Philip Bradley
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Research Interests
Philip Bradley has a total of 384 co-authors. Here are the top co-authors by number of shared publications:
- Paul Glyndwr Thomas
Department of Host Microbe Interactions, St Jude Children’s Research Hospital, Memphis, USA
13 shared publications - Frederick Albert Matsen
Computational Biology Program, Fred Hutchinson Cancer Research Center, Seattle, Washington, USA
12 shared publications - Andrew Fiore-Gartland
Vaccine and Infectious Disease Division, Fred Hutchinson Cancer Center, Seattle, USA
6 shared publications - David Baker
Department of Biochemistry, University of Washington, Seattle, WA, USA
6 shared publications - Koshlan Mayer-Blackwell
Fred Hutchinson Cancer Research Center
6 shared publications - Magdalena L. Russell
Computational Biology Program, Fred Hutchinson Cancer Center
6 shared publications
Latest Publications
- Open Access
- 24 October 2026
- Published by Elsevier in Cell Reports
- Vol. 45 (11)
Infection history and TCR sequence features shape the breadth of TCR cross-reactivity
- 24 October 2026
- Published by Elsevier in Cell Reports
- Vol. 45 (11)
Open AccessInfection history and TCR sequence features shape the breadth of TCR cross-reactivity
- 3 September 2026
- Published by American Association for the Advancement of Science (AAAS) in Science
- Vol. 393 (6815)
Uneven TCR chain pairing constraints govern epitope recognition
Combinatorial pairing of independently recombined T cell receptor (TCR) α- and β-chains is central to diversifying the TCR repertoire. Although sequence motifs in one chain correlate with epitope recognition, the extent to which a single chain dictates specificity remains unclear. Here, we systematically tested TCR chain coupling constraints by enforcing the pairing of individual chains with hundreds of thousands of partners. Although most chains paired stably, the preservation of epitope specificity was rare and highly variable, with the frequency of compatible partners ranging from ~10 to 70,000 epitope-specific TCRs across 10 epitopes. Our work illuminates the distinct contributions of TCR chains, highlights the limitations of single-chain data, and provides an experimental and analytical framework for refining TCR-peptide-major histocompatibility complex specificity inference.
- 3 September 2026
- Published by American Association for the Advancement of Science (AAAS) in Science
- Vol. 393 (6815)
Uneven TCR chain pairing constraints govern epitope recognition
Combinatorial pairing of independently recombined T cell receptor (TCR) α- and β-chains is central to diversifying the TCR repertoire. Although sequence motifs in one chain correlate with epitope recognition, the extent to which a single chain dictates specificity remains unclear. Here, we systematically tested TCR chain coupling constraints by enforcing the pairing of individual chains with hundreds of thousands of partners. Although most chains paired stably, the preservation of epitope specificity was rare and highly variable, with the frequency of compatible partners ranging from ~10 to 70,000 epitope-specific TCRs across 10 epitopes. Our work illuminates the distinct contributions of TCR chains, highlights the limitations of single-chain data, and provides an experimental and analytical framework for refining TCR-peptide-major histocompatibility complex specificity inference.
- Open Access
Deciphering antigen-driven T cell responses through vectorized TCRdist sequence neighborhood quantification
T cells provide precise mechanisms to defend the body against infection and malignancies, mediated through the expression of their hypervariable T cell receptors (TCRs). Interpreting similarity between TCRs, however, remains a significant challenge. While performant clustering methods exist, these often fail to distinguish between antigen-driven convergent selection and patterns arising stochastically from biases in the V(D)J recombination mechanism. Moreover, defining enrichment in sequence similarity among large repertoires is computationally taxing. To address these limitations, we present an efficient computational framework for rapid approximation of TCRdist distances using fixed-length vector embeddings and highly optimized nearest neighbor search, allowing sequence similarity enrichment testing at a multi-repertoire-wide scale. This framework leverages a novel shuffling-based background model that preserves important repertoire characteristics such as V gene frequency, CDR3 sequence length and generation probability more accurately than synthetic models. Together, these tools enable the efficient and robust identification of significantly neighbor enriched (SNE) TCR sequences at scale. We validate this approach by showing a significant enrichment of SNE clones in memory T cell fractions and further demonstrate its utility in identifying convergent T cell signatures of response to vaccination and viral infections, providing a scalable approach for antigen-agnostic T cell response profiling.
Open AccessDeciphering antigen-driven T cell responses through vectorized TCRdist sequence neighborhood quantification
T cells provide precise mechanisms to defend the body against infection and malignancies, mediated through the expression of their hypervariable T cell receptors (TCRs). Interpreting similarity between TCRs, however, remains a significant challenge. While performant clustering methods exist, these often fail to distinguish between antigen-driven convergent selection and patterns arising stochastically from biases in the V(D)J recombination mechanism. Moreover, defining enrichment in sequence similarity among large repertoires is computationally taxing. To address these limitations, we present an efficient computational framework for rapid approximation of TCRdist distances using fixed-length vector embeddings and highly optimized nearest neighbor search, allowing sequence similarity enrichment testing at a multi-repertoire-wide scale. This framework leverages a novel shuffling-based background model that preserves important repertoire characteristics such as V gene frequency, CDR3 sequence length and generation probability more accurately than synthetic models. Together, these tools enable the efficient and robust identification of significantly neighbor enriched (SNE) TCR sequences at scale. We validate this approach by showing a significant enrichment of SNE clones in memory T cell fractions and further demonstrate its utility in identifying convergent T cell signatures of response to vaccination and viral infections, providing a scalable approach for antigen-agnostic T cell response profiling.
- Open Access
IMMREP25: Unseen Peptides
T cell receptors (TCRs) can bind to peptides presented by MHC molecules (pMHC) as a first step to trigger a T cell response. Reliable approaches to predict TCR:pMHC binding would have broad applications in clinical diagnostics, therapeutics, and the fundamental understanding of molecular interactions. IMMREP is a community organized series of prediction contests that asks participants to predict TCR:pMHC binding on unpublished datasets. Previous iterations in 2022 and 2023 showed multiple approaches can predict TCR-pMHC binding with significant accuracy (median AUC_0.1≥0.7) for peptides where experimental data is available (“seen” peptides). In contrast, models did not outperform random guessing for peptides that have no such data available (“unseen” peptides). Here we report on the results of IMMREP25, which focused solely on unseen peptides in order to evaluate the cutting edge of the field. We received 126 named submissions predicting the specificity of 1,000 TCRs against twenty unseen peptides restricted by one of two MHC molecules (HLA-A*02:01 and HLA-B*40:01). The best performing methods showed a macro-AUC_0.1 of 0.60, significantly better than random, demonstrating significant advances in the field. The top performing methods incorporated structural modeling into their approach, indicating that especially for ‘unseen’ peptides, a structural understanding aids in the prediction of TCR:pMHC interactions. The results from this benchmark highlight the significant challenges remaining for TCR:pMHC predictions and will inform future method development.
Open AccessIMMREP25: Unseen Peptides
T cell receptors (TCRs) can bind to peptides presented by MHC molecules (pMHC) as a first step to trigger a T cell response. Reliable approaches to predict TCR:pMHC binding would have broad applications in clinical diagnostics, therapeutics, and the fundamental understanding of molecular interactions. IMMREP is a community organized series of prediction contests that asks participants to predict TCR:pMHC binding on unpublished datasets. Previous iterations in 2022 and 2023 showed multiple approaches can predict TCR-pMHC binding with significant accuracy (median AUC_0.1≥0.7) for peptides where experimental data is available (“seen” peptides). In contrast, models did not outperform random guessing for peptides that have no such data available (“unseen” peptides). Here we report on the results of IMMREP25, which focused solely on unseen peptides in order to evaluate the cutting edge of the field. We received 126 named submissions predicting the specificity of 1,000 TCRs against twenty unseen peptides restricted by one of two MHC molecules (HLA-A*02:01 and HLA-B*40:01). The best performing methods showed a macro-AUC_0.1 of 0.60, significantly better than random, demonstrating significant advances in the field. The top performing methods incorporated structural modeling into their approach, indicating that especially for ‘unseen’ peptides, a structural understanding aids in the prediction of TCR:pMHC interactions. The results from this benchmark highlight the significant challenges remaining for TCR:pMHC predictions and will inform future method development.
- Open Access
Targeting peptide–MHC complexes with designed T cell receptors and antibodies
Class I major histocompatibility complexes (MHCs), expressed on the surface of all nucleated cells, present peptides derived from intracellular proteins for surveillance by T cells. The precise recognition of foreign or mutated peptide–MHC (pMHC) complexes by T cell receptors (TCRs) is central to immune defense against pathogens and tumors. Although patient-derived TCRs specific for cancer-associated antigens have been used to engineer tumor-targeting therapies, their reactivity toward self- or near-self antigens may be constrained by negative selection in the thymus. Here, we introduce a structure-based deep learning framework, ADAPT (Antigen-receptor Design Against Peptide-MHC Targets), for the design of TCRs and antibodies that bind to pMHC targets of interest. We evaluate the ADAPT pipeline by designing and characterizing TCRs and antibodies against a diverse panel of pMHCs. Cryogenic electron microscopy structures of two designed antibodies bound to their respective pMHC targets demonstrate atomic-level accuracy at the recognition interface, supporting the robustness of our structure-based approach. Computationally designed TCRs and antibodies targeting pMHC complexes could enable a broad range of therapeutic applications, from cancer immunotherapy to autoimmune disease treatment, and insights gained from TCR–pMHC design should advance predictive understanding of TCR specificity with implications for basic immunology and clinical diagnostics.
Open AccessTargeting peptide–MHC complexes with designed T cell receptors and antibodies
Class I major histocompatibility complexes (MHCs), expressed on the surface of all nucleated cells, present peptides derived from intracellular proteins for surveillance by T cells. The precise recognition of foreign or mutated peptide–MHC (pMHC) complexes by T cell receptors (TCRs) is central to immune defense against pathogens and tumors. Although patient-derived TCRs specific for cancer-associated antigens have been used to engineer tumor-targeting therapies, their reactivity toward self- or near-self antigens may be constrained by negative selection in the thymus. Here, we introduce a structure-based deep learning framework, ADAPT (Antigen-receptor Design Against Peptide-MHC Targets), for the design of TCRs and antibodies that bind to pMHC targets of interest. We evaluate the ADAPT pipeline by designing and characterizing TCRs and antibodies against a diverse panel of pMHCs. Cryogenic electron microscopy structures of two designed antibodies bound to their respective pMHC targets demonstrate atomic-level accuracy at the recognition interface, supporting the robustness of our structure-based approach. Computationally designed TCRs and antibodies targeting pMHC complexes could enable a broad range of therapeutic applications, from cancer immunotherapy to autoimmune disease treatment, and insights gained from TCR–pMHC design should advance predictive understanding of TCR specificity with implications for basic immunology and clinical diagnostics.
- Open Access
- 16 October 2025
- Published by National Academy of Sciences in Proceedings of the National Academy of Sciences
- Vol. 122 (42)
T cell receptor specificity landscape revealed through de novo peptide design
T cells play a key role in adaptive immunity by mounting specific responses against diverse pathogens. Effective bindings between T cell receptors (TCRs) and pathogen derived peptides presented on major histocompatibility complexes (MHCs) mediate immune responses. However, predicting these interactions remains challenging due to limited functional data on T cell reactivities. Here, we introduce a computational approach to predict TCR interactions with peptides presented on MHC-I alleles, and to design immunogenic peptides for specified TCR–MHC complexes. Our method leverages HERMES, a structure-based machine learning model trained on the protein universe to predict amino acid preferences based on local structural environments. Despite no direct training on TCR-pMHC data, HERMES’s implicit physical reasoning enables us to make accurate predictions of both TCR–pMHC binding affinities and T cell activities across diverse viral and cancer epitopes, achieving up to 0.72 correlation with experimental data. Leveraging our TCR recognition model, we develop a computational protocol for de novo design of immunogenic peptides. Through experimental validation in three TCR–MHC systems, we demonstrate that our designs—with up to five substitutions from the native sequence—activate T cells at success rates of up to 50%. Last, we use our generative framework to quantify the diversity of the peptide recognition landscape for various TCR–MHC’s, offering key insights into T cell specificity. Our approach provides a platform for immunogenic peptide and neoantigen design, as well as for evaluating TCR specificity, offering a computational framework to inform design of engineered T cell therapies and vaccines.
- 16 October 2025
- Published by National Academy of Sciences in Proceedings of the National Academy of Sciences
- Vol. 122 (42)
Open AccessT cell receptor specificity landscape revealed through de novo peptide design
T cells play a key role in adaptive immunity by mounting specific responses against diverse pathogens. Effective bindings between T cell receptors (TCRs) and pathogen derived peptides presented on major histocompatibility complexes (MHCs) mediate immune responses. However, predicting these interactions remains challenging due to limited functional data on T cell reactivities. Here, we introduce a computational approach to predict TCR interactions with peptides presented on MHC-I alleles, and to design immunogenic peptides for specified TCR–MHC complexes. Our method leverages HERMES, a structure-based machine learning model trained on the protein universe to predict amino acid preferences based on local structural environments. Despite no direct training on TCR-pMHC data, HERMES’s implicit physical reasoning enables us to make accurate predictions of both TCR–pMHC binding affinities and T cell activities across diverse viral and cancer epitopes, achieving up to 0.72 correlation with experimental data. Leveraging our TCR recognition model, we develop a computational protocol for de novo design of immunogenic peptides. Through experimental validation in three TCR–MHC systems, we demonstrate that our designs—with up to five substitutions from the native sequence—activate T cells at success rates of up to 50%. Last, we use our generative framework to quantify the diversity of the peptide recognition landscape for various TCR–MHC’s, offering key insights into T cell specificity. Our approach provides a platform for immunogenic peptide and neoantigen design, as well as for evaluating TCR specificity, offering a computational framework to inform design of engineered T cell therapies and vaccines.
- Open Access
- 30 July 2025
- Published by Oxford University Press (OUP) in Molecular Biology and Evolution
- Vol. 42 (8)
A Sitewise Model of Natural Selection on Individual Antibodies via a Transformer–Encoder
During affinity maturation, antibodies are selected for their ability to fold and to bind a target antigen between rounds of somatic hypermutation. Previous studies have identified patterns of selection in antibodies using B cell repertoire sequencing data. However, these studies are constrained by needing to group many sequences or sites to make aggregate predictions. In this paper, we develop a transformer–encoder selection model of maximum resolution: given a single antibody sequence, it predicts the strength of selection on each amino acid site. Specifically, the model predicts for each site whether evolution will be slower than expected relative to a model of the neutral mutation process (purifying selection) or faster than expected (diversifying selection). We show that the model does an excellent job of modeling the process of natural selection on held out data, and does not need to be enormous or trained on vast amounts of data to perform well. The patterns of purifying vs diversifying natural selection do not neatly partition into the complementarity-determining vs framework regions: for example, there are many sites in framework that experience strong diversifying selection. There is a weak correlation between selection factors and solvent accessibility. When considering evolutionary shifts down a tree of antibody evolution, affinity maturation generally shifts sites towards purifying natural selection, however this effect depends on the region, with the biggest shifts toward purifying selection happening in the third complementarity-determining region. We observe distinct evolution between gene families but a limited relationship between germline diversity and selection strength.
- 30 July 2025
- Published by Oxford University Press (OUP) in Molecular Biology and Evolution
- Vol. 42 (8)
Open AccessA Sitewise Model of Natural Selection on Individual Antibodies via a Transformer–Encoder
During affinity maturation, antibodies are selected for their ability to fold and to bind a target antigen between rounds of somatic hypermutation. Previous studies have identified patterns of selection in antibodies using B cell repertoire sequencing data. However, these studies are constrained by needing to group many sequences or sites to make aggregate predictions. In this paper, we develop a transformer–encoder selection model of maximum resolution: given a single antibody sequence, it predicts the strength of selection on each amino acid site. Specifically, the model predicts for each site whether evolution will be slower than expected relative to a model of the neutral mutation process (purifying selection) or faster than expected (diversifying selection). We show that the model does an excellent job of modeling the process of natural selection on held out data, and does not need to be enormous or trained on vast amounts of data to perform well. The patterns of purifying vs diversifying natural selection do not neatly partition into the complementarity-determining vs framework regions: for example, there are many sites in framework that experience strong diversifying selection. There is a weak correlation between selection factors and solvent accessibility. When considering evolutionary shifts down a tree of antibody evolution, affinity maturation generally shifts sites towards purifying natural selection, however this effect depends on the region, with the biggest shifts toward purifying selection happening in the third complementarity-determining region. We observe distinct evolution between gene families but a limited relationship between germline diversity and selection strength.
- Open Access
TCR2HLA: calibrated inference of HLA genotypes from TCR repertoires enables identification of immunologically relevant metaclonotypes
T cell receptors (TCRs) recognize peptides presented by polymorphic human leukocyte antigen (HLA) molecules, but HLA genotype data are often missing from TCR repertoire sequencing studies. To address this, we developed TCR2HLA, an open-source tool that infers HLA genotypes from TCRβ repertoires. Expanding on work linking public TRBV-CDR3 sequences to HLA genotypes, we incorporated “quasi-public” metaclonotypes – composed of rarer TCRβ sequences with shared amino acid features – enriched by HLA genotypes. Using four TCRβseq datasets from 3,150 individuals, we applied TRBV gene partitioning and locality-sensitive hashing to identify ∼96,000 TCRβ features strongly associated with specific HLA alleles from 71M input TCRs. Binary HLA classifiers built with these features achieved high balanced accuracy (>0.9) across common HLA-A (9/12), B (9/12), C (6/13), DRB1 (11/11) alleles and prevalent DPA1/DPB1 (6/10), DQA1/DQB1 (8/17) heterodimers. We also introduced a high-sensitivity calibration to support predictions in samples with as few as 5,000 unique clonotypes. Calibrated predictions with confidence filtering improved reliability. Beyond genotype imputation, TCR2HLA enables the discovery of novel HLA- and exposure-associated TCRs, as shown by the identification of SARS-CoV-2 related TCRs in a large COVID-19 dataset lacking HLA data. TCR2HLA provides a scalable framework for bridging the gap between TCRseq data and HLA genotype for biomarker discovery.
Open AccessTCR2HLA: calibrated inference of HLA genotypes from TCR repertoires enables identification of immunologically relevant metaclonotypes
T cell receptors (TCRs) recognize peptides presented by polymorphic human leukocyte antigen (HLA) molecules, but HLA genotype data are often missing from TCR repertoire sequencing studies. To address this, we developed TCR2HLA, an open-source tool that infers HLA genotypes from TCRβ repertoires. Expanding on work linking public TRBV-CDR3 sequences to HLA genotypes, we incorporated “quasi-public” metaclonotypes – composed of rarer TCRβ sequences with shared amino acid features – enriched by HLA genotypes. Using four TCRβseq datasets from 3,150 individuals, we applied TRBV gene partitioning and locality-sensitive hashing to identify ∼96,000 TCRβ features strongly associated with specific HLA alleles from 71M input TCRs. Binary HLA classifiers built with these features achieved high balanced accuracy (>0.9) across common HLA-A (9/12), B (9/12), C (6/13), DRB1 (11/11) alleles and prevalent DPA1/DPB1 (6/10), DQA1/DQB1 (8/17) heterodimers. We also introduced a high-sensitivity calibration to support predictions in samples with as few as 5,000 unique clonotypes. Calibrated predictions with confidence filtering improved reliability. Beyond genotype imputation, TCR2HLA enables the discovery of novel HLA- and exposure-associated TCRs, as shown by the identification of SARS-CoV-2 related TCRs in a large COVID-19 dataset lacking HLA data. TCR2HLA provides a scalable framework for bridging the gap between TCRseq data and HLA genotype for biomarker discovery.
- Open Access
Diverse modes of T cell receptor sequence convergence define unique functional and cellular phenotypes
Single-cell techniques allow concurrent study of gene activity and T cell receptor (TCR) sequences, identifying connections between TCR structure and cell traits. Expanding on our CoNGA software, we present a “metaCoNGA” analysis of 6 million T cells from 91 diverse studies, mapping TCR sequence similarity across tissues and diseases. This approach exposes shared TCR features within specific T cell subsets, including those associated with infection, cancer, and autoimmunity. We introduce a method to identify T cell groups with similar gene expression and biased TCR amino acid composition, providing a systematic framework for classifying diverse unconventional T cells, including KIR+ CD8+ T cells, CD4+ regulatory T cells, and subsets of NKT and MAIT cells. A new TCR clustering approach identifies thousands of convergent TCR sequence clusters hypothesized to target shared antigens. These clusters show coherent gene expression, highlighting the role of antigen exposure in shaping T cell behavior. Finally, we provide a tool for users to merge new data with this resource and rapidly identify T cell features in their data sets. This resource empowers investigations into the complex relationship between TCR sequence and T cell function in human health.
Open AccessDiverse modes of T cell receptor sequence convergence define unique functional and cellular phenotypes
Single-cell techniques allow concurrent study of gene activity and T cell receptor (TCR) sequences, identifying connections between TCR structure and cell traits. Expanding on our CoNGA software, we present a “metaCoNGA” analysis of 6 million T cells from 91 diverse studies, mapping TCR sequence similarity across tissues and diseases. This approach exposes shared TCR features within specific T cell subsets, including those associated with infection, cancer, and autoimmunity. We introduce a method to identify T cell groups with similar gene expression and biased TCR amino acid composition, providing a systematic framework for classifying diverse unconventional T cells, including KIR+ CD8+ T cells, CD4+ regulatory T cells, and subsets of NKT and MAIT cells. A new TCR clustering approach identifies thousands of convergent TCR sequence clusters hypothesized to target shared antigens. These clusters show coherent gene expression, highlighting the role of antigen exposure in shaping T cell behavior. Finally, we provide a tool for users to merge new data with this resource and rapidly identify T cell features in their data sets. This resource empowers investigations into the complex relationship between TCR sequence and T cell function in human health.
- Open Access
- 20 March 2025
- Published by Oxford University Press (OUP) in Nucleic Acids Research
- Vol. 53 (6)
Statistical analysis of repertoire data demonstrates the influence of microhomology in V(D)J recombination
V(D)J recombination generates the diverse B and T cell receptors essential for recognizing a wide array of antigens. This diversity arises from the combinatorial assembly of V(D)J genes and the junctional deletion and insertion of nucleotides. While previous in vitro studies have shown that microhomology—short stretches of sequence homology between gene ends—can bias the recombination process, the extent of microhomology’s impact in vivo, particularly in humans, remains unknown. In this paper, we assess how germline-encoded microhomology influences trimming and ligation during V(D)J recombination using statistical inference on previously published high-throughput TCRα repertoire sequencing data. We find that microhomology increases both trimming and ligation probabilities, making it an important predictor of recombination outcomes. These effects are consistent across other receptor loci and sequence types. Further, we demonstrate that accounting for germline microhomology effects significantly alters sequence annotation probabilities and rankings, highlighting its practical importance for accurately inferring the V(D)J recombination events that generated an observed sequence. Together, these results enhance our understanding of how germline-encoded microhomologous nucleotides shape the human V(D)J recombination process.
- 20 March 2025
- Published by Oxford University Press (OUP) in Nucleic Acids Research
- Vol. 53 (6)
Open AccessStatistical analysis of repertoire data demonstrates the influence of microhomology in V(D)J recombination
V(D)J recombination generates the diverse B and T cell receptors essential for recognizing a wide array of antigens. This diversity arises from the combinatorial assembly of V(D)J genes and the junctional deletion and insertion of nucleotides. While previous in vitro studies have shown that microhomology—short stretches of sequence homology between gene ends—can bias the recombination process, the extent of microhomology’s impact in vivo, particularly in humans, remains unknown. In this paper, we assess how germline-encoded microhomology influences trimming and ligation during V(D)J recombination using statistical inference on previously published high-throughput TCRα repertoire sequencing data. We find that microhomology increases both trimming and ligation probabilities, making it an important predictor of recombination outcomes. These effects are consistent across other receptor loci and sequence types. Further, we demonstrate that accounting for germline microhomology effects significantly alters sequence annotation probabilities and rankings, highlighting its practical importance for accurately inferring the V(D)J recombination events that generated an observed sequence. Together, these results enhance our understanding of how germline-encoded microhomologous nucleotides shape the human V(D)J recombination process.