Large Scale Structure Sampling for Protein Fold Prediction using the Generalized Simulated Annealing

Abstract

Proteins are the building blocks of cells and the executioners of nearly all cellular functions. Their structure is of paramount importance to understand their dynamics and function, as well as the interactions with other molecules.In this work, we apply the Generalized Simulated Annealing (GSA) to guide the exploration of the energy hyper surface of the protein folding process, looking for the global minimum and, hence, the native fold of the protein. The GSA is a stochastic search algorithm employed in energy minimization and used in global optimization problems, such as gravity models, fitting of numerical data and conformation optimization of small molecules. Our software applies the analytical inverse of the probability distribution from GSA, a new method to apply rotations to the phi and psi angles of the peptide bonds and side chains, faster connection with NAMD for potential energy calculation and the possibility of parallel execution, granting a new take on ab-initio protein structure prediction. The new design also allows for an easier inclusion of knowledge derived potentials, based on experimentally determined protein structures.We present results for the 14 amino acid protein mastoparan-X. The chain folds with RMSD of 3,0 angstroms after 500.000 GSA steps. Currently, for this system, the software calculates 5 million GSA steps in under 6 hours using 4 processors in one node.Predicted structures can be refined with molecular dynamics simulations and used to study proteins whose conformation can not be determined with experimental methods. These structures can be used in protein engineering, drug development and biotechnological research.