pvSOAR: detecting similar surface patterns of pocket and void surfaces of amino acid residues on proteins

Abstract

Detecting similar protein surfaces provides an important route for discovering unrecognized or novel functional relationship between proteins. The web server pvSOAR (pocket and void Surfaces Of Amino acid Residues) provides an online resource to identify similar protein surface regions. pvSOAR can take a structure either uploaded by a user or obtained from the Protein Data Bank, and identifies similar surface patterns based on geometrically defined pockets and voids. It provides several search modes to compare protein surfaces by similarity in local sequence, local shape and local orientation. Statistically significant search results are reported for visualization and interactive exploration. pvSOAR can be used to predict biological functions of proteins with known three-dimensional structures but unknown biological roles. It can also be used to study functional relationship between proteins and for exploration of the evolutionary origins of structural elements important for protein function. We present an example using pvSOAR to explore the biological roles of a protein whose structure was solved by the structural genomics project. The pvSOAR web server is available at http://pvsoar.bioengr.uic.edu/.

Figure 1

Interactive environment for the visualization of pvSOAR search results. This example depicts a pocket surface pattern of (A) protein BioH from E.coli (CASTp pocket id = 35, PDB id = 1m33) and (D) haloperoxidase (CASTp pocket id = 33, PDB id = 1a8s) from Pseudomonas fluorescens. The superposition of conserved pocket residues and the superposition after projection on to a unit sphere are shown in (B) and (C), respectively. The alignment of sequence patterns of the pocket residues is shown in (E). These two surface patterns share strong similarity (cRMSD P-value = 1.073 × 10⁻³, oRMSD P-value = 3.836 × 10⁻⁶), indicating that there may be functional similarity between BioH protein and haloperoxidase.

Figure 2

Screenshot of statistically significant search results between all surface patterns on myoglobin from Physeter catodon (PDB id = 109 m) and the CASTp database. Each surface has three histograms, reporting the distribution of significance score for E-values, P-value for cRMSD, and P-value for oRMSD. The drop-down box (A) allows for interactively setting up a threshold value to filter hits presented.