Protein similarity search with subset seeds on a dedicated reconfigurable hardware

International audienceGenome sequencing of numerous species raises the need of complete genome comparison with precise and fast similarity searches. Today, advanced seed-based techniques (spaced seeds, multiple seeds, subset seeds) provide better sensitivity/specificity ratios. We present an implementation of such a seed-based technique onto parallel specialized hardware embedding reconfigurable architecture (FPGA), where the FPGA is tightly connected to large capacity Flash memories. This parallel system allows large databases to be fully indexed and rapidly accessed. Compared to traditional approaches like the Blastp software, we obtain both significant speed-up and better results. As our knowledge, this is the first attempt to exploit modern seed features for parallelizing similarity search