- Article -
| In Silico Biology 4, 0037 (2004); ©2004, Bioinformation Systems e.V. |
Recognition of remotely related structural homologues using sequence profiles of aligned homologous protein structures
Seema Namboori, Narayanaswamy Srinivasan* and Shashi B. Pandit
Molecular Biophysics Unit, Indian Institute of Science, Bangalore 560 012, India
* corresponding author; Email: ns@mbu.iisc.ernet.in
Phone: +91-80-2293 2837; Fax: +91-80-2360 0535
Edited by E. Wingender; received April 25, 2004; revised June 17 and July 22, 2004; accepted July 24; published August 06, 2004
Abstract
In order to bridge the gap between proteins with three-dimensional (3-D) structural information and those without 3-D structures, extensive experimental and computational efforts for structure recognition are being invested. One of the rapid and simple computational approaches for structure recognition makes use of sequence profiles with sensitive profile matching procedures to identify remotely related homologous families. While adopting this approach we used profiles that are generated from structure-based sequence alignment of homologous protein domains of known structures integrated with sequence homologues. We present an assessment of this fast and simple approach. About one year ago, using this approach, we had identified structural homologues for 315 sequence families, which were not known to have any 3-D structural information. The subsequent experimental structure determination for at least one of the members in 110 of 315 sequence families allowed a retrospective assessment of the correctness of structure recognition. We demonstrate that correct folds are detected with an accuracy of 96.4% (106/110). Most (81/106) of the associations are made correctly to the specific structural family. For 23/106, the structure associations are valid at the superfamily level. Thus, profiles of protein families of known structure when used with sensitive profile-based search procedure result in structure association of high confidence. Further assignment at the level of superfamily or family would provide clues to probable functions of new proteins. Importantly, the public availability of these profiles from us could enable one to perform genome wide structure assignment in a local machine in a fast and accurate manner.
Key words: protein fold, superfamily, profiles, protein family, fold recognition, structural genomics, remote homologues