Counterfactual Peptide Editing for Causal TCR--pMHC Binding Inference

Sanjar Khudoyberdiev; Arman Bekov

cs.LGApr 14, 2026

Counterfactual Peptide Editing for Causal TCR--pMHC Binding Inference

Authors: Sanjar Khudoyberdiev, Arman Bekov

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 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.

Figures & tables

Split	Train	Val	Test	Pos. rate
Random	118 742	16 963	33 927	0.050
Family-HO	92 314	11 082	21 843	0.042
Distance-aware	87 508	10 501	19 624	0.039

TABLE I: Dataset and Split Summary

Split	Method	AUROC	AUPRC	ECE $\downarrow$	BS $\downarrow$	NLL $\downarrow$
FHO	Baseline	0.779	0.421	0.138	0.091	0.219
	+Edit Aug	0.798	0.447	0.112	0.084	0.197
	CIP (ours)	0.831	0.491	0.083	0.073	0.171
DA	Baseline	0.758	0.399	0.151	0.097	0.234
	+Edit Aug	0.779	0.428	0.121	0.089	0.211
	CIP (ours)	0.812	0.463	0.095	0.079	0.186

TABLE II: Main Results. Full-coverage evaluation.

\uparrow

: higher is better;

\downarrow

: lower is better.

Method	OOD AUROC $\uparrow$	SI $\downarrow$	CFC $\uparrow$	AFR $\uparrow$
Baseline	0.741	0.412	0.518	0.231
+Edit Aug	0.768	0.311	0.634	0.287
CIP (ours)	0.803	0.249	0.724	0.401

TABLE III: Causal Diagnostic Metrics under Family-Held-Out Split. SI

\downarrow

: Shortcut Index; CFC

\uparrow

: Counterfactual Consistency; AFR

\uparrow

: Anchor Flip Rate.

Configuration	AUROC	AUPRC	CFC $\uparrow$	AFR $\uparrow$
CIP (full)	0.831	0.491	0.724	0.401
w/o $\mathcal{L}_{\text{sens}}$	0.816	0.473	0.701	0.264
w/o $\mathcal{L}_{\text{inv}}$	0.808	0.461	0.591	0.388
w/o anchor masking	0.819	0.477	0.662	0.319
w/o BLOSUM constraint	0.822	0.479	0.689	0.371
Baseline (no auxiliary loss)	0.779	0.421	0.518	0.231

TABLE IV: Ablation Study (FHO split). Best in bold .

Counterfactual Data Augmentation

Counterfactual data augmentation creates plausible examples by changing a chosen feature while preserving relevant context, helping models learn robust distinctions.

See all 38 papers

Oct 19, 2026

EMBER: Predicting TCR--pMHC Binding Without a Docked Complex

We propose EMBER, a structure-based framework for TCR--pMHC binding prediction that eliminates the need for a predicted docked complex, and with it the placement error that dominates such predictions. Traditional structure-based predictors utilize the co-folded complex structure of the receptor and the antigen and interpret binding from its contact surface. Consequently, every downstream feature inherits any deviation the structure-prediction tool introduced. Methodologically, EMBER folds the TCR and pMHC independently, encodes them as SE(3)-invariant residue graphs, and compares their learned representations without constructing a shared coordinate frame. A relative pose is therefore not an input of the model, so placement error cannot propagate into the prediction. Computationally, EMBER preprocesses

O(N{+}M)

monomer folds where a complex-based predictor needs

O(NM)

operations. Experimentally, under epitope-disjoint evaluation with six retrained baselines, EMBER is first on every metric under all four folding tools, by 0.9 to 14.4% in micro-AUROC, 18.5 to 25.3% in micro-AUPRC and 10.6 to 24.1% in MCC. Its ranking is also up to 85% more stable across folding tools.

Lin Ha, Chen Zhou, Haozheng Luo +3

Northwestern University

Protein-Protein Interaction Prediction Protein Representation Learning

Ask PDF

Jun 29, 2026

Structure-Regularized Interpretable TCR-Epitope Prediction

T cell receptor (TCR)-epitope binding prediction is essential for understanding adaptive immunity and developing immunotherapies. Existing sequence- and structure-based models often generalize poorly to unseen epitopes and provide limited interpretability. Furthermore, the impact of generated structures on model learning remains unclear. We present TCR-SRIM, a structure-regularized interpretable-by-design model that combines protein language model embeddings with interpretable contact prototypes to capture residue-level TCR-epitope interactions. TCR-SRIM achieves state-of-the-art predictive performance and improved interpretation quality on the TCR-XAI benchmark. Using its inherent interpretability, we further evaluate the effect of generated structures on model learning. While structures predicted by AlphaFold3, TCRModel2, and tFold-TCR yield competitive performance, they lead to less accurate interaction patterns and reduced binding-site diversity than experimentally-resolved structures. Our results highlight limitations of current structure prediction models for TCR-epitope learning and demonstrate the value of interpretable-by-design models for studying generated biological structures.

Jiarui Li, Zixiang Yin, Yunbei Zhang +4

Department of Computer Science, Tulane University · Department of Biochemistry and Molecular Biology, Tulane University School of Medicine

Explainable Artificial Intelligence Interpretable ML

Ask PDF

Jun 3, 2026

New Benchmarking Shows Limited Generalization Power of TCR Antigenic Epitope Prediction Models

Accurate computational prediction of T cell receptor (TCR) antigen specificity would transform the study of T cell biology and enable scalable immune engineering, yet existing models lack sufficient sensitivity and specificity for broad applications. A major limitation is the absence of rigorously defined, unseen benchmark datasets that allow unbiased evaluation of model performance and generalizability. Here, we describe two complementary classes of datasets that meet this criterion and argue that they provide both a robust framework for model assessment and a foundation for next-generation TCR-antigen prediction algorithm development.

Yiming Liao, Yiheng Li, Ning Jiang +2

Trustworthy and Intelligent Computing Lab (TAIC), Department of Computer Science and Electrical Engineering, University of Maryland, Baltimore County, Baltimore, Maryland 21250, USA · Children’s Hospital of Philadelphia, Philadelphia, PA 19104, USA · Department of Bioengineering, University of Pennsylvania, Philadelphia, PA 19104, USA +5

Cross-Dataset Generalization

Ask PDF

Abstract

Counterfactual Data Augmentation

Explore similar work

EMBER: Predicting TCR--pMHC Binding Without a Docked Complex

Structure-Regularized Interpretable TCR-Epitope Prediction

New Benchmarking Shows Limited Generalization Power of TCR Antigenic Epitope Prediction Models