SAbDab2 Machine Learning Dataset

Abstract

Clean, curated antibody and antibody–antigen-complex structures from SAbDab2, specifically post-processed for ML applications, alongside standardised train/test-splits. The current release contains 15,810 structures from 8,796 PDB IDs, corresponding to 5,424 unique antibodies (SAbDab2 IDs), composed as follows: antibody type with antigen no antigen total structures paired-chain (FV-like) 8230 (52.1%) 4137 (26.2%) 12367 (78.2%) single-domain heavy (VHH-like) 2083 (13.2%) 1235 (8.0%) 3318 (21.0%) single-domain light (VL-like) 37 (0.2%) 27 (0.2%) 64 (0.4%) VNAR 45 (0.3%) 16 (0.1%) 61 (0.4%) total structures 10395 (65.7%) 5415 (34.3%) 15810 (100.0%) Each structure file contains a single paired-chain or single-chain antibody, cropped to the variable region, alongside any antigen chains. Around 7.9% of structures bind antigens consisting of multiple polymer chains. We provide two distinct train/test splits of these structures, both based on sequence similarity. The ab-split (ab_split.csv) accounts for similarity between antibody sequences. To avoid data leakage via the antigen, use this only for antigen-agnostic tasks like prediction of antibody structure in solution. The ab-ag-split (abag_split.csv) additionally considers similarity between any protein, peptide, DNA and RNA antigen sequences. Use this for antigen-aware tasks like antibody–antigen complex modelling. These same splits are also available filtered down to just single-domain (VHH-like and VL-like) structures (ab_split_sd.csv and abag_split_sd.csv). Unless otherwise noted, splits are backward-compatible, allowing models trained on previously published versions of this dataset to be benchmarked on the most recent test split. More details in the README.md accompanying this download. How to Cite If you use this data in your work, please cite the specific data release with its corresponding DOI (see right-hand "Versions" panel), alongside the SAbDab2 publication: ```bibtex @article {Capel2026.06.16.732554, author = { Capel, Henriette L. and Vavourakis, Odysseas and Williams,Benjamin H. and Taylor, Christopher }, title = {{SAbDab2}: The structural antibody database in the age of machine learning}, elocation-id = {2026.06.16.7325543}, year = {2026}, doi = (10.64898/2026.06.16.732554}, publisher = {Cold Spring Harbor Laboratory}, URL = {https://www.biorxiv.org/content/early/2026/06/20/2026.06.16.732554}, eprint = (https://www.biorxiv.org/content/early/2026/06/20/2026.06.16.732554.full.pdf}, journal = {bioRxiv}} ```