A hybrid ant colony optimization algorithm for the far from most string problem

Abstract

The far from most string problem belongs to the family of string selection and comparison problems known as sequence consensus problems, where a finite set of sequences is given and one is interested in finding their consensus, that is, a new sequence that represents as much as possible all the given sequences. Among the consensus problems, the far from most string problem is computationally one of the hardest ones with applications in several fields, including molecular biology where one is interested in creating diagnostic probes for bacterial infections or in discovering potential drug targets. This paper comes with several contributions. On one side, the first linear integer programming formulation for the considered problem is introduced. On the other side, a hybrid ant colony optimization approach for finding good approximate solution to the problem is proposed. Both approaches are compared to the current state of the art, which is a recently proposed hybrid GRASP with path-relinking. Computational results on a large set of randomly generated test instances indicate that the hybrid ACO is very competitive.