Reduced-Search Dynamic Programming for Linear-Space Sequence Alignment

Abstract

Dynamic programming (DP) algorithms for the sequence alignment can easily incorporate the length of sequences, gap penalties, backtracking preference, and alignment score which present difficulties for algorithms based on linear or nonlinear programming formulations and for many sequence alignment heuristics. However, exact DP algorithms for the sequence alignment need exponential storage and computational time and can only be applied to small problems. In this paper, a heuristic algorithm is proposed for applying to banding algorithms on handling the alignment. The approach uses only linear space with an obviously reducing time complexity. This heuristic has several advantages over the previous fixed-band methods for aligning two sequences along a similarity-sensitive, adaptable band. Test results are conducted to show that the more precise similarity deliberated, the process is preformed more efficiently.