Exploring the Potential of Structure-Based Deep Learning Approaches for T cell Receptor Design

This repository contains data and results from the study titled "Exploring the Potential of Structure-Based Deep Learning Approaches for T cell Receptor Design". It includes the following compressed files: TCR_design_data_results: Contains data and results from the TCR design process with ProteinMPNN, ESM-IF1, and RosettaDesign, as well as design evaluation with TCRModel2 and Rosetta. ATLAS_Benchmark: Contains data, scripts, and results of Molecular Dynamics simulations and MM/PBSA calculations of modeled TCR:pMHC complexes (native and mutants with solved 3D structure) collected from ATLAS for the binding affinity benchmarking. ATLAS_Benchmark_models_[part1 and part2]: Contain data, scripts, and results of Molecular Dynamics simulations and MM/PBSA calculations of modeled TCR:pMHC complexes collected from ATLAS for the binding affinity benchmarking. MD_MMPBSA_ProteinMPNN_[part1, part2 and part3]: Contain the input and output files from the topology generation process, molecular dynamics simulations, and MM/PBSA calculations of CDR3 interface designs generated by ProteinMPNN and their corresponding Native complex (the first two parts) and ProteinMPNN designs of all CDR positions (part3). MD_MMPBSA_ESM-IF_[part1, part2 and part3]: Contain the input and output files from the topology generation process, molecular dynamics simulations, and MM/PBSA calculations of ESM-IF designs (the first two are CDR3 interface designs, and the third includes designs for all CDR positions). MD_example_additional_scripts: contains additional scripts used in the MD system preparation and trajectory analyses. OLGA_analysis: contains the probability of generation (pgen) values obtained from OLGA for native and TCR designs. Sequence_variability: contains the raw sequences of TCR V genes, J genes, and CDR3s obtained from IMGT and VDJDB. TCR_design_data_results_part2: Contains data and results from the TCR design process of all CDRs positions and all TCR positions with ProteinMPNN and ESM-IF1, as well as design evaluation with TCRModel2 and Rosetta. Solvent_accessibility: contains the raw data of relative solvent accessibility (RSA) estimates per residue for TCR:pMHC complexes in the test cases (MHC-I and MHC-II). For further details, refer to the README.md files within each directory. The abstract of the corresponding study is provided below: "Deep learning methods, trained on the increasing set of available protein 3D structures and sequences, have substantially impacted the protein modeling and design field. These advancements have facilitated the creation of novel proteins, or the optimization of existing ones designed for specific functions, such as binding a target protein. Despite the demonstrated potential of such approaches in designing general protein binders, their application in designing immunotherapeutics remains relatively underexplored. A relevant application is the design of T cell receptors (TCRs). Given the crucial role of T cells in mediating immune responses, redirecting these cells to tumor or infected target cells through the engineering of TCRs has shown promising results in treating diseases, especially cancer. However, the computational design of TCR interactions presents challenges for current physics-based methods, particularly due to the unique natural characteristics of these interfaces, such as low affinity and cross-reactivity. For this reason, in this study, we explored the potential of two structure-based deep learning protein design methods, ProteinMPNN and ESM-IF1, in designing fixed-backbone TCRs for binding target antigenic peptides presented by the MHC through different design scenarios. To evaluate TCR designs, we employed a comprehensive set of sequence- and structure-based metrics, highlighting the benefits of these methods in comparison to classical physics-based design methods and identifying deficiencies for improvement. "