Multiple molecular sequence alignment by island parallel genetic algorithm

This paper presents an evolution-based approach for solving multiple molecular sequence alignment. The approach is based on the island parallel genetic algorithm that relies on the fitness distribution over the population of alignments. The algorithm searches for an alignment among the independent isolated evolving populations by optimizing weighted sum of pairs objective function which measures the alignment quality. The parallel approach is implemented on PARAM 10000, a parallel machine developed at the Center for Development of Advanced Computing, Pune, and is shown to consistently perform better than the sequential genetic algorithm. The algorithm yields alignments that are qualitatively better than an alternative method, ClustalW.