A Memetic Algorithm for Protein Structure Prediction Based on Conformational Preferences of Aminoacid Residues

Memetic Algorithms are intrinsically concerned with exploiting all available knowledge about the problem under study. The incorporation of problem domain knowledge is not an optional mechanism, but a fundamental feature of the MAs. In this paper we present a Memetic Algorithm to tackle the three dimension protein structure prediction problem. The algorithm uses a structured population and incorporates a simulated annealing algorithm as a local search strategy. The algorithms takes advantage of the knowledge stored in the PDB. The algorithm was tested on six proteins and results show its ability to predict good tertiary structures.