Book of Abstracts: Albany 2007Conversation 15 June 19-23 2007 New structural classification uncovers common sequence motifs for proteins from different superfamiliesWe describe a novel method for determining common sequence pattern for proteins with non-similar amino acid sequences, but similar tertiary structures. The two main steps of this method are: 1) grouping together structurally similar proteins; and 2) developing a multiple sequence alignment of proteins in each such group. Here we focus on a large set of beta sandwich?like proteins.
To uncover underlying structural similarity among proteins, we introduce a unit of supersecondary structure termed ?strandon?. Strandon is defined as a set of sequentially consecutive strands connected by hydrogen bonds in 3D structure. Representing proteins as an assembly of strandons allowed us to reveal the rules that define the spatial arrangement of strands in the beta sheets and to developed a new supersecondary structure classification that reflects these supersecondary regularities. The new classification is presented in the SuperSecondary Structure (SSS) database (http://binfs.umdnj.edu/sssdb/). It is important to emphasize that classification of proteins in SSS database does not rely on sequence homology, but only on structural similarities. Thus, proteins grouped together in SSS database can have very different amino acid sequences, and often belong to different families and superfamilies. The second step of our program - multiple sequence alignment for proteins with non-similar amino acid sequences - is based on alignment of residues that form inter-strand hydrogen bonds. For each group of proteins we found a number of conserved positions, about 30% of sequence length, occupied by few closely related residues. This set of residues could be considered to be the structural determinants responsible for the 3D shape of the protein molecules. The main conclusion of this work is that a group of proteins with non-similar amino acid sequences can share common structural determinants, if they have the same arrangement of the strands. References and Footnotes
Alexander E. Kister Dept of Health Informatics, SHRP, |