Parallelizing and optimizing a hybrid differential evolution with Pareto tournaments for discovering motifs in DNA sequences

Transcriptional regulation is the main regulation of gene expression, the process by which all prokaryotic organisms and eukaryotic cells transform the information encoded by the nucleic acids (DNA) into the proteins required for their operation and development. A crucial component in genetic regulation is the bindings between transcription factors and DNA sequences that regulate the expression of genes. These specific locations are short and share a common sequence of nucleotides. The discovery of these small DNA strings, also known as motifs, is labor intensive and therefore the use of high-performance computing can be a good way to address it. In this work, we present a parallel multiobjective evolutionary algorithm, a novel hybrid technique based on differential evolution with Pareto tournaments (H-DEPT). To study whether this algorithm is suitable to be parallelized, H-DEPT has been used to solve instances of different sizes on several multicore systems (2, 4, 8, 16, and 32 cores). As we will see, the results show that H-DEPT achieves good speedups and efficiencies. We also compare the predictions made by H-DEPT with those predicted by other biological tools demonstrating that it is also capable of performing quality predictions.