Authors:
Fernando José Mateus da Silva
1
;
Juan Manuel Sánchez Pérez
2
;
Juan Antonio Gómez Pulido
2
and
Miguel A. Vega Rodríguez
2
Affiliations:
1
Polytechnic Institute of Leiria, Portugal
;
2
Universidad de Extremadura, Spain
Keyword(s):
Multiple sequence alignments, Genetic algorithms, Multiobjective optimization, Niched Pareto, Equivalence class sharing, Bioinformatics.
Related
Ontology
Subjects/Areas/Topics:
Artificial Intelligence
;
Computational Intelligence
;
Evolutionary Computing
;
Soft Computing
Abstract:
The alignment of molecular sequences is a recurring task in bioinformatics, but it is not a trivial problem. The size and complexity of the search space involved difficult the task of finding the optimal alignment of a set of sequences. Due to its adaptive capacity in large and complex spaces, Genetic Algorithms emerge as good candidates for this problem. Although they are often used in single objective domains, its use in multidimensional problems allows finding a set of solutions which provide the best possible optimization of the objectives – the Pareto front. Niching methods, such as sharing, distribute these solutions in space, maximizing their diversity along the front. We present a niched Pareto Genetic Algorithm for sequence alignment which we have tested with six BAliBASE alignments, taking conclusions regarding population evolution and quality of the final results. Whereas methods for finding the best alignment are mathematical, not biological, having a set of solutions whi
ch facilitate experts’ choice, is a possibility to consider.
(More)