A fast method for linear space pairwise sequence alignment

Pairwise sequence alignment is an important technique for finding the optimal arrangement between two sequences. A basic dynamic-programming strategy for sequence alignment needs O(mn) time and also O(mn) space. Hirschberg's divide-and-conquer method reduces the required space to roughly 2m (m <= n), but it also doubles the computing time. On the other hand, the FastLSA approach adds extra rows and columns to generalize Hirschberg's algorithm, reducing the number of recomputations at the cost of more memory space. In this paper, we shall present an efficient linear space algorithm, called the NFLSA algorithm, to reduce the ratio of recalculation greater than FastLSA while using the same memory. The NFLSA algorithm is superior to Hirschberg's algorithm and the FastLSA algorithm in our simulations.