Improving scoring functions for protein-protein docking with LambdaLoss

Abstract

Modeling protein-protein interactions requires accurate scoring functions that can rank potential poses (conformations) of a protein-protein complex to differentiate near-native poses from incorrect ones. Here, we propose a general framework for improving protein-protein pose ranking and other biomolecular interaction models using the LambdaLoss loss function from the Learning-to-Rank field. We test this framework by fine-tuning the energy prediction head of DFMDock with the LambdaLoss on an augmented dataset of 2.9M decoy poses derived from the DIPS dataset. On targets from the CAPRI score set benchmark, our fine-tuned ranking model LambdaDockScore is better at identifying correct poses in its top-1 and top-5 predictions compared to EuDockScore, a state-of-the-art method. LambdaDockScore also improves upon baseline DFMDock ranking performance for scoring antibody-antigen complexes and protein-protein complexes with very large or small binding interfaces.

Unified Sampling and Ranking for Protein Docking With DFMDock
Lee‐Shin Chu, Sudeep Sarma, Da Xu +1
Proteins Structure Function and Bioinformatics6 Aug 2026Peer-reviewedLicence not statedFull text7 citedMolecular biology & gene regulationOpenAlex
GroScore: Automated Scoring of Protein–Protein Interactions Using Bidirectional Steered Molecular Dynamics
Jan Walther Perthold, Dražen Petrov, Chris Oostenbrink
Journal of Chemical Information and Modeling2 Oct 2026Peer-reviewedCC BY 4.0Full textMolecular biology & gene regulationPMC
Structural insights into an aptainhibitor of the SARS-CoV-2 E protein using molecular docking, molecular dynamics, and biophysical analyses
Miyssa Imam Abdelmageed Mohamed, Jin-Pyo Lee, Woo-Ri Shin +6
Molecular therapy. Nucleic acids23 Sep 2026Peer-reviewedCC BY-NC-NDFull textMolecular biology & gene regulationEurope PMC
PTM-PAL Pipeline: A Python Web Platform of Integrative Scoring System for Assessing the Functional Impact of Protein S-Palmitoylation.
Xin-Yu Bi, Jia-Ni Chen, Xiong-Fei Pan +1
Membranes14 Oct 2026Peer-reviewedCC BYFull textMolecular biology & gene regulationEurope PMC
Harmonic Torsional Diffusion for Protein-Ligand Flexible Docking
Maksim Zhdanov, Pavel Strashnov, Vladislav Kurenkov
arXiv18 Oct 2026PreprintCC BY 4.0Full textBiophysicsarXiv

Cite

BibTeX RIS

BibTeX

@misc{zhu2026improving,
  title = {{Improving scoring functions for protein-protein docking with LambdaLoss}},
  author = {Richard Zhu and Darren Xu and Lee-Shin Chu and Jeffrey J. Gray},
  howpublished = {arXiv},
  year = {2026},
  month = oct,
  doi = {10.48550/arxiv.2610.00191},
  eprint = {2610.00191},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/2610.00191},
  note = {CC BY 4.0},
}

RIS

TY  - UNPB
TI  - Improving scoring functions for protein-protein docking with LambdaLoss
AU  - Richard Zhu
AU  - Darren Xu
AU  - Lee-Shin Chu
AU  - Jeffrey J. Gray
PY  - 2026
DA  - 2026/10/01
PB  - arXiv
DO  - 10.48550/arxiv.2610.00191
UR  - https://arxiv.org/abs/2610.00191
AB  - Modeling protein-protein interactions requires accurate scoring functions that can rank potential poses (conformations) of a protein-protein complex to differentiate near-native poses from incorrect ones. Here, we propose a general framework for improving protein-protein pose ranking and other biomolecular interaction models using the LambdaLoss loss function from the Learning-to-Rank field. We test this framework by fine-tuning the energy prediction head of DFMDock with the LambdaLoss on an augmented dataset of 2.9M decoy poses derived from the DIPS dataset. On targets from the CAPRI score set benchmark, our fine-tuned ranking model LambdaDockScore is better at identifying correct poses in its top-1 and top-5 predictions compared to EuDockScore, a state-of-the-art method. LambdaDockScore also improves upon baseline DFMDock ranking performance for scoring antibody-antigen complexes and protein-protein complexes with very large or small binding interfaces.
ER  -

Abstract

Related

Cite