SARFing the PDB

Fast growth of the number of the solved protein structures is increasing the role of their comparative analysis. In this paper I describe a new program, SARF2, for protein structure comparison and discuss new examples of the non-topological structural resemblance. SARF2 is designed to detect ensembles of secondary structure elements, which form similar spatial arrangements with possible different topological connections. The program is available to everyone through the World Wide Web (URL http://www-lmmb.ncifcrf.gov/similar to nicka/sarf2.html). The performance of the program is demonstrated by previously unnoticed cases of the significant similarities. One similarity discussed in this paper, between heme-binding proteins (cytochrome P450 and globin), consists of six alpha-helices, arranged into a globin fold. Another pair of structures (pectate lyase and snowdrop lectin) achieve similar beta-prism architecture through different topologies. The significance of these similarities is validated by (i) the distribution of a similarity score, (ii) the comparison of the aligned contact maps and/or (iii) the location of the active site. The observation of recurrent non-topological structural motifs implies their energetic stability and opens new possibilities for sequence-structure alignment (threading) methods.