Protein structures prediction (PSP) is a computationally complex problem. Simplified models of the protein molecule (such as the HP Model) and the use of evolutionary algorithms (EAs) are among the most investigated techniques for PSP. However, the evaluation of a structure represented by the HP model considers only the number of hydrophobic contacts, which doesn't enable the EA to distinguish between structures with the same number of contacts. This paper presents a new multi-objective formulation for PSP in HP Model. Two metrics are evaluated: the number of hydrophobic contacts and the distance between the hydrophobic amino acids. Both metrics are used by the Multi-objective EA in Tables. We showed that the algorithm is fast and robust.
Protein structure prediction; Evolutionary algorithms; Multi-objective optimization; HP Model
