Leakage-controlled benchmarking reveals generalization limits of deep learning for protein-ligand binding affinity prediction
- 1University of Missouri - Columbia
Abstract
To address widespread data leakage and inconsistent evaluation in protein-ligand affinity prediction, we introduce PLABench, a leakage-controlled and target-centric benchmarking framework that enables standardized comparison across sequence- and structure-based models. We benchmark nine deep learning methods across blind CASP16 targets, leakage-controlled ChEMBL35 sets, and established Davis and KIBA datasets under rigorous data split settings, standardizing structural input via AlphaFold3 to ensure fair comparison. Although pretrained structure-based models achieve the highest overall accuracy, they show severe target-dependent variability, and increasing structural fidelity from predicted to experimental conformations yields no consistent gains. Meanwhile, sequence-based approaches surpass some structure-based methods on select targets, and protein family-level evaluations reveal uneven performance across families and substantial inter-model complementarity obscured by aggregate metrics. These findings demonstrate that training scale and structural input alone cannot guarantee cross-target generalization, highlighting the need for context-aware interaction modeling. PLABench provides an extensible open-source platform to facilitate these developments.
Related
- Benchmarking Large Language Models for Assay-Aware Drug-Target Binding Affinity Prediction
- Benchmarking Peptide-Protein Affinity Prediction Across Peptide and Target Shifts
- Active Learning framework for Antibody Binding Affinity Prediction
- MIRAGE: Measuring Interpolation and Redundancy in Affinity GEneralization
- PMO-Dock: Benchmarking Docking, Specificity, and Generalization in Molecular Optimization
BibTeX
@misc{wang2026leakagecontrolled,
title = {{Leakage-controlled benchmarking reveals generalization limits of deep learning for protein-ligand binding affinity prediction}},
author = {L. Wang and J. Cheng},
howpublished = {bioRxiv},
year = {2026},
month = sep,
doi = {10.64898/2026.09.17.752433},
url = {https://www.biorxiv.org/content/10.64898/2026.09.17.752433v1},
note = {CC BY-NC-ND},
}
RIS
TY - UNPB
TI - Leakage-controlled benchmarking reveals generalization limits of deep learning for protein-ligand binding affinity prediction
AU - L. Wang
AU - J. Cheng
PY - 2026
DA - 2026/09/17
PB - bioRxiv
DO - 10.64898/2026.09.17.752433
UR - https://www.biorxiv.org/content/10.64898/2026.09.17.752433v1
AB - To address widespread data leakage and inconsistent evaluation in protein-ligand affinity prediction, we introduce PLABench, a leakage-controlled and target-centric benchmarking framework that enables standardized comparison across sequence- and structure-based models. We benchmark nine deep learning methods across blind CASP16 targets, leakage-controlled ChEMBL35 sets, and established Davis and KIBA datasets under rigorous data split settings, standardizing structural input via AlphaFold3 to ensure fair comparison. Although pretrained structure-based models achieve the highest overall accuracy, they show severe target-dependent variability, and increasing structural fidelity from predicted to experimental conformations yields no consistent gains. Meanwhile, sequence-based approaches surpass some structure-based methods on select targets, and protein family-level evaluations reveal uneven performance across families and substantial inter-model complementarity obscured by aggregate metrics. These findings demonstrate that training scale and structural input alone cannot guarantee cross-target generalization, highlighting the need for context-aware interaction modeling. PLABench provides an extensible open-source platform to facilitate these developments.
ER -