Beyond AlphaFold 3: a standardised benchmark of open-source co-folding models for antibody–antigen complexes, from docking accuracy to interaction specificity

Co-folding models are increasingly used to predict antibody-antigen complexes for epitope mapping, candidate triage and structure-guided de novo design or engineering, yet the latest generation of these models has not been compared under standardized conditions. We benchmark AlphaFold 3, Protenix-v2, ESMFold2, ESMFold2-fast and OpenDDE on 200 test-set complexes (121 Fvs and 79 VHHs) from SAbDab2-Bench, a newly curated dataset, and a reconstructed antibody-antigen subset of FoldBench, Ab-FoldBench. Using standardized inputs, sampling, and interface metrics that isolate the antibody–antigen interface, we compare docking accuracy, inference-time scaling, confidence-based candidate structure selection and sensitivity to interaction perturbations. OpenDDE achieved the highest median DockQ and docking pass rate, followed by Protenix-v2, although their difference was not statistically significant. Both models outperformed AlphaFold 3 on these measures, alongside offering fewer licensing restrictions on model and prediction use. We further find that model confidence was generally effective at identifying accurate predictions but showed model-dependent variations. To probe interaction specificity, which is essential for antibody design, we generated decoys through non-conservative interface mutations, and non-cognate antigen shuffling. Confidence decreased for both perturbations, but more consistently for non-cognate antigens than for interface mutations, with OpenDDE retaining comparatively high confidence for many decoys. Overall, Protenix-v2 combined high docking accuracy with strong rejection of perturbed interfaces, OpenDDE achieved the highest docking accuracy, albeit at the longest runtime, while ESMFold2-fast offered the fastest inference for large-scale triage. These results provide an up-to-date guidance framework for selecting and configuring co-folding models in antibody discovery, design, and engineering.

Authors

Institutions

Publication Details

Journal
mAbs
Published
2026-11-12
DOI
https://doi.org/10.1080/19420862.2026.2760737
Primary Topic
Protein Structure and Dynamics
Type
article
Field-Weighted Citation Impact
0.00

Funders

Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
article

Beyond AlphaFold 3: a standardised benchmark of open-source co-folding models for antibody–antigen complexes, from docking accuracy to interaction specificity

Pietro Sormanni, Hongjie Zhao, Aubin Ramon, Arnav Cheruku
mAbs
Protein Structure and Dynamics
article

Beyond AlphaFold 3: a standardised benchmark of open-source co-folding models for antibody–antigen complexes, from docking accuracy to interaction specificity

Pietro Sormanni, Hongjie Zhao, Aubin Ramon, Arnav Cheruku
article en

Abstract

Co-folding models are increasingly used to predict antibody-antigen complexes for epitope mapping, candidate triage and structure-guided de novo design or engineering, yet the latest generation of these models has not been compared under standardized conditions. We benchmark AlphaFold 3, Protenix-v2, ESMFold2, ESMFold2-fast and OpenDDE on 200 test-set complexes (121 Fvs and 79 VHHs) from SAbDab2-Bench, a newly curated dataset, and a reconstructed antibody-antigen subset of FoldBench, Ab-FoldBench. Using standardized inputs, sampling, and interface metrics that isolate the antibody–antigen interface, we compare docking accuracy, inference-time scaling, confidence-based candidate structure selection and sensitivity to interaction perturbations. OpenDDE achieved the highest median DockQ and docking pass rate, followed by Protenix-v2, although their difference was not statistically significant. Both models outperformed AlphaFold 3 on these measures, alongside offering fewer licensing restrictions on model and prediction use. We further find that model confidence was generally effective at identifying accurate predictions but showed model-dependent variations. To probe interaction specificity, which is essential for antibody design, we generated decoys through non-conservative interface mutations, and non-cognate antigen shuffling. Confidence decreased for both perturbations, but more consistently for non-cognate antigens than for interface mutations, with OpenDDE retaining comparatively high confidence for many decoys. Overall, Protenix-v2 combined high docking accuracy with strong rejection of perturbed interfaces, OpenDDE achieved the highest docking accuracy, albeit at the longest runtime, while ESMFold2-fast offered the fastest inference for large-scale triage. These results provide an up-to-date guidance framework for selecting and configuring co-folding models in antibody discovery, design, and engineering.

mAbsVol. 18(1)
University of Cambridge (GB), Imperial College London (GB)
UK Research and Innovation, Directorate for Biological Sciences, Engineering and Physical Sciences Research Council, Biotechnology and Biological Sciences Research Council
Protein Structure and Dynamics
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Laika access coming soon