Section: Scientific Foundations

High Performance Computing

HPC for bioinformatics aims to bring efficient computing solutions in the two following challenging areas: processing of high throughput genomic data and processing of computational intensive algorithms. These two areas have in common to be highly time-consuming, but for different reasons. The first has to handle very huge amounts of data while the second has to solve very large optimization problems. More precisely, the first area required to organize, to structure, to index, and more generally, to manipulate very large data structures, making memory management issue a real challenge. The second area involves high complexity algorithms, but on a much more reduced data set.

In both cases, space and time limitations can be pushed away by the use of parallel machines as they can provide large aggregated memory space and/or high computational power. We believe that the design of parallel and optimized parallel algorithms is a key issue to face the avalanche of genomic data, and that all forms of parallelism must be exploited, from cloud computing to hardware accelerators such as GPGPU.

Corresponding disciplinary fields are optimization, parallelism, processing mass of data, hardware accelerator and advanced indexing structures.