Global pairwise sequence alignment through mixed-integer linear programming: a template-free approach

The problem of protein sequence alignment is the main starting point for biological analysis of genomic information. A new approach to global pairwise sequence alignment is proposed. This approach is formulated in a mathematically rigorous way, as a mixed integer linear optimization problem. Not only does the proposed formulation guarantee the identification of the global optimal alignment, but it also allows for a complete rank-ordered list of any number of user specified top ranking pairwise alignments and for the incorporation of restraints based on restrictions necessary to maintain biological function. This approach has been applied to the alignment of transmembrane helices and serine proteases to demonstrate its utility and advantages over other algorithms.