Multiple Sequence Alignment Based on Indicator-Based Multi-Objective Evolutionary Algorithm

In this paper, multiple sequence alignment (MSA) is modeled as a two-objective minimization problem, the score of the alignment serves as an objective, the penalty of the gaps serves as the other objective. Three scoring methods are compared to test the effectiveness of the proposed model. Indicator-based evolutionary algorithm (IEA) is used to resolve MSA. The experimental results of 12 state-of-the art aligners are compared with those of the model proposed. The results show that, MSA-IEA achieves uniformly distributed Pareto optimal front. Among the three scoring methods, the linear combination of Q-score, total column score and weighted sum of pair performs the best in the exploration and convergence of the algorithm. The results convey that, more effort should be focused on the design of the scoring method in the MSA based on multi-objective evolutionary algorithm, so that the subjective and the objective scoring criteria can precisely guide the population towards the Pareto optimal front.