Genetic Algorithms and Their Application to the Protein Folding Problem
Abstract
The protein folding problem involves the prediction of the secondary and tertiary structure of a molecule given the primary structure. The primary structure defines sequence of amino-acid residues, while the secondary structure describes the local 3-dimensional arrangement of amino-acid residues within the molecule. The relative orientation of the secondary structural motifs, namely the tertiary structure, defines the shape of the entire biomolecule. The exact, mechanism by which a sequence of amino acids (protein) folds into its 3- dimensional conformation is unknown Current approaches to the protein folding problem include calculus-based methods, systematic search, model building and symbolic methods, random methods such as Monte Carlo simulation and simulated annealing, distance geometry, and molecular dynamics. Many of these current approaches search for conformations which minimize the internal energy of the molecule. A genetic algorithm (GA), a stochastic search technique modeled after natural adaptive systems, potentially offers significant speedup over other search algorithms because of its inherent parallelizability. The results of applying a parallel GA to the protein folding problem show significant improvement in execution time when compared to serial implementations of the GA. In addition, the parallel GA demonstrates good scalability characteristics since the communications strategy used to manage the population can be tailored to the parallel architecture.
Document Details
- Document Type
- Technical Report
- Publication Date
- Dec 01, 1993
- Accession Number
- ADA274389
Entities
People
- Donald J. Brinkman
Organizations
- Air Force Institute of Technology