Computational Tools for Exploration of the Energy Landscape in Protein-Protein Association
Albert Solernou, Juan Fernández‐Recio, Jesús Clemente-Gallardo, Pierpaolo Bruscolini, Francisco M. Castejón, Pablo Echenique, José Félix Sáenz-Lorenzo · AIP conference proceedings · 2008
We show here our work in structural prediction of protein‐protein interactions. Our computational docking methodology has two major components: the sampling of mutual orientations of the interacting molecules, and the scoring and clustering of these orientations for the identification of near‐native docking poses. Our procedure can generate a uniformly distributed set of rigid‐body docking poses, which can be easily extended to explore the flexibility of the side‐chains. The method is able to find near‐native orientations in line with other state‐of‐the‐art docking programs, and it has been successfully applied together with our pyDock scoring scheme in the most recent rounds of the CAPRI worl‐wide experiment (http://capri.ebi.ac.uk). We have also devised a new measure to compare rigid‐body docking poses based on angular distance (instead of RMSD), which describes the relative orientations of the two molecules in the different docking poses, as evaluated in a large benchmark of known protein‐protein cases. The low computational cost of this measure (as compared with the RMSD calculation) makes it ideal for clustering docking poses in a more effective way. The new docking angular index of similarity (DASI) can also be used to evaluate the validity of the docking predictions, when comparing with the X‐ray structures.