Design of selective cationic antibacterial peptides: A multiobjective genetic algorithm approach

Selective Cationic Antibacterial Peptides (SCAPs) are becoming a potential alternative to known antibiotics to deal with multi-drug resistant pathogens. Conventional techniques for the discovery and design of SCAPs can be time consuming and expensive, therefore the use of computers to aid such discovery or design is becoming attractive since they can help to reduce the number of sequences to be evaluated in the lab. A recent work shows that if a peptide fulfill a set of constraints on four physicochemical properties then the peptide is a candidate for being a SCAP. This result suggests that an exhaustive search can be used to discover new SCAPs. However, the size of the search space, 20n for peptides of length n, makes it prohibitively expensive to search for peptides longer than 11 amino acids on current desktop machines. As an alternative to the exhaustive search the present work proposes to model the SCAPs design problem as a multi-objective optimization problem. In order to solve the resulting problem an evolutionary algorithm based on the well known NSGA-II with a variable length representation and problem domain mutation operators is proposed. Preliminary results show that the approach can effectively and efficiently design SCAPs of different lengths. The predicted antibacterial peptides are validated in silico by using a publicly available Cationic Antimicrobial Peptides predictor.