DPAM Domain Classification of Human Proteins against ECOD Reference

Domain partitions and ECOD (Evolutionary Classification of protein Domains) classifications for AlphaFold Database models of the human proteome, plus raw DPAM domain results for five comparison proteomes (Mus musculus, Danio rerio, Drosophila melanogaster, Caenorhabditis elegans, Pan paniscus). This is the data set behind Schaeffer et al. (2023) PNAS 120:e2214069120 (doi:10.1073/pnas.2214069120).

Version 2 (2026) keeps the version 1 domain definition files unchanged and adds the classification as served on the ECOD website (prodata.swmed.edu/ecod/index_human.php, now retired), domain structures and sequences, and the full database behind the site.

Contents: 47,576 human domains (46,610 assigned to ECOD T-groups, 966 unassigned after manual inspection) from 18,037 proteins.

• ecod_human.domains.tsv: one row per domain, with protein annotation, residue range, classification method and full ECOD lineage (main table)
• ecod.af2_human.pdb.tar.gz: domain coordinates (PDB, pLDDT in the B-factor column)
• ecod.af2_human.fa.tar.gz and ecod.af2_human.missing_seq.fa: domain sequences
• ecod.af2_human.domains.txt: the domain list as distributed from the website
• ecod_human.tables.tar.zst and ecod_human.pgdump: the full PostgreSQL database, as per-table TSV and as a pg_dump
• HomSa_combined_domains, *_raw_domains, 0README: version 1 files (2022)

See README.md for column definitions, provenance and known differences between files. The domain structures are derived from AlphaFold DB models (CC BY 4.0).