Section: New Results

Structural Systems Biology

Participants : Marie-Dominique Devignes, Anisah Ghoorah, Van-Thai Hoang, Bernard Maigret, David Ritchie, Malika Smaïl-Tabbone.

Keywords:

bioinformatics, chemistry, docking, knowledge discovery, screening, systems biology

Structural systems biology aims to describe and analyze the many components and interactions within living cells in terms of their three-dimensional (3D) molecular structures. We are currently developing advanced computing techniques for molecular shape representation, protein-protein docking, protein-ligand docking, high-throughput virtual drug screening, and knowledge discovery in databases dedicated to protein-protein interactions.

Accelerating protein docking calculations using graphics processors

We have recently adapted the Hex protein docking software [113] to use modern graphics processors (GPUs) to carry out the expensive FFT part of a docking calculation [114] . Compared to using a single conventional central processor (CPU), a high-end GPU gives a speed-up of 45 or more. This software is publicly available at http://hex.loria.fr . A public GPU-powered server has also been created (http://hexserver.loria.fr ) [106] . The docking server has performed some 14,000 docking runs during 2013.

Our docking work has facilitated further developments on modeling the assembly of multi-component molecular structures using a particle swarm optimization technique [123] , and on modeling protein flexibility during docking [122] . In 2013, in collaboration with the Nano-D team at Inria Grenoble, we developed a new docking algorithm called “DockTrina” [31] , which can rapidly model trimers of protein structures by combining multiple pair-wise docking results from Hex. We also used Hex successfully to model a challenging protein complex containing water molecules at the protein-protein interface [29] .

KBDOCK: Protein docking using Knowledge-Based approaches

In order to explore the possibilities of using structural knowledge of protein-protein interactions, Anisah Ghoorah recently developed the KBDOCK system as part of her doctoral thesis project [95] . KBDOCK is available at http://kbdock.loria.fr . KBDOCK combines coordinate data from the Protein Data Bank [87] with the Pfam protein domain family classification [91] in order to describe and analyze all known protein-protein interactions for which the 3D structures are available. We have demonstrated the utility of KBDOCK [94] for template-based docking using 73 complexes from the Protein Docking Benchmark [97] . We recently presented results obtained using KBDOCK at the CAPRI conference on protein docking in Utrecht [21] . In 2013, we updated KBDOCK with the latest data from Pfam and the Protein Data Bank. An article describing the new version of KBDOCK was accepted by the Database Issue of Nucleic Acids Research [6] .

Kpax: A new algorithm for protein structure alignment

We have developed a new protein structure alignment approach called Kpax [112] . The approach exploits the fact that each amino acid residue has a carbon atom with a highly predictable tetrahedral geometry. This allows the local environment of each residue to be transformed into a canonical orientation, thus allowing easy comparison between the canonical orientations of residues within pairs of proteins using a novel scoring function based on Gaussian overlaps. The overall approach is two or three orders of magnitude faster than most contemporary protein structure alignment algorithms, while still being almost as accurate as the state-of-the-art TM-Align approach [124] . The Kpax program is available at http://kpax.loria.fr/ . The Kpax program is now used heavily behind the scenes in the new KBDOCK web server [6] to find structural templates for docking which might be beyond the reach of sequence-based homology modeling approaches.

gEMpicker and gEMfitter: GPU-accelerated tools for cryo-electron microscopy

Solving the structures of large protein assemblies is a difficult and computationally intensive task. Multiple two-dimensional (2D) images must be processed and classified to identify protein particles in different orientations. These images may then be averaged and stacked to deduce the three-dimensional (3D) structure of a protein. In order to help accelerate the first of these tasks we have recently developed a novel and highly parallel algorithm called “gEMpicker” which uses multiple graphics processors to detecting 2D particles in cryo-electron microscopy images [112] . We have also developed a 3D shape matching algorithm called “gEMfitter” which also exploits graphics processors, and which will provide a useful tool for the final 3D assembly step [112] . Both programs have been made publicly available at http://gem.loria.fr/ .

DOVSA: Developing new algorithms for virtual screening

In 2010, Violeta Pérez-Nueno joined the Orpailleur team thanks to a Marie Curie Intra-European Fellowship (IEF) award to develop new virtual screening algorithms (DOVSA). The aim of this project was to advance the state of the art in computational virtual drug screening by developing a novel consensus shape clustering approach based on spherical harmonic (SH) shape representations [110] . As a continuation of this project, and in collaboration with colleagues from the University of Bari in Italy, we recently published a review on drug discovery relating to the GPCR receptor proteins [15] . We also published a book chapter describing the ParaFit program for fast spherical harmonic shape matching [70] .