An FPGA-Based Bioinformatics Accelerator and Application to Ultra-High-Throughput Sequencing
William Dally (Computer Science) CO-PI’s:
Arend Sidow (Genetics)
Serafim Batzoglou (Computer Science)

Ultra-high-throughput sequencing technology holds the promise of rapidly sequencing the genome of an individual to identify risk factors and personalize treatment. We propose to develop algorithms to assemble the output of ultra-high-throughput sequencing machines and for analysis of individualized genome sequence. The computation required, however, will be a bottleneck that may prevent this technology from seeing widespread clinical application. To remove this bottleneck, we propose to develop an FPGA-based hardware accelerator that will perform this computation 100x to 1000x faster than a conventional machine of similar cost. Our accelerator will be easily programmable using a high-level language (“C”) to enable the development of new algorithms and methods for processing sequence data. Our accelerator will also include internal storage for key data structures to eliminate the I/O bottleneck that has plagued previous accelerators.