TY - JOUR T1 - EC-PSI: Associating Enzyme Commission Numbers with Pfam Domains JF - bioRxiv DO - 10.1101/022343 SP - 022343 AU - Seyed Ziaeddin Alborzi AU - Marie-Dominique Devignes AU - David W. Ritchie Y1 - 2015/01/01 UR - http://biorxiv.org/content/early/2015/07/10/022343.abstract N2 - With the growing number of protein structures in the protein data bank (PDB), there is a need to annotate these structures at the domain level in order to relate protein structure to protein function. Thanks to the SIFTS database, many PDB chains are now cross-referenced with Pfam domains and enzyme commission (EC) numbers. However, these annotations do not include any explicit relationship between individual Pfam domains and EC numbers. This article presents a novel statistical training-based method called EC-PSI that can automatically infer high confidence associations between EC numbers and Pfam domains directly from EC-chain associations from SIFTS and from EC-sequence associations from the SwissProt, and TrEMBL databases. By collecting and integrating these existing EC-chain/sequence annotations, our approach is able to infer a total of 8,329 direct EC-Pfam associations with an overall F-measure of 0.819 with respect to the manually curated InterPro database, which we treat here as a “gold standard” reference dataset. Thus, compared to the 1,493 EC-Pfam associations in InterPro, our approach provides a way to find over six times as many high quality EC-Pfam associations completely automatically. ER -