A hybrid algorithm for the DNA sequencing problem

• Published in: Discrete Applied Mathematics Vol. 163; pp. 87 - 99
• Main Authors: Caserta, Marco, Voß, Stefan
• Format: Journal Article
• Language: English
• Published: Elsevier B.V 30.01.2014
• Subjects: DNA sequencing; Health care: bioinformatics; Integer programming; Matheuristics; Metaheuristics; Integer programming; Metaheuristics; Health care: bioinformatics; Matheuristics; DNA sequencing
• ISSN: 0166-218X; 1872-6771
• DOI: 10.1016/j.dam.2012.08.025

The DNA sequencing problem is aimed at reconstructing an unknown fragment of DNA (deoxyribonucleic acid) based upon a set of oligonucleotides that makes up the fragment’s spectrum. Such a spectrum is the result of a hybridization experiment, in which oligonucleotides hybridize with the unknown DNA sequence. The introduction of errors (positive as well as negative) in the biological experiment phase gives rise to a challenging combinatorial optimization problem. The DNA sequencing problem can be modeled as a variation of the classical traveling salesman problem and, due to its computational complexity, it is a candidate for the design and implementation of metaheuristic algorithms. We present a hybrid algorithm, which may be seen as an approach within the recently introduced area of matheuristics, i.e., an approach in which mathematical programming techniques and metaheuristic schemes are effectively intertwined. The algorithm is tested on 400 benchmark instances from the literature and compares favorably with the best known algorithm. In addition, an explanation concerning the relation between error distribution and algorithmic performance is provided, illustrating that the way in which negative errors are distributed within the spectrum has a bearing on the overall algorithmic performance.