Section: New Results

Modeling Interfaces and Contacts

Docking, scoring, interfaces, protein complexes, scoring functions, Voronoi diagrams, arrangements of balls.

Modeling Macro-molecular Complexes : a Journey Across Scales

Participants : Frédéric Cazals, Tom Dreyfus.

In collaboration with C. Robert (IBPC / CNRS, Paris, France).

While proteins and nucleic acids are the fundamental components of an organism, Biology itself is based on the interactions they make with each other. Analyzing macromolecular interactions typically requires handling systems involving from two to hundreds of polypeptide chains. After a brief overview of the modeling challenges faced in computational structural biology, the text [16] reviews concepts and tools aiming at improving our understanding of the link between the static structures of macromolecular complexes and their biophysical/biological properties. We discuss geometrical approaches suited to atomic-resolution complexes and to large protein assemblies; for each, we also present examples of their successful application in quantifying and interpreting biological data. This methodology includes state-of-the-art geometric analyses of surface area, volume, curvature, and topological properties (isolated components, cavities, voids, cycles) related to Voronoi constructions in the context of structure analysis. On the applied side, we present novel insights into real biological problems gained thanks to these modeling tools.

CSA: Comprehensive Comparison of Pairwise Protein Structure Alignments

Participant : Noël Malod-Dognin.

In collaboration with I. Wohlers (CWI / VU University Amsterdam, Netherlands), R. Andonov (Irisa / Rennes University, France), G.W. Klau (CWI / VU University Amsterdam, Netherlands).

Protein structural alignment is a key method for answering many biological questions involving the transfer of information from well-studied proteins to less well-known proteins. Since structures are more conserved during evolution than sequences, structural alignment allows for the most precise mapping of equivalent residues. Many structure-based scoring schemes have been proposed and there is no consensus on which scoring is the best. Comparative studies also show that alignments produced by different methods can differ considerably. Based on the alignment engine derived from A_purva, we designed CSA (Comparative Structural Alignment), the first web server for computation, evaluation and comprehensive comparison of pairwise protein structure alignments at single residue level [15] . It offers the exact computation of alignments using the scoring schemes of DALI, Contact Map Overlap (CMO), MATRAS and PAUL. In CSA, computed or uploaded alignments can be explored in terms of many inter-residue distances, RMSD, and sequence-based scores. Intuitive visualizations also help in grasping the agreements and differences between alignments. The user can thus make educated decisions about the structural similarity of two proteins and, if necessary, post-process alignments by hand. CSA is available at http://csa.project.cwi.nl .

Upon publication [15] , CSA was selected by Nucleic Acids Research as featured article of July 2012 (top 5% of papers in terms of originality, significance and scientific excellence.