Global Pairwise Sequence Alignments with Multiple Scoring Matrices

A necessity of repeating alignments on a same pair of sequences to select an appropriate scoring function to detect a significance of score motivates us to work on higher performance matching technique. While overlapping each alignments obtained through a set of typical scoring matrix with its default gap cost, we observe significant parameters that may induce a maximum deviation from a reference global alignment. We propose a percent identity and a wideness of the alignment as a potential candidate of parameters. We sample pairs of protein sequences from ASTRAL database to generate the corresponding probability distribution with respect to the proposed parameters. When biologists conduct trial-and-error global matching procedure which is based on highly cost dynamic programming, our ‘static’ searching selection scheme can be 12 times faster than a normal alignment algorithm when repeating alignments with 3 scoring matrices, having a reasonably high accuracy of alignment between 85% and 95%. keywords: global sequence alignment, multiple scoring matrices