Technical Highlight - January 2015
Short description: An expanded ModBase comparative protein structure database and toolkit enables insights into protein function, including determinants of HIV-1 protease specificity.
The rise in experimentally determined structures has been outpaced by a swell of protein sequence information, creating a need for robust structural prediction. ModBase is a database of annotated comparative protein structure models for protein sequences that match at least one known structure.
Sali and colleagues (PSI NYSGRC and EFI) have updated ModBase with new features and expanded structural coverage to include almost 30 million models for domains found in 4.7 million unique protein sequences. Researchers can browse existing predictions or take advantage of tools offered through the ModWeb server.
Predictions are based on the ModPipe pipeline, which identifies and aligns structures that match a given target sequence, and then builds and assesses the models. ModPipe uses Modeller, PSI-BLAST or HHBlits and HHSearch for fold assignment and target-template alignment. Associated servers enable structural alignment, loop and protein-protein interaction modeling, structural annotation of human non-synonymous single nucleotide polymorphisms and multiple-sequence and structure alignment.
New functionality includes the use of statistically optimized atomic potentials scores to assess protein loops and docking interfaces in ModPipe. AllosMod predicts conformational transitions such as allosteric changes and ligand-induced protein dynamics, and three web tools incorporate small-angle X-ray scattering (SAXS) data: AllosMod-FoXS finds conformations with best fits to SAXS profiles, FoXSDock filters protein docking models with a SAXS profile and SAXS Merge merges profiles. Pose & Rank uses statistical potential to score protein–ligand complexes.
The authors used ModPipe to identify 100 optimal targets whose structures would more than double the modelable fraction of the human α-helical transmembrane proteome. The authors also assessed the specificity of HIV-1 protease, a target for anti-retroviral drugs. HIV-1 protease cleaves the Gag and Pol polypeptides during virion maturation as well as human proteins bearing an eight–amino acid non-homologous sequence. Based on the structural features of octopeptides that are cleaved or not in vitro, they identified structural determinants of HIV-1 protease activity and generated a ModBase dataset bearing annotated models of 118 in vitro human protein targets.
U. Pieper et al. ModBase, a database of annotated comparative protein structure models and associated resources.
Nucleic Acids Res., D336-46 (2014). doi:10.1093/nar/gkt1144