METHOD AND SYSTEM FOR T-CELL RECEPTOR (TCR) ASSAY DESIGN

20240303488

18/601946

March 11, 2024

A system and method of designing a T-cell receptor (TCR) assay includes the use of processor-based predictive modeling of an HLA binding classifier, T-cell response, sequencing T-cells, and TCR classifier/regression. Particularly, embodiments include feeding a representation of various peptides into a trained HLA binding classifier model configured to determine average binding predictions of overlapping peptides at each position of the viral or cancer protein. Based upon the average binding predictions, one or more peptide pools can be selected and fed into the T-cell response model, along with representative blood samples associated with a patient/patient population. Further, a sequenced resultant T-cell response can be used to detect T-cell response patterns. These detected patterns can be used to train the TCR classifier/regression model to predict or estimate a patient state. Ultimately, a primer can be designed using a detected minimum set of T-cell receptors for classifying or estimating the patient state.

ImmunityBio, Inc. (Culver City, CA, US)

1. A method of designing a T-cell receptor (TCR) assay performed by a processor-based TCR assay module, the method comprising: obtaining a first plurality of inputs representing a plurality of peptides; training an Artificial Neural Network (ANN) defining a pan-human leukocyte antigen (HLA) binding classifier model using the first plurality of inputs, wherein the trained HLA binding classifier model is configured to determine average binding predictions of overlapping peptides at each position of the viral or cancer protein independently for each of a plurality of test HLAs comprising HLA-I and HLA-II functional groupings; obtaining a second plurality of inputs representing a viral or cancer protein encoded into a plurality of peptides; feeding the second plurality of inputs into the trained HLA binding classifier model, wherein the trained HLA binding classifier is configured to determine average binding predictions of overlapping peptides of the plurality of peptides; selecting, based on the average binding predictions, one or more peptide pools from the plurality of peptides; obtaining a third plurality of inputs associated with a plurality of blood samples, wherein the blood samples are representative of a patient or patient population; instantiating, based on the one or more peptide pools and the third plurality of inputs, a T-cell response model; wherein the T-cell response model is trained to predict peptides and protein fragments associated with a high probability of eliciting T-cell response, based on validated T-cell epitopes and peptides failing to elicit a T-cell response; sequencing, by a sequencer, responding T-cells identified based on T-cell response criteria; detecting, based on data obtained from sequencing the responding T-cells, one or more T-cell response patterns common to the patient or patient population; training a TCR classifier/regression model to predict or estimate a patient state using datasets based on the one or more T-cell response patterns; determining, using the trained TCR classifier/regression model, a minimum set of T-cell receptors for classifying or estimating the patient state; and designing one or more primers based on the determined minimum set of T-cell receptors, the one or more primers defining a TCR assay for classifying or estimating the patient state.

2. The method of claim 1, wherein the training of the HLA binding classifier model comprises, obtaining a plurality of test HLAs encoded into variable-length proteins, wherein the plurality of test HLAs comprises HLA-I and HLA-II functional groupings; processing the encoded variable-length peptides corresponding to the viral protein and the variable-length proteins corresponding to the plurality of test HLAs using the classifier model such that, independently per test HLA, the classifier model is operable to determine an average binding prediction of overlapping peptides at each position of the viral protein; independently per test HLA: mapping in aggregate average binding predictions to locations along the test viral protein such that peptide-HLA interaction is indicated; determining nearest max locations for the average binding predictions using a sliding window having a fixed length; determining top max regions by selecting the nearest max locations having average binding predictions within a top percentage of values; selecting peptides classified as binders that overlap the top max regions; and determining a pan-HLA max region, wherein the determining includes setting unselected locations to zero, calculating a mean along an HLA axis of the average binding prediction, and selecting pan-HLA maxima within a top percentage of values based on the mean; independently for each of the HLA-I and HLA-II functional groupings: filtering the selected peptides classified as binders to identify candidate peptides that overlap the top max regions based on an aggregate of the pan-HLA max regions; and including one or more of the candidate peptides in an mRNA-based vaccine or therapeutic treatment for a patient.

3. The method of claim 1, wherein the training of the TCR classifier/regression model to predict or estimate a patient state comprises: differentiating TCR sequences specific to a patient state from general TCR sequences associated with patients not representative of a condition of interest associated with the patient state; identifying T-cell response patterns based on the differentiation; and generating the TCR classifier/regression model based on the identified T-cell response patterns.

4. The method of claim 1, wherein the training of the TCR classifier/regression model to predict or estimate a patient state comprises: differentiating TCR sequences specific to a patient state from general TCR sequences associated with patients not representative of a condition of interest associated with the patient state; identifying, based on the differentiation, TCR sequences in a sequence embedding space associated with the condition of interest; and generating the TCR classifier/regression model with the identified TCR sequences.

5. The method of claim 1, wherein the determining of the minimum set of T-cell receptors comprises: retrieving prediction scores of the trained TCR classifier/regression model for a plurality of patients; and selecting, based on the retrieved prediction scores, one or more TCR patterns.

6. The method of claim 1, wherein the method further comprises selecting the one or more peptide pools based on one or more of: a specific site, a hotspot, or a receptor-binding domain of the viral or cancer protein.

7. The method of claim 1, wherein the method further comprises selecting the one or more peptide pools based on multiple regions or hotspots of the viral or cancer protein.

8. The method of claim 1, wherein the method further comprises selecting the one or more peptide pools based on an entire viral or cancer protein.

9. The method of claim 1, wherein the method further comprises selecting the one or more peptide pools based on at least one of CD4 T-cell interaction or CD8 T-cell interaction.

10. The method of claim 1, wherein the method further comprises selecting the one or more peptide pools based on the average binding predictions for the HLA-I functional groupings.

11. The method of claim 1, wherein the method further comprises selecting the one or more peptide pools based on the average binding predictions for the HLA-II functional groupings.

12. The method of claim 1, wherein the method further comprises selecting the one or more peptide pools based on areas of predicted binding frequency across the HLA-I and HLA-II functional groupings.

13. The method of claim 1, wherein the method further comprises selecting the one or more peptide pools based on a pan-HLA binding prediction.

14. The method of claim 1, wherein the test for T cell response comprises at least one of the following: an enzyme-linked immunosorbent spot (ELISpot) assay test, a cytotoxic T Lymphocyte (CTL) assay test, and a DNA barcoded peptide-MHC (pMHC) multimers test.

15. The method of claim 1, wherein the test for T-cell response further comprises testing a synthetic TCR assay for T-cell response.

16. The method of claim 15, wherein the synthetic TCR assay is designed to supplement T-cell response data for the patient or patient population.

17. The method of claim 1, wherein the method further comprises using the TCR assay to classify or estimate a patient state.

18. The method of claim 17, wherein the method further comprises administering a therapeutic treatment to a patient based on the classified or estimated patient state.

19. A computer program product comprising a non-transitory computer readable medium comprising processor-executable instructions that, when executed by one or more processors, cause the one or more processors to perform operations to: obtain a first plurality of inputs representing a plurality of peptides; train an Artificial Neural Network (ANN) defining a pan-human leukocyte antigen (HLA) binding classifier model using the first plurality of inputs, wherein the trained HLA binding classifier model is configured to determine average binding predictions of overlapping peptides at each position of the viral or cancer protein independently for each of a plurality of test HLAs comprising HLA-I and HLA-II functional groupings; obtain a second plurality of inputs representing a viral or cancer protein encoded into a plurality of peptides; feed the second plurality of inputs into the trained HLA binding classifier model, wherein the trained HLA binding classifier is configured to determine average binding predictions of overlapping peptides of the plurality of peptides; select, based on the average binding predictions, one or more peptide pools from the plurality of peptides; obtain a third plurality of inputs associated with a plurality of blood samples, wherein the blood samples are representative of a patient or patient population; instantiate, based on the one or more peptide pools and the third plurality of inputs, a T-cell response model; wherein the T-cell response model is trained to predict peptides and protein fragments associated with a high probability of eliciting T-cell response, based on validated T-cell epitopes and peptides failing to elicit a T-cell response; sequence, by a sequencer, responding T-cells identified based on T-cell response criteria; detect, based on data obtained from sequencing the responding T-cells, one or more T-cell response patterns common to the patient or patient population; train a TCR classifier/regression model to predict or estimate a patient state using datasets based on the one or more T-cell response patterns; determining, using the trained TCR classifier/regression model, a minimum set of T-cell receptors for classifying or estimating the patient state; and designing one or more primers based on the determined minimum set of T-cell receptors, the one or more primers defining a TCR assay for classifying or estimating the patient state.

20. A computer system comprising: a memory storing one or more instructions for designing a T-cell receptor (TCR) assay; and one or more processors, coupled with the memory, the one or more processors configured to execute the one or more instructions to perform operations to: obtain a first plurality of inputs representing a plurality of peptides; train an Artificial Neural Network (ANN) defining a pan-human leukocyte antigen (HLA) binding classifier model using the first plurality of inputs, wherein the trained HLA binding classifier model is configured to determine average binding predictions of overlapping peptides at each position of the viral or cancer protein independently for each of a plurality of test HLAs comprising HLA-I and HLA-II functional groupings; obtain a second plurality of inputs representing a viral or cancer protein encoded into a plurality of peptides; feed the second plurality of inputs into the trained HLA binding classifier model, wherein the trained HLA binding classifier is configured to determine average binding predictions of overlapping peptides of the plurality of peptides; select, based on the average binding predictions, one or more peptide pools from the plurality of peptides; obtain a third plurality of inputs associated with a plurality of blood samples, wherein the blood samples are representative of a patient or patient population; instantiate, based on the one or more peptide pools and the third plurality of inputs, a T-cell response model; wherein the T-cell response model is trained to predict peptides and protein fragments associated with a high probability of eliciting T-cell response, based on validated T-cell epitopes and peptides failing to elicit a T-cell response; sequence, by a sequencer, responding T-cells identified based on T-cell response criteria; detect, based on data obtained from sequencing the responding T-cells, one or more T-cell response patterns common to the patient or patient population; train a TCR classifier/regression model to predict or estimate a patient state using datasets based on the one or more T-cell response patterns; determine, using the trained TCR classifier/regression model, a minimum set of T-cell receptors for classifying or estimating the patient state; and design one or more primers based on the determined minimum set of T-cell receptors, the one or more primers defining a TCR assay for classifying or estimating the patient state.

06; 16

edspap.20240303488