An integrated machine learning system to computationally screen protein databases for Protein binding peptide ligands.

Abstract

A fairly large set of protein interactions are mediated by families of peptide binding domains, such as SH2, SH3, PDZ, and MHC etc. To identify their ligands by experimental screening is not only labor intensive but almost futile in screening low abundant species, due to the suppression of high abundant species. The ideal way of studying protein-protein interactions is to use high-throughput computational approaches to screen protein sequence databases, direct the validating experiments towards the most promising peptides. Predictors with only good cross validation were not good enough to screen protein database. In this method, only information relevant to interaction was extracted; a family of domains and their ligands were collected and aligned respectively, then combined into a prediction system. An integrated machine learning systems was built using three novel coding methods, and screened the Swissprot and Genbank protein database for ligands of 10 SH3 and 3 PDZ domains. A large proportion of predictions have already been experimentally confirmed by other independent research groups, indicating a satisfying generalization capability in protein interaction identification.