Section: New Results

HPC and Parallelism

Participants : Rumen Andonov, Guillaume Chapuis, Charles Deltel, Dominique Lavenier, Fabrice Legeai, François Moreews.

• High performance pipelines for annotation. We participed to effort of URGI (INRA Versailles) to set up TriAnnot, a modular architecture allowing for the annotation of genomes. The TriAnnot pipeline is parallelized on a 712 CPU computing cluster that can run a 1-Gb sequence annotation in less than 5 days. [13]

• Bioinformatics Workflows. SLICEE is an environment to capture and parallelize time-consuming bioinformatics applications on grid or cloud platforms. In 2012, a web interface has been designed to interactively draw and run workflows from standard browsers ([workflow portal] ). Several workflows used in the BioWIC ANR project have been successfully tested on this platform (http://biowic.inria.fr/ )

• Parallelization of a pseudo-clique solver. Following such solvers as DAST and A_purva, we develop a pseudo-clique solver for alignment graphs. Looking for pseudo-cliques allows us to relax some of the constraints that are inherent to clique finding and thus maintain polynomial run times. We focus on defining a parallel algorithm and developping an implementation that benefits from multiple levels of parallelism: fine grain parallelism (bit-level parallelism, SSE instructions) and coarse grain parallelism (mutli-core parallelism). Intended applications range from protein local similarity search to protein surface similarity search or even docking.