A genetic algorithm for simplifying the amino acid alphabet

Abstract

Simplified amino acid alphabets have been successful in several areas of bioinformatics, including predicting protein structure, predicting protein function, and protein classification. Since the number of possible simplifications is large, it is not practical to search through all possible simplifications to find one suitable for a specific application. A previous study conducted by the authors indicate that algorithms with heavy reliance on randomness tend to produce poor simplifications. Genetic algorithms have been generally successful in producing quality solutions to problems with a large solution space, though their reliance on randomness makes it difficult to create quality simplifications. This study's goal is to overcome these difficulties, and create a genetic simplification algorithm. The presented results include the genetic simplification algorithm, as well as the difficulties of creating such an algorithm. The described algorithm has led to the development of a computer program that uses a genetic algorithm to produce simplified alphabets, and these outputs are listed and analyzed.