Avbelj,F. and Moult,J. (a) Biochemistry (In press)
Avbelj,F. and Moult,J. (b) (submitted)
Moult,J. and Unger,R. Biochemistry 30 3816-3823 1991.
Determination of the Structure of Small Protein Fragments using Torsion Space Monte Carlo and Genetic Algorithm Methods. We have adopted an approach to determining structure from sequence that seeks to understand protein folding pathways (Moult and Unger, 1991), with the aim of then imitating that pathway in folding simulations. So far, we are able to fold up selected fragments of proteins to conformations close to that found in the final full structures.
A torsion space representation of conformation is used, together with an all atom model. The force field uses scaled empirical point atomic charge electrostatics together with a surface area term representing electrostatic and hydrophobic solvation contributions. Both the electrostatics and surface area terms have been parameterized by a potential of mean force analysis of protein structures (Avbelj, 1992; Avbelj and Moult (a),(b)).
Initial conformations are generated starting from a random coil, and changing mainchain and side chain angles at random, drawing replacement angles from a library of observed values, and accepting or rejecting moves using the Metropolis criterion. Samples from the Monte Carlo run (Avbelj and Moult (b)) are used to provide a starting population for a Genetic Algorithm. New generations are created by performing cross overs on the population, annealing the side chain conformations of the hybrids so formed, and then using the Metropolis criterion to decide whether to accept them.
In tests on small (12-16 residues) protein fragments selected for their independent folding properties, both Monte Carlo and Genetic Algorithm methods are able to generate conformations close to the experimental ones in most cases, although there are occasional convergence and potential problems.
As part of the protein structure prediction experiment, we attempted to use the Genetic Algorithm method to determine the conformation of three peptides: residues 7 to 22 of the subtilisin propiece (prosub), residues 1 to 15 of Eosinophil Derived Nuerotoxin (EDN), and the membrane binding domain of the C2 domain of human coagulation factor VIII (membind), a 22 residue peptide. Results of these experiments will be presented.
Avbelj,F. Biochemistry 29 2403-2408 1992
Avbelj,F. and Moult,J. (a) Biochemistry (In press)
Avbelj,F. and Moult,J. (b) (submitted)
Moult,J. and Unger,R. Biochemistry 30 3816-3823 1991.
Structural Biology home page Asilomar Conference home page