RNA Secondary Structure Prediction using Ant Colony Optimisation

It is important to know the secondary structure of RNA for applications such as drug development and modelling single stranded viruses. Predictive methods have various degrees of accuracy but are significantly faster and cheaper than empirical methods such as X-ray crystallography. This project explores how Ant Colony Optimisation (ACO) performs on the task of RNA secondary structure prediction (RNASSP). An ant colony system is developed and experiments are conducted to examine its behaviour on this problem and to determine a good set of parameters. The performance and accuracy of this approach is then compared with alternative methods. The main findings are that whilst the accuracy of ACO is as good as dynamic programming for small sequences it is significantly slower to execute. For longer sequences both slower and less accurate than dynamic programming.