SEAL: a divide-and-conquer approach for sequence alignment

Sequence similarity search and sequence alignment methods are fundamental steps in comparative genomics and have a wide spectrum of application in the field of medicine, agriculture, and environment. The dynamic programming sequence alignment methods produce optimal alignments but are impractical for a similarity search due to their large running time. Heuristic methods like BLAST run much faster but may not provide optimal alignments. In this paper, we introduce a novel sequence alignment algorithm, SEAL. SEAL is a parallelizable algorithm that does not require gap penalty parameter as in heuristic methods. SEAL uses a combination of divide-and-conquer paradigm and the maximum contiguous subarray solution. SEAL is also improved by the use of borders in every contiguous segment. The alignment scores obtained by SEAL are consistently higher than those obtained by heuristic methods. Since the dependencies are minimized among intermediate steps, the complexity of SEAL can be reduced to (Formula presented.) in the presence of satisfactory number of parallel processors. © 2015, Springer-Verlag Wien.