Variable neighborhood search for solving the DNA fragment assembly problem

Abstract

The fragment assembly problem consists in the building of the DNA sequence from several hundreds (or even, thousands) of fragments obtained by biologists in the laboratory. This is an important task in any genome project, since the accuracy of the rest of the phases depends of the result of this stage. In addition, real instances are very large and therefore, the efficiency is also a very important issue in the design of fragment assemblers. In this paper, we propose two Variable Neighborhood Search variants for solving the DNA fragment assembly problem. These algorithms are specifically adapted for the problem being the difference between them the optimization orientation (fitness function). One of them maximizes the Parsons’s fitness function (which only considers the overlapping among the fragments) and the other estimates the variation in the number of contigs during a local search movement, in order to minimize the number of contigs. The results show that doesn’t exist a direct relation between these functions (even in several cases opposite values are generated) although for the tested instances, both variants allow to find similar and very good results but the second option reduces significatively the consumed-time.