Parallelization of local BLAST service on workstation clusters

摘要: Abstract This paper describes approaches to improve the performance of one most common and increasingly important aspects Human Genome Project (HGP) — large-volume, batch comparison DNA sequence data. basic operation, usually carried out by well-known BLAST program on subject against internationally available databases nearly five million target sequences, is already used hundreds thousands times each day researchers around world. At present, it still primarily in single query, or small query mode. As entire human genome nears completion, area functional genomics, use micro-arrays sets genes, coming fore. These developments will demand ever more efficient means BLASTing data that make processor implementation powerful workstations infeasible. We describe three primary parallel components BLAST. The first at sequence-to-sequence level. second parallelizes a across partitioned distributed database. Finally, set queries themselves are servers with replicated databases. methods may be employed alone concert. Our current described which requests, our plans for other levels also described. results ultimately applied hardware assistance this soon-to-be primitive computer operation.