Counterfactual Peptide Editing for Causal TCR--pMHC Binding Inference

TL;DR

This paper introduces CIP, a training framework that improves TCR-pMHC binding prediction by generating biologically constrained counterfactual peptide edits to reduce shortcut learning and enhance out-of-distribution robustness.

Contribution

CIP is a novel training method that enforces invariance to non-anchor peptide positions and sensitivity to anchor residues, improving causal TCR specificity modeling.

Findings

CIP achieves AUROC 0.831 on a challenging benchmark.

CIP reduces shortcut index by 39.7%.

Anchor-aware edits drive out-of-distribution gains.

Abstract

Neural models for TCR-pMHC binding prediction are susceptible to shortcut learning: they exploit spurious correlations in training data -- such as peptide length bias or V-gene co-occurrence -- rather than the physical binding interface. This renders predictions brittle under family-held-out and distance-aware evaluation, where such shortcuts do not transfer. We introduce \emph{Counterfactual Invariant Prediction} (CIP), a training framework that generates biologically constrained counterfactual peptide edits and enforces invariance to edits at non-anchor positions while amplifying sensitivity at MHC anchor residues. CIP augments the base classifier with two auxiliary objectives: (1) an invariance loss penalizing prediction changes under conservative non-anchor substitutions, and (2) a contrastive loss encouraging large prediction changes under anchor-position disruptions. Evaluated on a curated VDJdb-IEDB benchmark under family-held-out, distance-aware, and random splits, CIP achieves AUROC 0.831 and counterfactual consistency (CFC) 0.724 under the challenging family-held-out protocol -- a 39.7\% reduction in shortcut index relative to the unconstrained baseline. Ablations confirm that anchor-aware edit generation is the dominant driver of OOD gains, providing a practical recipe for causally-grounded TCR specificity modeling.