumor cells; and the same HLA class II protein predicted to bind to the mutant peptide either (a) does not bind, or (b) binds to the corresponding non-mutated wild-type peptide with a significantly lower affinity than the mutant peptide of the subject. The preferential binding of HLA to mutant peptides is advantageous in the development of immunotherapeutics because cells expressing wild-type peptides will be protected from immune attacks by T cells reactive to HLA-presented peptides. In some embodiments, the predicted peptide that specifically binds to an HLA class II protein is a peptide with a post-translational modification. Exemplary post-translational modifications include, but are not limited to, phosphorylation, ubiquitination, dephosphorylation, glycosylation, methylation, or acetylation. In some embodiments, the predicted peptide is post-translationally modified prior to use in immunotherapy.

[0499] In some embodiments, the immunotherapy methods and strategies disclosed herein can also be applied to inhibit unwanted immune activation, such as in autoimmune reactions. Specifically, peptides identified as potential binders to specific HLA subtypes can be tailored to bind to specific HLA molecules and induce tolerance rather than cause an immunogenic response.

[0500] On the one hand, the present invention proposes a customized or personalized immunotherapy method for a specific subject. Each subject or patient expresses a specific set of HLA class I and HLA class II proteins. HLA typing is a well-known technology that allows the determination of a specific repertoire of HLA proteins expressed by a subject. Once the HLA heterodimers expressed by a specific subject are known, it is possible to generate a specific immune response specifically customized for the subject with improved, complex and reliable methods for predicting peptides that can bind to specific HLA class II alpha and beta chain heterodimers with high fidelity as described herein.

[0501] The genes encoding HLA heterodimers are highly polymorphic, with more than 4,000 HLA class II allelic variants identified throughout the human population. For each of the HLA class II loci, an individual can inherit different alleles from maternal and paternal HLA haplotypes, and each HLA class II heterodimer consists of an α chain and a β chain. Due to the large number of α-chain and β-chain pairing combinations, especially for HLA-DP and HLA-DQ alleles, the population of possible HLA heterodimers is very complex. HLA class II heterodimers are translated in the endoplasmic reticulum (ER) and assembled into a stable complex with the invariant chain (Ii) derived from the protein CD74. Ii stabilizes the class II complex by allowing correct protein folding and enables HLA class II heterodimers to be exported into endosomal / lysosomal compartments. Within these HLA class II loading compartments, Ii is proteolytically cleaved by cathepsins into placeholder peptides known as CLIPs. CLIP is then exchanged for higher affinity peptides in a low pH environment by the molecular chaperone HLA-DM (a non-classical HLA class II heterodimer). The HLA class II complex loaded with high affinity peptides then reaches the trans-Golgi apparatus and finally the cell surface for display by CD4+ T cells.

[0502] Each HLA heterodimer is estimated to bind thousands of peptides with allele-specific binding preferences. In fact, each HLA allele is estimated to bind and present approximately 1,000-10,000 unique peptides to T cells. Given this diversity in HLA binding, accurate prediction of whether a peptide is likely to bind to a specific HLA allele is very challenging. Little is known about the allele-specific peptide binding properties of HLA class II molecules because of the heterogeneity of α and β chain pairing, the complexity of the data that limits the ability to confidently assign core binding epitopes, and the lack of allele-specific antibodies of the immunoprecipitation grade required for high-resolution biochemical analysis. Furthermore, analysis of peptide epitopes derived from a given HLA allele introduces uncertainty when multiple HLA alleles are presented on the cell surface.

[0503] Prediction of candidate neoantigens has been performed primarily for HLA class I epitopes (given the availability of experimental data for class I prediction algorithms compared to class II), but CD4+ T cell responses have been frequently observed in preclinical and clinical personalized neoantigen vaccination studies. These observations suggest that HLA class II epitope processing and presentation may also play a key role in cancer therapy. Although HLA class II prediction algorithms exist, they are inaccurate because the open peptide binding groove on the HLA class II heterodimer allows longer peptides (typically 15-40 amino acids) to bind, which increases the heterogeneity and complexity of epitope presentation. Therefore, further work is needed to better understand the properties of the HLA class II peptide binding core and the cellular processes involved in class II epitope processing and presentation. The field of proteomics is currently limited by the complexity of HLA class II heterodimer formation and the availability of immunoprecipitation-grade antibodies for isolation of HLA class II-peptide complexes. To overcome these challenges, a monoallelic HLA profiling workflow was developed that relies on LC-MS / MS to characterize allele-specific HLA class II-ligand sets and class II epitope prediction methods. The following definitions supplement those in the art and are for the present application and should not be attributed to any related or unrelated circumstances, such as any commonly owned patents or applications. Exemplary materials and methods are described herein, but any methods and materials similar or equivalent to those described herein can be used in the practice of testing the present disclosure. Therefore, the terms used herein are for the purpose of describing specific embodiments only and are not intended to be limiting.

[0504] Disclosed herein is a method for preparing a personalized cancer vaccine. The method for preparing a personalized cancer vaccine may include identifying a peptide sequence with a mutation expressed in a subject's cancer cells; inputting the amino acid position information of the identified peptide sequence into a machine learning HLA-peptide presentation prediction model using a computer processor to generate a set of presentation predictions for the identified peptide sequence, each presentation prediction representing the probability that one or more proteins encoded by the class II MHC alleles of the subject's cancer cells will present a given sequence of the identified peptide sequence; and selecting a subset of the peptide sequences identified based on the set of presentation predictions for use in preparing a personalized cancer vaccine.

[0505] In some embodiments, one or more results obtained from the methods described herein may provide one or more quantitative values ​​indicating one or more of the following: the likelihood of diagnostic accuracy, the likelihood of the presence of a condition in a subject, the likelihood of a subject developing a condition, the likelihood of a particular treatment being successful, or any combination thereof. In some embodiments, the methods described herein may predict the risk or likelihood of developing a condition. In some embodiments, the methods described herein may be an early diagnostic indication of the occurrence of a condition. In some embodiments, the methods described herein may confirm the diagnosis or presence of a condition. In some embodiments, the methods described herein may monitor the progression of a condition. In some embodiments, the methods described herein may monitor the efficacy of a treatment on a subject's condition.

[0506] Methods for identification of MHC-II peptides

[0507] In one aspect, the present invention provides a method for identifying one or more peptides presented by MHC-II proteins for immune activation. In some embodiments, the one or more peptides comprise an epitope. In some embodiments, the method relates to a computational prediction of the likelihood that a particular epitope is presented by an MHC-II protein. In some embodiments, the method relates to a computational prediction of the specificity of an epitope for MHC-II presentation. In some embodiments, the computational prediction method relates to an assessment of peptide-MHC interactions. In some embodiments, the computational prediction method relates to predicting the allele specificity of a peptide for antigen presentation.

[0508] In some embodiments, the computational prediction method involves the integration of bioinformatics information, such as nucleotide sequences, structural motifs of biomolecules, protein-protein interaction characteristics and functional efficacy such as immunogenicity. In some embodiments, the computational prediction method involves machine learning. Based on machine learning methods, such as simple pattern motifs, support vector machines (SVM), hidden Markov models (HMM), neural network (NN) models, quantitative structure-activity relationship (QSAR) analysis, structure-based methods and biophysical methods, many immunoinformatics methods for predicting peptide-MHC interactions have been developed for MHC class I and class II. These methods can be divided into two categories, namely, intra-allelic (allele-specific) and trans-allelic (pan-specific) methods. The intra-allelic method is trained on a limited set of experimental peptide binding data for a specific MHC molecule and is applied to predict peptides bound to the molecule. Due to the extreme polymorphism of MHC molecules, the existence of thousands of allelic variants, and the lack of sufficient experimental binding data, it is impossible to establish a prediction model for each allele. Therefore, cross-allele and universal methods such as NetMHCIIpan (Karosiene E et al., NetMHCIIpan-3.0, a common pan-specific MHC classII prediction method including all three human MHC class II isotypes, HLA-DR, HLA-DP and HLADQ. Immunogenetics (2013) 65(10): 711–24) and TEPITOPEpan (Zhang L et al., TEPITOPEpan: extending TEPITOPE for peptide binding prediction covering over700HLA-DR molecules. PLoS One (2012) 7(2): e30483) have been developed using peptide binding data that extend over many alleles or across species. Similar methods for MHC-I are also available, such as NetMHCpan and KISS.

[0509] In some embodiments, the peptide sequence may not be expressed in normal cells of the subject. In some embodiments, each and every cell of the subject may not be a cancer cell. The cancer cell may be produced by different cancers, including but not limited to thyroid cancer, adrenocortical cancer, anal cancer, aplastic anemia, bile duct cancer, bladder cancer, bone cancer, bone metastasis, central nervous system (CNS) cancer, peripheral nervous system (PNS) cancer, breast cancer, Castleman disease, cervical cancer, childhood non-Hodgkin lymphoma, lymphoma, colorectal cancer, endometrial cancer, esophageal cancer, Ewing tumor family (e.g., Ewing sarcoma), eye cancer, gallbladder cancer, gastrointestinal carcinoid tumor, gastrointestinal stromal tumor, gestational trophoblastic disease, hairy cell leukemia, Hodgkin disease, Kaposi's sarcoma, kidney cancer, laryngeal cancer and hypopharyngeal cancer, acute lymphocytic leukemia, acute myeloid leukemia, leukemia, childhood leukemia, chronic lymphocytic leukemia, chronic myeloid leukemia, liver cancer, lung cancer, lung carcinoid tumor, non-Hodgkin lymphoma, male breast cancer, malignant mesothelioma, multiple myeloma, myelodysplastic syndrome, myeloproliferative disorders, nasal and paranasal cancer, nasopharyngeal cancer, neuroblastoma, oral cavity and oropharyngeal cancer, osteosarcoma, ovarian cancer, pancreatic cancer, penile cancer, pituitary tumor, prostate cancer, retinoblastoma, rhabdomyosarcoma, salivary gland cancer, sarcoma (soft tissue cancer in adults), melanoma skin cancer, non-melanoma skin cancer, stomach cancer, testicular cancer, thymic cancer, uterine cancer (e.g., uterine sarcoma), vaginal cancer, vulvar cancer, or Waldenstrom's macroglobulinemia.

[0510] The identification may include comparing the DNA, RNA or protein sequence from the subject's cancer cell with the DNA, RNA or protein sequence from the subject's normal cell. The DNA, RNA or protein sequence from the subject's cancer cell may be different from the DNA, RNA or protein sequence from the subject's normal cell. The identification may identify nucleic acid variants with high sensitivity.

[0511] The machine learning HLA-peptide presentation prediction model may include a plurality of predictor variables identified based at least on training data. The training data may include sequence information of sequences of peptides presented by HLA proteins expressed in cells and identified by mass spectrometry; training peptide sequence information including amino acid position information, wherein the training peptide sequence information is associated with HLA proteins expressed in cells; and a function representing the relationship between the amino acid position information received as input and the presentation likelihood generated as output based on the amino acid position information and the predictor variables.

[0512] In some embodiments, the training data may further include structured data, time series data, unstructured data, and relational data. Unstructured data may include audio data, image data, video, mechanical data, electrical data, chemical data, and any combination thereof, for accurately simulating or training a robot or simulation. Time series data may include data from one or more of a smart meter, smart appliance, smart device, monitoring system, telemetry device, or sensor. Relational data includes data from a client system, an enterprise system, an operating system, a website, a network accessible application program interface (API), or any combination thereof. This may be accomplished by a user by any method in a file or other data format input into a software or system.

[0513] In some embodiments, the training data can be stored in a database. The database can be stored in a computer-readable format. The computer processor can be configured to access data stored in a computer-readable memory. In some embodiments, a computer system can be used to analyze data to obtain a result. The result can be stored remotely or internally on a storage medium and transmitted to personnel such as a drug expert. In some embodiments, a computer system can be operably coupled with a component for transmitting the result. The component for transmission can include wired and wireless components. Examples of wired communication components can include universal serial bus (USB) connections, coaxial cable connections, Ethernet cables such as Cat5 or Cat6 cables, fiber optic cables or telephone lines. Examples of wireless communication components can include Wi-Fi receivers, components for accessing mobile data standards such as 3G or 4GLTE data signals, or Bluetooth receivers. In some embodiments, all these data in the storage medium are collected and archived to build a data warehouse.

[0514] In some embodiments, the database includes an external database. The external database can be a medical database, such as, but not limited to, an adverse drug reaction database, an AHFS supplement file, an allergen selection list file, an average WAC pricing file, a brand probability file, a Canadian drug file v2, a comprehensive price history, a controlled substance file, a drug allergy cross reference file, a drug application file, a drug administration and administration database, a drug image database v2.0 / drug imprint database v2.0, a drug expiration date file, a drug indication database, a drug laboratory conflict database, a drug therapy monitoring system (DTMS) v2.2 / DTMS consumer monograph, a duplicate treatment database, a federal government pricing file, a healthcare common procedure coding system code (HCPCS) database, an ICD-10 mapping file, an immunization cross reference file, a comprehensive A to Z drug fact module, a comprehensive patient education, a master parameter database, a half span electronic drug file (MED-File) v2, a Medicaid Rebate file, health care plan file, medical condition selection list file, medical condition master database, medication order management database (MOMD), monitoring parameter database, patient safety program file, payment limits-part B (PAF-B) v2.0, precautionary measures database, RxNorm cross-reference file, standard drug identifier database, substitution group file, supplemental name file, uniform system of classification cross-reference file, or warning label database.

[0515] In some embodiments, training data may also be obtained through other data sources. Data sources may include sensors or smart devices, such as appliances, smart meters, wearable devices, monitoring systems, data stores, customer systems, billing systems, financial systems, crowd source data, weather data, social networks, or any other sensor, enterprise system, or data store. Examples of smart meters or sensors may include meters or sensors located at customer locations, or meters or sensors located between customers and generation or source locations. By integrating data from a wide range of sources, the system may be able to perform complex and detailed analysis. In some embodiments, data sources may include, but are not limited to, sensors or databases for other medical platforms.

[0516] HLA typing is conventionally performed by serological methods using antibodies or by PCR-based methods such as sequence-specific oligonucleotide probe hybridization (SSOP) or sequence-based typing (SBT). The first is hampered by a potentially high degree of cross-reactivity and limited resolution, while the second has difficulties associated with PCR efficiency, since the possibilities for positioning primers are very limited due to the polymorphic positions.

[0517] In some embodiments, sequence information is identified by a sequencing method or a method using mass spectrometry, such as liquid chromatography-mass spectrometry (LC-MS or LC-MS / MS, or HPLC-MS or HPLC-MS / MS). These sequencing methods may be well known to the skilled person and are reviewed in Medzihradszky KF and Chalkley RJ. Mass Spectrom Rev. 2015 Jan-Feb; 34(1):43-63. In some embodiments, the mass spectrometry is monoallelic mass spectrometry. In some embodiments, the mass spectrometry may be MS analysis, MS / MS analysis, LC-MS / MS analysis, or a combination thereof. In some embodiments, MS analysis may be used to determine the mass of the intact peptide. For example, the determination may include determining the mass of the intact peptide (e.g., MS analysis). In some embodiments, MS / MS analysis may be used to determine the mass of a peptide fragment. For example, the determination may include determining the mass of a peptide fragment, which may be used to determine the amino acid sequence of a peptide or a portion thereof (e.g., MS / MS analysis). In some embodiments, the mass of a peptide fragment may be used to determine the amino acid sequence within the peptide. In some embodiments, LC-MS / MS analysis can be used to separate complex peptide mixtures. For example, the determination can include, for example, separation of complex peptide mixtures by liquid chromatography, and determination of the mass of the complete peptide, the mass of the peptide fragments, or a combination thereof (e.g., LC-MS / MS analysis). The data can be used, for example, for peptide sequencing.

[0518] In some embodiments, the training peptide sequence information includes amino acid position information of the training peptide. In some embodiments, the training peptide sequence information includes at most about 90%, 80%, 70%, 60%, 50%, 40%, 30%, 20%, 10% or less of the sequence of the peptide presented by the HLA protein expressed in the cell and identified by mass spectrometry. In some embodiments, the training peptide sequence information may include at least about 10%, 20%, 30%, 40%, 50%, 60%, 70%, 80%, 90% or more of the sequence of the peptide presented by the HLA protein expressed in the cell and identified by mass spectrometry.

[0519] Any information and data can be paired with the subject as the source of the information and data. The subject or medical professional can retrieve information and data from the storage or server through the subject identity. The subject identity can include a patient's photo, name, address, social security number, birthday, telephone number, zip code, or any combination thereof. The subject identity can be encrypted and encoded into a visible graphic code. The visible graphic code can be a one-time barcode that can be uniquely associated with the subject identity. The barcode can be a UPC barcode, an EAN barcode, a Code 39 barcode, a Code 128 barcode, an ITF barcode, a CodaBar barcode, a GS1 DataBar barcode, an MSI Plessey barcode, a QR barcode, a Datamatrix code, a PDF417 code, or an Aztec barcode. The visible graphic code can be configured to be displayed on a display screen. The barcode can include a QR that can be optically captured and read by a machine. The barcode can define elements such as the version, format, position, alignment, or timing of the barcode to enable the reading and decoding of the barcode. Barcodes can encode various types of information, such as binary or alphanumeric information, in any suitable format. QR codes can have various symbol sizes, as long as the QR code can be scanned from a reasonable distance by an imaging device. QR codes can be in any image file format (such as EPS or SVG vector graphics, PNG, TIF, GIF, or JPEG raster graphics formats).

[0520] In some embodiments, the function representing the relationship between the amino acid position information received as input and the presentation probability generated as output based on the amino acid position information and the predictor variable comprises a linear or nonlinear function. The function can be, for example, a rectified linear unit (ReLU) activation function, a Leaky ReLu activation function, or other functions, such as saturated hyperbolic tangent, identity, binary step function, logistic function, arcTan, softsign, parametric rectified linear unit, exponential linear unit, softPlus, bent identity, softExponential, sinusoid, Sinc, Gaussian or sigmoid function, or any combination thereof.

[0521] In some embodiments, the linear function is obtained by linear regression. In some embodiments, linear regression is a method for predicting a target variable by fitting the best linear relationship between a dependent variable and an independent variable. The best fit may mean that the sum of all distances between the shape at each point and the actual observed value is the smallest. Linear regression may include simple linear regression or multiple linear regression. Simple linear regression can use a single independent variable to predict the dependent variable. Multiple linear regression can use more than one independent variable to predict the dependent variable by fitting the best linear relationship. Nonlinear function can be obtained by nonlinear regression. Nonlinear regression can be a form of regression analysis, in which observed data is modeled by a function that is a nonlinear combination of model parameters and depends on one or more independent variables. Nonlinear regression may include step functions, piecewise functions, splines, and generalized additive models.

[0522] In some embodiments, the presentation possibility is presented by a one-dimensional value (e.g., probability). In some embodiments, probability is configured to measure the possibility that an event may occur. In some embodiments, the probability range is about 0 to 1, 0.1 to 0.9, 0.2 to 0.8, 0.3 to 0.7 or 0.4 to 0.6. The higher the probability of an event occurring, the more likely the event is to occur. In some embodiments, an event includes any type of situation, including, as a non-limiting example, whether HLA-peptides will present certain peptides with specific amino acid position information, and whether people will get sick based on amino acid position information. In some embodiments, the possibility can be presented by a multidimensional value. Multidimensional values ​​can be presented by multidimensional space, heat map or electronic table.

[0523] In one embodiment, a subset of peptide sequences identified based on a group of presentation predictions is selected to be configured to prepare a personalized cancer vaccine. In some embodiments, the subset comprises up to about 90%, 80%, 70%, 60%, 50%, 40%, 30%, 20%, 10% or less of peptide sequences identified based on the group of presentation predictions. In other cases, the subset may comprise at least about 10%, 20%, 30%, 40%, 50%, 60%, 70%, 80%, 90% or more of peptide sequences identified based on the group of presentation predictions. A cancer vaccine may be a vaccine for treating an existing cancer or preventing cancer development. The vaccine may be prepared by a sample taken from a patient and may be specific to the patient.

[0524] In some embodiments, poxviruses are used in disease (e.g., cancer) vaccines or immunogenic compositions. These include orthopoxviruses, avipox, vaccinia, MVA, NYVAC, canarypox, ALVAC, fowlpox, TROVAC, and the like. Advantages of this vector may include simple construction, the ability to accommodate large amounts of foreign DNA, and high expression levels. With regard to poxviruses that may be used to practice the present disclosure, such as Chordopoxvirinae poxviruses (poxviruses of vertebrates), e.g., orthopoxviruses and avipox viruses, such as vaccinia viruses (e.g., Wyeth strains, WR strains (e.g., VR-1354), Copenhagen strain, NYVAC, NYVAC.1, NYVAC.2, MVA, MVA-BN), canarypox virus (e.g., Wheatley C93 strain, ALVAC), fowlpox virus (e.g., FP9 strain, Webster strain, TROVAC), dovepox, pigeonpox, quailpox and raccoonpox, particularly information on their synthetic or non-naturally occurring recombinants, their uses and methods for making and using such recombinants can be found in the scientific and patent literature.

[0525] In some embodiments, vaccinia virus is used in disease vaccines or immunogenic compositions to express antigens. Recombinant vaccinia virus may be able to replicate in the cytoplasm of infected host cells, so the polypeptide of interest can induce an immune response.

[0526] In some embodiments, ALVAC is used as a vector in a disease vaccine or immunogenic composition. ALVAC may be a canarypox virus that can be modified to express foreign transgenes and has been used as a vaccination method against prokaryotic and eukaryotic antigens.

[0527] In some embodiments, a modified vaccinia Ankara (MVA) virus is used as a viral vector for an antigen vaccine or immunogenic composition. MVA can be a member of the Orthopoxviridae family and has been produced by approximately 570 serial passages of the Ankara strain of vaccinia virus (CVA) in chicken embryo fibroblasts. As a result of these passages, the resulting MVA virus contains 31 kilobases less genomic information than CVA and is highly host cell restricted. MVA can be characterized by its extreme attenuation, i.e., reduced virulence or infectivity, but still excellent immunogenicity. When tested in a variety of animal models, MVA can be shown to be non-toxic, even in immunosuppressed individuals. In addition, MVA- -HER2 may be a candidate immunotherapy for the treatment of HER-2-positive breast cancer and is currently undergoing clinical trials.

[0528] In some embodiments, a positive predictive value (PPV) is used as part of a prediction model. PPV, also referred to as a precision measurement, is the probability that an individual diagnosed with a disease or condition by, for example, a test or model actually suffers from the disease or condition. It can be calculated by dividing the number of true positive results by the total number of positive results (including false positive results) returned. PPV = true positive / (true positive + false positive). For example, if in a group of 100 patients, the model determines a positive result in 50 patients, 25 of which are true positives, then the PPV will be 25 / 50 = 0.5. The closer the PPV is to 1, the more accurate the diagnostic method, such as a test or model, is. PPV can be used to determine the accuracy of a prediction model. PPV can be used to adjust the prediction model to accommodate the false positive results that the model may generate.

[0529] The decision rate can be used as part of a predictive model. The decision rate can be considered as the percentage of true positive results to the total number of positives in the sample set. Decision rate = true positive / (true positive + false negative). For example, if in a group of 100 patients, the model determines positive results in 50 patients, 25 of which are true positives, and there are a total of 75 positives in the group of patients, then the decision rate is {25 / (25+25)}x100=50%. The decision rate can be used to determine the accuracy of the predictive model. The decision rate can be used to adjust the predictive model to adapt to the false positive results or false negative results that the model may generate.

[0530] In some embodiments, the prediction model has a positive predictive value of at least 0.05, 0.1, 0.2, 0.25, 0.3, 0.4, 0.5, 0.6, 0.7, 0.8, 0.9 or more at a decision rate of 0.1%-10%. In some embodiments, the prediction model can have a positive predicti...

Claims

1. A method for preparing an HLA class II tetramer or multimer containing an epitope, the method comprising contacting purified soluble HLA-DM loaded with a peptide epitope with an HLA class II tetramer or multimer, thereby forming an HLA class II tetramer or multimer loaded with the peptide epitope.

2. The method of claim 1, wherein the HLA class II tetramer or multimer comprises HLA-DR or HLA-DP or HLA-DQ heterodimers, each heterodimer comprising an α and a β chain.

3. The method of claim 1, wherein the method comprises expressing the α and β chains of the HLA class II tetramer or multimer in a cell.

4. The method of claim 3, wherein the expressing comprises expressing the α and β chains of the HLA class II tetramer or multimer from a polynucleotide molecule comprising a sequence encoding the α chain and a sequence encoding the β chain, wherein the sequence encoding the α chain and the sequence encoding the β chain are separated by a ribosomal skipping sequence or a sequence encoding a protease cleavage site.

5. The method of claim 3, wherein expressing comprises expressing the α and β chains of the HLA class II tetramer or multimer in a eukaryotic cell.

6. The method of claim 1, wherein the method comprises purifying the HLA class II tetramer or multimer.

7. The method of claim 6, wherein the purification comprises gel filtration chromatography.

8. A method for purifying HLA class II tetramers or multimers loaded with an epitope, the method comprising (a) loading the HLA class II tetramers or multimers with the peptide epitope by contacting purified soluble HLA-DM loaded with the peptide epitope with the HLA class II tetramers or multimers, and (b) purifying the HLA class II tetramers or multimers loaded with the epitope using a suitable protein purification method; wherein the purification comprises purifying HLA class II tetramers or multimers loaded with the peptide epitope having a low level of post-translational modification from HLA class II tetramers or multimers having a high level of post-translational modification.

9. A vector comprising a sequence encoding an α chain of an HLA class II tetramer or multimer and a sequence encoding a β chain of the HLA class II tetramer or multimer, wherein the sequence encoding the α chain and the sequence encoding the β chain are separated by a ribosomal skipping sequence or a sequence encoding a protease cleavage site.

10. A method for identifying an epitope of an HLA class II tetramer or multimer, the method comprising (a) incubating HLA class II tetramers or multimers in the presence of (i) soluble HLA-DM, (ii) a peptide probe comprising a detectable label, and (iii) a candidate peptide epitope, thereby forming a first complex comprising the HLA class II tetramer or multimer and the peptide probe and a second complex comprising the HLA class II tetramer or multimer and the candidate peptide epitope; (b) measuring the labeling of the peptide probe; (c) identifying the candidate peptide epitope as an epitope of an HLA class II tetramer or multimer based on (b).

11. The method of claim 10, wherein the detectable label is a fluorescent label and measuring the label of the peptide probe comprises measuring fluorescence polarization.

12. The method of claim 10 or 11, wherein the HLA class II tetramer or multimer is loaded with a placeholder peptide prior to incubation.

13. The method of claim 12, wherein the placeholder peptide is a peptide selected from the peptides of Table 19.

14. The method of claim 10, wherein the candidate peptide epitope is encoded by the genome or exome of the subject or a pathogen or virus in the subject.

15. The method of claim 10, wherein the peptide probe is a validated epitope of the HLA class II tetramer or multimer.

16. The method of claim 10, wherein the peptide probe comprises the sequence of a probe of Table 20 and the HLA class II tetramer or multimer comprises a protein encoded by a corresponding HLA allele of Table 20.

17. A library of HLA class II tetramers or multimers and matched validated epitopes of said HLA class II tetramers or multimers, wherein said validated epitopes comprise the sequences of the probes of Table 20 and said HLA class II tetramers or multimers comprise proteins encoded by the corresponding HLA alleles of Table 20.

18. The library of claim 17, wherein the validated epitope is fluorescently labeled.

19. A composition comprising purified soluble HLA-DM, wherein the purified soluble HLA-DM is present at a concentration greater than 1 mg / L.

20. A method for preparing purified soluble HLA-DM, the method comprising expressing a polynucleic acid sequence encoding an HLA-DM protein in a eukaryotic cell, wherein the HLA-DM protein is secreted from the eukaryotic cell; and purifying the HLA-DM protein from the supernatant of the eukaryotic cell.

21. An HLA class II tetramer or multimer comprising HLA-DR or HLA-DP or HLA-DQ heterodimers, each heterodimer comprising an α and a β chain, wherein the heterodimers are purified and present at a concentration greater than 16 mg / L.

22. A method comprising: (a) processing amino acid information of a plurality of candidate peptide sequences using a machine learning HLA peptide presentation prediction model to generate a plurality of presentation predictions, wherein each candidate peptide sequence in the plurality of candidate peptide sequences is encoded by a genome or exome of a subject or a pathogen or virus in the subject, wherein the plurality of presentation predictions comprises an HLA presentation prediction for each of the plurality of candidate peptide sequences, wherein each HLA presentation prediction indicates a likelihood that one or more proteins encoded by a class II HLA allele of a cell of the subject is capable of presenting a given candidate peptide sequence in the plurality of candidate peptide sequences, wherein the machine learning HLA peptide presentation prediction model is trained using training data comprising sequence information of training peptide sequences, wherein the training peptides are identified by a method comprising: (A) incubating HLA class II tetramers or multimers in the presence of (i) soluble HLA-DM, (ii) a fluorescently labeled peptide probe, and (iii) a training peptide epitope loaded with an epitope having an HLA class II tetramer or multimer, thereby forming a first complex comprising the HLA class II tetramer or multimer and the fluorescently labeled peptide probe and a second complex comprising the HLA class II tetramer or multimer and the training peptide epitope; (B) measuring polarization; and (C) determining the affinity of the training peptide epitope for the HLA class II tetramer or multimer based on (B); as well as (b) identifying, based at least on the plurality of presentation predictions, a peptide sequence from the plurality of peptide sequences as being presented by at least one of the one or more proteins encoded by the class II HLA alleles of cells of the subject.

23. The method according to any one of the preceding claims, wherein the HLA protein is an HLA class II protein selected from the following: HLA-DPB1*01:01 / HLA-DPA1*01:03, HLA-DPB1*02:01 / HLA-DPA1*01:03, HLA-DPB1*03:01 / HLA-DPA1*01:03, HLA-DPB1*04:01 / HLA-DPA1*01:03, HLA-DPB1*04:02 / HLA-DPA1*01:03, HLA-DPB1*06:01 / HLA-DPA1*01:03, HLA-DQB1*02:01 / HLA-DQA1*05:01, HLA-DQB1*02:02 / HLA-DQA1*02:01, HLA-DQB1*06:02 / HLA-DQA1*01:02, HLA-DQB1*06:04 / HLA-DQA1*01:02, HLA-DRB1*01:01, HLA-DRB1*01:02, HLA-DRB1*03:01, HLA-DRB1*03:02, HLA-DRB1*04:01, HLA-DRB1*04:02, HLA-DRB1*04:03, HLA-DRB1*04:04, HLA-DRB1*04:05, HLA-DRB1*04:07, HLA-DRB1*07:01, HLA-DRB1*08:01, HLA-DRB1*08:02, HLA-DRB1*08:03, HLA-DRB1*08:04, HLA-DRB1*09:01, HLA-DRB1*10:01, HLA-DRB1*11:01, HLA-DRB1*11:02, HLA-DRB1*11:04, HLA-DRB1*12:01, HLA-DRB1*12:02, HLA-DRB1*13:01, HLA-DRB1*13:02, HLA-DRB1*13:03, HLA-DRB1*14:01, HLA-DRB1*15:01, HLA-DRB1*15:02, HLA-DRB1*15:03, HLA-DRB1*16:01, HLA-DRB3*01:01, HLA-DRB3*02:02, HLA-DRB3*03:01, HLA-DRB4*01:01 and HLA-DRB5*01:

01.

24. A composition comprising the library according to any one of the above claims, wherein the HLA protein is an HLA class II protein selected from: HLA-DPB1*01:01 / HLA-DPA1*01:03, HLA-DPB1*02:01 / HLA-DPA1*01:03, HLA-DPB1*03:01 / HLA-DPA1*01:03, HLA-DPB1*04:01 / HLA-DPA1*01:03, HLA-DPB1*04:02 / HLA-DPA1*01:03, HLA-DPB1*06:01 / HLA-DPA1*01:03, HLA-DQB1*02:01 / HLA-DQA1*05:01, HLA-DQB1*02:02 / HLA-DQA1*02:01, HLA-DQB1*06:02 / HLA-DQA1*01:02, HLA-DQB1*06:04 / HLA-DQA1*01:02, HLA-DRB1*01:01, HLA-DRB1*01:02, HLA-DRB1*03:01, HLA-DRB1*03:02, HLA-DRB1*04:01, HLA-DRB1*04:02, HLA-DRB1*04:03, HLA-DRB1*04:04, HLA-DRB1*04:05, HLA-DRB1*04:07, HLA-DRB1*07:01, HLA-DRB1*08:01, HLA-DRB1*08:02, HLA-DRB1*08:03, HLA-DRB1*08:04, HLA-DRB1*09:01, HLA-DRB1*10:01, HLA-DRB1*11:01, HLA-DRB1*11:02, HLA-DRB1*11:04, HLA-DRB1*12:01, HLA-DRB1*12:02, HLA-DRB1*13:01, HLA-DRB1*13:02, HLA-DRB1*13:03, HLA-DRB1*14:01, HLA-DRB1*15:01, HLA-DRB1*15:02, HLA-DRB1*15:03, HLA-DRB1*16:01, HLA-DRB3*01:01, HLA-DRB3*02:02, HLA-DRB3*03:01, HLA-DRB4*01:01 and HLA-DRB5*01:

01.

25. A composition comprising a vector comprising a sequence encoding an HLA class II tetramer or multimer according to any one of the preceding claims, wherein the HLA protein is an HLA class II protein selected from the group consisting of HLA-DPB1*01:01 / HLA-DPA1*01:03, HLA-DPB1*02:01 / HLA-DPA1*01:03, HLA-DPB1*03:01 / HLA-DPA1*01:03, HLA-DPB1*04:01 / HLA-DPA1*01:03, HLA-DPB1*04:02 / HLA-DPA1*01:03, HLA-DPB1*06:01 / HLA-DPA1*01:03, HLA-DQB1*02:01 / HLA-DQ A1*05:01, HLA-DQB1*02:02 / HLA-DQA1*02:01, HLA-DQB1*06:02 / HLA-DQA1*01:02, HLA-DQB1*06:04 / HLA-DQA1*01:02, HLA-D RB1*01:01, HLA-DRB1*01:02, HLA-DRB1*03:01, HLA-DRB1*03:02, HLA-DRB1*04:01, HLA-DRB1*04:02, HLA-DRB1*04:03, HLA- DRB1*04:04, HLA-DRB1*04:05, HLA-DRB1*04:07, HLA-DRB1*07:01, HLA-DRB1*08:01, HLA-DRB1*08:02, HLA-DRB1*08:03, HLA -DRB1*08:04, HLA-DRB1*09:01, HLA-DRB1*10:01, HLA-DRB1*11:01, HLA-DRB1*11:02, HLA-DRB1*11:04, HLA-DRB1*12:01, HL A-DRB1*12:02, HLA-DRB1*13:01, HLA-DRB1*13:02, HLA-DRB1*13:03, HLA-DRB1*14:01, HLA-DRB1*15:01, HLA-DRB1*15:02, H LA-DRB1*15:03, HLA-DRB1*16:01, HLA-DRB3*01:01, HLA-DRB3*02:02, HLA-DRB3*03:01, HLA-DRB4*01:01 and HLA-DRB5*01:

01.

26. The method of claim 10 or 11, wherein the method has a detection sensitivity limit of 0.001% by standard fluorescence detection methods.

27. Use of HLA class II tetramers or multimers loaded with peptide epitopes prepared according to the method of claim 1 for identifying antigen-specific CD4+ T cells.

28. Use of HLA class II tetramers or multimers loaded with peptide epitopes purified according to the method of claim 8 for identifying antigen-specific CD4+ T cells.

29. Use of the HLA class II tetramer or multimer according to claim 17 and a library of matched validated epitopes of the HLA class II tetramer or multimer for identifying antigen-specific CD4+ T cells.

30. Use of the composition according to claim 19 for identifying antigen-specific CD4+ T cells.

31. Use of the purified soluble HLA-DM prepared according to the method of claim 20 for identifying antigen-specific CD4+ T cells.

32. Use of the HLA class II tetramer or multimer according to claim 21 for identifying antigen-specific CD4+ T cells.

Images