Seminar

Protein-protein docking in the interactomics era: integrating theoretical modeling and low-resolution structural data

Dr. Juan Fernández-Recio

Protein interaction networks are controlling the majority of cell processes, and therefore their study has attracted the interest of many inter-disciplinary experimental and theoretical areas. Among them, computational docking aims to predict the structure of a protein-protein complex based only on their individual components, and thus can contribute to push structural information into proteomics projects. Our approach pyDock, which evaluates the binding energy based on atomic solvation parameters, shows good predictive rates in rigid-body docking cases, as assessed in the recent community-wide CAPRI experiments
(http://www.ebi.ac.uk/msd-srv/capri/). Indeed, we have already run large-scale docking experiments in the yeast proteome. However, before attempting a systematic application to complete interactomes, major challenges need to be solved. For instance, one bottleneck in docking is the treatment of conformational flexibility during binding. One line we are exploring is the use of pre-existing native conformations from NMR-based unbound ensembles as starting coordinates for docking. Related to this, we need to efficiently sample and score millions of conformers during docking, which will necessarily require high-performance computing and coarser-grained scoring alternatives. Thus, we have derived new coarse-grained potentials, based on statistical interface propensities from known protein-protein interfaces (SIPPER) and have developed simpler molecular models for scoring and refinement (pyDockCG). Another major challenge for proteomic applications is the efficient conversion of low-resolution structural data of protein complexes into useful models. In this line, we have shown that the combination of SAXS data and docking predictions can overcome the limitations of both techniques when applied independently. We believe this is an exciting time for the study and prediction of protein-protein interactions, and the complementarity of theoretical and experimental approaches will undoubtedly bring major advances in the field.