Comparison of Parallel and Serial Genetic Algorithms for RNA Secondary Structure Prediction

Abstract

The basic function of a biomolecule is determined by its 3 dimensional shape, otherwise known as the tertiary structure. However, existing empirical methods to determine this shape are too costly and lengthy to be practical. RNA is of interest as a biomolecule because it is central in several stages of protein synthesis. Also, its secondary structure dominates its tertiary structure. In our model, RNA secondary structure develops as a consequence of bonds which form between specific pairs of nucleotides known as the canonical base pairs. Searching a sequence of nucleotides for all possible base pairs is rapid and straightforward; the challenge comes from attempting to predict which specific canonical base pairs will form bonds in the real structure. Various algorithms have been used for RNA structure prediction such as dynamic programming, and comparative methods [1] and stochastic methods such as genetic algorithms (GA).