Paradigme d'inference bayesienne pour la Biologie structurale in silico

The project bip:bip promotes an integrated use of known and modelled three-dimensional structures of proteins as an aid in complete genome annotation. The architectural organization of biological macromolecules is essential for the understanding and manipulation of their function. In parallel, more and more experimental techniques (transcriptomics, proteomics, high-resolution microscopies, ...) are now providing new and detailed insights into cellular processes. However, each method gives only access to a small fraction of the information necessary for a complete view of each system. Connecting them efficiently will be instrumental for proper simulations of living organisms.

The project follows a multi-scale and multi-techniques approach for integrative structural bioinformatics: (i) from the structure of proteins alone to protein-ligand complexes and large multi-protein assemblies, (ii) from isolated structures to dynamically relevant structure ensembles and (iii) from individual proteins to whole families and entire proteomes.

The project brings together different fields in modern bioinformatics that we think will mostly benefit from novel multi-scale solutions. The algorithmic framework for is Bayesian statistics, in order to make inferences in an optimal way and to obtain rigorous estimates of the reliability of the results. The developed methods will be made available to the researchers with the help of the French bioinformatics platform, and by deposition of results in dedicated databases