An Efficient Parallel Algorithm for Global Sequence Alignment on Multi-cores

Due to the gigantic explosion in the amount of molecular data, a high-quality global sequence alignment (GA) becomes a big dilemma nowadays. the similarity matrix computations phase is the most consuming time and space. This paper addresses this issue by proposing the most common parallel implementation of the algorithm Needleman Wunch (NW) by breaking the similarity matrix to blocks. The basis is the grid scheme of the algorithm, which allows both save memory and parallelize calculations. The results have shown, that the memory complexity was reduced from quadratic to linear, and the speedup of more than 5 was obtained when the algorithm was implemented by C++ language on core-i7-8550U, with the biological dataset sequence with lengths 300k.