Convey Computer said this week that it has implemented a version of the Burrows-Wheeler aligner algorithm on its hybrid-core computing platform that increased genome reference mapping rates by a factor of 15 compared to commodity servers.
This latest algorithm adds to Convey's suite of bioinformatics tools, which includes Convey GraphConstructor used for de novo short read assembly (BI 5/20/2011) and its implementation of the Smith-Waterman algorithm for local sequence alignment (BI 11/20/2009).
Bruce Toal, Convey's CEO and co-founder, said in a statement that the firm's latest release both “improves efficiency" and "allows researchers to get results faster."
That’s because, according to the firm, software-only applications are limited by the performance of commodity servers that execute a stream of general-purpose instructions. Convey's hybrid-core architecture, on the other hand, pairs Intel x86 microprocessors with a field programmable gate array-based coprocessor to allow important segments of applications — such as traversal of a Burrows-Wheeler index, for instance — to run directly in hardware.
Furthermore, where bioinformatics algorithms are hampered by access to memory on traditional architectures, Convey’s highly parallel memory subsystem allows application-specific logic to concurrently access 8,192 individual words in memory, thus increasing the effective memory bandwidth over cache-based memory systems, the company said.