Vector alignment search tool (VAST) automated protein structure comparison using special structural elements

The rapid growth in the size of protein structure databases has led to the development of numerous protein structural comparison algorithms. These protein structure comparison methods can reveal structural and functional similarities from remote homologs which cannot be detected by sequence information alone. They can also reveal evolutionary relationships and provide information for phylogenetic studies. The Vector Alignment Search Tool (VAST), developed by the Structure Group at the National Center for Biotechnology Information (NCBI), offers a fast and efficient way to perform structural similarity comparisons. The VAST algorithm uses vectors to represent secondary structure elements (SSEs) in protein structures and the structural comparison step relies on the relative orientation, length and alignment of those vectors. Comparisons that align only one or two vectors between two proteins are never considered significant. As a result, the presence of at least three SSEs is a prerequisite in performing structure similarity searches using VAST. Many small proteins which have fewer than three secondary structure elements are therefore not comparable by VAST. Furthermore, slight variations in a pair of structures can occasionally result in large differences in vector representations. For proteins with only a few SSEs, this can sometimes cause comparisons between similar structures to be judged insignificant. We have constructed an extension of the VAST algorithm that improves comparison results by incorporating other conserved special structural elements such as disulfide bond information. This enhances the current VAST algorithms by improving structure alignment for proteins which do not contain conventional SSEs or those which have only a few SSEs, thus allowing otherwise non-comparable small proteins to be included in the database. Our results demonstrate that VAST performance has been improved by the implementation of this new structural element. This more accurate structural comparison algorithm has the potential to help provide further insight into the relationship between structure and the function of proteins. VAST structural neighbor information is available on the Internet at: www.ncbi.nlm.nih.gov/Structure.