GAMMA: A NEW DOCKING PROGRAM UTILIZING AN ADVANCED EVOLUTION SYSTEM ALGORITM AS AN ENGINE. Steven Ness, Trevor Hart, Randy Read, Department of Medical Microbiology and Immunology, University of Alberta, Canada

Gamma (Genetic Algorithm for MacroMolecular-ligand Analysis) is a new program that uses an advanced Genetic Algorithm/Evolution System (GA/ES) engine for docking flexible ligands to macromolecular targets. Genetic Algorithms are powerful minimization/adaptation engines that have been used previously in many problem domains, and are beginning to be applied to the docking problem.

Previous approaches to the docking problem using GAs have mainly used the simple canonical GA. The canonical GA is defined as an algorithm that iteratively uses crossover and mutation operators on a population of bit-string organisms, selecting and reproducing organisms that have higher scores in terms of an objective fitness function. Good results have been obtained using this simplified evolutionary algorithm applied to the docking problem.

Theoretical research into GA/ES systems is progressing at a rapid pace, and new methods that eclipse the performance of the canonical GA are now appearing. Gamma utilizes some of these important new methods including: the use of integers and real numbers as genes, multiple gene types, prevention of lethal crossovers, hybrid minimization, multiple simultaneous runs with migration, and Demes (local neighourhoods). These methods are combined together within an object-oriented framework that allows rapid prototyping and rapid design changes with little global program reorganization.

Preliminary results are very encouraging. Compared with a Monte Carlo Simulated Annealing engine, Gamma finds better (lower energy) solutions in considerably less time.