Abordagem baseada em algoritmos meméticos para descoberta de motivos biológicos

Abstract

The location of Transcription Factor Binding Sites (TFBS) is considered one of the main problems of Bioinformatics. Their correct identification plays an important role in understanding the mechanisms of genetic regulation and development of new drugs. The de novo motif discovery is a difficult task and the construction of computationally effective programs is necessary to improve the understanding and study of cell transcripts. This allowed to point and choose recurring elements in a set of sequences for further biological investigation, such as the results of high performance differential expression experiments. In this work we present the Memetic Framework for Motif Discovery (MFMD), an algorithm whose construction was inspired by the theory of memes and based on two heuristics - a semi-greedy construct based on GRASP and another based on VNS - as well as a global optimizer based on the evolutionary algorithms. When evaluated in synthetic and real datasets, MFMD has outperformed the main existing motif detection tools. This new approach was compared to other techniques well known in the literature and the results suggested a significant improvement in the performance measures achieved by MFMD in relation to the algorithms faced.