Towards a Better Understanding of Heuristic Approaches Applied to the Biological Motif Discovery

Abstract

The detection of transcription factor binding sites (TFBS) play a important role inside bioinformatics challenges. Its correct identification in the promoter regions of co-expressed genes is a crucial step for understanding gene expression mechanisms and creating new drugs and vaccines. The problem of finding motifs consists of looking for conserved patterns in biological datasets of sequences through the use of unsupervised learning algorithms. For that reason, it is considered one of the classic problems of computational biology, which in its simplest formulation has been proven to be NP-HARD. Moreover, heuristic and meta-heuristic algorithms have been shown to be very promising in solving combinatorial problems with very large search spaces. In this work, we propose an evaluation of different heuristics and meta-heuristics approaches in order to measure its performance: Variable Neighborhood Search (VNS), Expectation Maximization (EM) and Iterated Local Search (ILS). For each of them, two sets of experiments were carried out: In the first, the heuristics were performed alone and in the second, a constructive procedure was introduced with respect to improve the quality of initial solutions. Finally, the metrics were compared with the state-of-art MEME algorithm, which is very used in biological motif discovery. The results obtained suggest that the heuristics are more efficient when used together and also, a constructive procedure was very promising, managing to improve the performance metrics of the evaluated heuristics in most experiments. Also, the combination between a constructive procedure and EM proved to be quite competitive, managing to outperform the MEME algorithm in several datasets.