Decomposition-based multi-objective optimization approach for PPI network alignment

Abstract

The alignment of protein–protein interaction (PPI) networks is in the spotlight of the research in bioinformatics. The main goal is to find structural or functional complexes that are evolutionarily conserved between species. Recent works in the area struggle to produce alignments of both good topological and functional quality, since those two objectives conflict more than expected. To this end, we introduce a decomposition-based multi-objective algorithm with two new problem-aware mutation operators, being each of them focused on the improvement of one of both objectives. The experiments have been performed over 10 scenarios of PPI network alignment with real data from five different species. The results have been evaluated with five quality metrics. The performed experiments have confirmed that the solutions resulting from our proposal obtain statistically significant improvements and are of higher quality than the solutions from related works.