Moesch, A. | bioRxiv Science

bioRxiv2026

Generative AI Guided Design of High-Affinity T cell Receptors

Developing T cell receptors (TCRs) with sufficiently high affinity for tumor antigens (TAs) remains a fundamental challenge in TCR-T immunotherapy. Experimental methods such as affinity maturation and high-throughput screening have enabled the identification of TCRs with enhanced activity. However, their efficiency is often constrained by limited throughput, insufficient coverage, and the generally lower affinities of naturally occurring TCRs toward TAs. To address these challenges, we present TCRPPO2, an integrated AI-driven, in silico affinity maturation framework for peptide-specific TCR optimization. Using reinforcement learning, TCRPPO2 learns mutation policies that iteratively enhance the TCR binding affinity to the target peptide, derived from predictive models trained on carefully curated interaction data. The model is further augmented by a generative AI critic model that discourages implausible designs to ensure the biophysical validity. The designs are further screened by robust post-screening methods that leverage diverse functional annotations and physical prior knowledge. We applied TCRPPO2 to the clinically relevant MART-1 antigen and experimentally validated the designed candidates in Jurkat cell-based functional assays. Among the five engineered TCRs, all of which demonstrated positive cellular responses, three showed significantly increased activities relative to their templates and one showed substantial enhancement. These functional gains were consistent with more favorable interaction energy from structural and physical modeling. Together, our results support a generalizable paradigm for TCR engineering, in which learned mutation policies can efficiently navigate the peptide-specific binding landscape of TCRs and propose biologically enhanced candidates without explicit structural supervision, offering a practical route for early-stage computational TCR optimization for challenging tumor antigens.

immunology↗

bioRxiv2023

Assessing the Generalization Capabilities of TCR Binding Predictors via Peptide Distance Analysis

Understanding the interaction between T Cell Receptors (TCRs) and peptide-bound Major Histocompatibility Complexes (pMHCs) is crucial for comprehending immune responses and developing targeted immunotherapies. While recent machine learning (ML) models show remarkable success in predicting TCR-pMHC binding within training data, these models often fail to generalize to peptides outside of their training distributions, raising concerns about their applicability in therapeutic settings. Understanding and improving the generalization of these models is therefore critical to ensure real-world applications. To address this issue, we evaluate the effect of the distance between training and testing peptide distributions on ML model empirical risk assessments, using sequence-based and 3D structure-based distance metrics. In our analysis we use several state-of-the-art models for TCR-peptide binding prediction: Attentive Variational Information Bottleneck (AVIB), NetTCR-2.0 and -2.2, and ERGO II (pre-trained autoencoder) and ERGO II (LSTM). In this work, we introduce a novel approach for assessing the generalization capabilities of TCR binding predictors: the Distance Split (DS) algorithm. The DS algorithm controls the distance between training and testing peptides based on both sequence and structure, allowing for a more nuanced evaluation of model performance. We show that lower 3D shape similarity between training and test peptides is associated with a harder out-of-distribution task definition, which is more interesting when measuring the ability to generalize to unseen peptides. However, we observe the opposite effect when splitting using sequence-based similarity. These findings highlight the importance of using a distance-based splitting approach to benchmark models. This could then be used to estimate a confidence score on predictions on novel and unseen peptides, based on how different they are from the training ones. Additionally, our results may hint that employing 3D shape to complement sequence information could improve the accuracy of TCR-pMHC binding predictors.

immunology↗