A CLASSIC EVOLUTIONARY ALGORITHM TO PROTEIN STRUCTURE PREDICTION

FQ1

Rodrigo Antonio Faccioli, Telma Woerle de Lima, Ivan Nunes da Silva, Alexandre Cláudio Botazzo Delbem

Engineering School of Sao Carlos, Institute of Mathematics and Computer Sciences

IntroductionThe Protein Structure Prediction (PSP) can be an optimization problem and it aims to determine the protein tertiary structure from its amino acids sequence. This paper presents a classic Evolutionary Algorithm (EA) to PSP problem using an ab initio approach. ObjectiveDevelopment a classic evolutionary algorithm to PSP problem with hydrophobic interactions. MethodologyThe algorithm starts initializing a random conformation. The torsion angles (Φ,Ψ,Χ_i) are generated at random from the constrained regions. Afterwards, the energy of the conformation is evaluated. First, the protein’s structure in internal coordinates (backbone and side-chain torsion angles) is transformed into Cartesian coordinates. We proposed three kinds of crossover operators: one based on BLX-α operator; one using uniform crossover and the last is a two-point crossover. Three kinds of mutation operators were proposed: the first acts on the peptide chain; the second and the third apply a uniform mutation, modifying all the values of the backbone and side-chain torsion angles. Results

This classic EA used the population size is 200 chromosomes and the maximum number of generations is 100. The cost function has dielectric constant equal to 4.0. Table 1 presents the lowest energy of 1A11 protein tested and your Distance Matrix Error (DME).