G. Allen et al. (Eds.): ICCS 2009, Part I, LNCS 5544, pp. 954–963, 2009. © Springer-Verlag Berlin Heidelberg 2009 Pairwise Distance Matrix Computation for Multiple Sequence Alignment on the Cell Broadband Engine Adrianto Wirawan, Bertil Schmidt, and Chee Keong Kwoh School of Computer Engineering, Nanyang Technological University, Singapore 639798 {adri0004,asbschmidt,asckkwoh}@ntu.edu.sg Abstract. Multiple sequence alignment is an important tool in bioinformatics. Although efficient heuristic algorithms exist for this problem, the exponential growth of biological data demands an even higher throughput. The recent emergence of accelerator technologies has made it possible to achieve a highly improved execution time for many bioinformatics applications compared to general-purpose platforms. In this paper, we demonstrate how the PlayStation®3, powered by the Cell Broadband Engine, can be used as a computational platform to accelerate the distance matrix computation utilized in multiple sequence alignment algorithms. Keywords: multiple sequence alignment, cell broadband engine. 1 Introduction Multiple sequence alignment (MSA) of many nucleotides or amino acids is an important tool in bioinformatics. It can identify patterns or motifs to characterize protein families, and is therefore utilized to detect homology between sequences as well as to perform phylogenetic analysis. Many MSA heuristics have been proposed to reduce the exponential complexity of computing optimal MSAs. Heuristic MSA implementations include MSA[1], ClustalW[2], T-Coffee[3], MAFFT[4], DIALIGN P[5] and PRALINE[6]. ClustalW[2] has over 26,000 citations in the ISI Web of Science and is considered to be one of the most popular MSA tools. It is based on the progressive alignment method. Although not optimal, this method can produce reasonably good alignments at a good efficiency. However, the exponential growth of biological data demands an even better throughput. Thus, software approaches to improve the performance of ClustalW have been introduced, including caching[8, 9] and parallel processing[10-12]. The recent emergence of accelerator technologies such as FPGAs, GPUs and specialized processors have made it possible to achieve an improvement in execution time for many bioinformatics applications compared to current general-purpose platforms at a low cost. Recent usage of easily accessible accelerator technologies to improve the ClustalW algorithm include FPGA[13] and GPU[14]. Our profiling of ClustalW has revealed that distance matrix computation is the most time consuming stage and typically takes up more than 90% of the overall runtime.