Accelerating a charged single α-helix search algorithm in protein sequences using FPGA

Inhalt:
Processing and analyzing the immense amount of biological data generated each day requires fast and accurate bioinformatics algorithms. The main characteristic of the these algorithms is the low precision input data which can be stored on 2-5bits. Therefore reconfigurable architectures can be used more efficiently compared to conventional 32 or 64 bit microprocessors, because the accuracy of the functional units can be customized and a relatively small amount of configurable logic is required. In the paper a new architecture is introduced and implemented on FPGA to speed-up a structural motif search algorithm. The architecture can be implemented on a versatile medium range FPGA providing about two orders of magnitude speedup. The system can be efficiently scaled to process several sequences in parallel on a larger FPGA resulting in even higher speedup.