PEP-SiteFinder: a tool for the blind identification of peptide binding sites on protein surfaces

Abstract

Peptide–protein interactions are important to many processes of life, particularly for signal transmission or regulatory mechanisms. When no information is known about the interaction between a protein and a peptide, it is of interest to propose candidate sites of interaction at the protein surface, to assist the design of biological experiments to probe the interaction, or to serve as a starting point for more focused in silico approaches. PEP-SiteFinder is a tool that will, given the structure of a protein and the sequence of a peptide, identify protein residues predicted to be at peptide–protein interface. PEP-SiteFinder relies on the 3D de novo generation of peptide conformations given its sequence. These conformations then undergo a fast blind rigid docking on the complete protein surface, and we have found, as the result of a benchmark over 41 complexes, that the best poses overlap to some extent the experimental patch of interaction for close to 90% complexes. In addition, PEP-SiteFinder also returns a propensity index we have found informative about the confidence of the prediction. The PEP-SiteFinder web server is available at http://bioserv.rpbs.univ-paris-diderot.fr/PEP-SiteFinder.

Fully Blind Peptide-Protein Docking with pepATTRACT

“…PEP-SiteFinder is based on the PTOOLS implementation of ATTRACT (Fiorucci and Zacharias, 2010;Saladin et al, 2009;Zacharias, 2003) and performs rigid body docking with peptide structures generated by the PEP-FOLD method ; it was benchmarked on 41 unbound-unbound complexes from the peptiDB set. PEP-SiteFinder identified at least 50% of the correct interface residues in the top ten poses in 71% of these cases (Saladin et al, 2014), whereas our protocols was able to achieve this for 85% of these 41 complexes. In sum, the pepATTRACT protocol performed very well in interface post-prediction.…”

“…Although good performance was found when using ATTRACT for peptide binding site prediction (Saladin et al, 2014), it has not yet been applied systematically to peptide-protein complexes. To test the performance of the ATTRACT force field with regard to sampling and scoring peptide-protein complexes, we first performed bound-bound rigid body docking for all cases yielding a theoretical limit for the performance of unbound-unbound docking.…”

“…Several groups have proposed that contact analysis of docking models can be used to predict the interface of the proteins (interface post-prediction) (Ferná ndez- Recio et al, 2004;Hwang et al, 2010b;Lensink and Wodak, 2010a;Sacquin-Mora et al, 2008;Saladin et al, 2014;de Vries and Bonvin, 2011). We thus also wanted to evaluate how well the binding site was predicted regardless of the peptide conformation.…”

“…A number of peptide-protein docking and binding site prediction tools have been developed to date (Antes, 2010;Bordner and Abagyan, 2006;Dagliyan et al, 2011;Donsky and Wolfson, 2011;Dundas et al, 2006;Heté nyi and van der Spoel, 2002;Lavi et al, 2013;Luitz and Zacharias, 2014;Niv and Weinstein, 2005;Petsalaki et al, 2009;Raveh et al, 2011;Rosenfeld et al, 1995;Saladin et al, 2014;Staneva and Wallin, 2009;Trabuco et al, 2012;Unal et al, 2010;Verschueren et al, 2013). Global docking and binding site prediction methods (Ben-Shimon and Eisenstein, 2010;Dagliyan et al, 2011;Dundas et al, 2006;Lavi et al, 2013;Petsalaki et al, 2009;Saladin et al, 2014;Trabuco et al, 2012;Verschueren et al, 2013) often identify the correct binding site but do not yield high-quality models for the peptide conformation (London et al, 2013b). The ligand docking approach Autodock was adapted to fully blind peptide docking, but is limited to short fragments (Heté nyi and van der Spoel, 2002;London et al, 2013b;Unal et al, 2010).…”

Role of computer-aided drug design in modern drug discovery

“…The binding site of a small molecule compound can be predicted using available tools such as (Somarowthu and Ondrechen 2012), and many others. Alternatively, specific approaches that can identify peptide binding sites include PepSite (Trabuco et al 2012), PeptiMap (Lavi et al 2013), and PEPSiteFinder (Saladin et al 2014). Lastly, difficulties in detecting allosteric sites can be alleviated by recently developed open-access web servers such as SPACER (Goncearenco et al 2013), MCPath (Kaya et al 2013), Allosite (Huang et al 2013), and PARS (Panjkovich and Daura 2014).…”