An adaptive fuzzy thresholding algorithm for exon prediction

Abstract

Thresholding is always critical and decisive in problem solving. In this paper, we propose an adaptive fuzzy logic-based approach to thresholding for exon prediction problem, which is an important problem in bioinformatics. Rather than using the same threshold for the entire dataset, we allow the thresholds to vary along the dataset based on the local statistical properties. We incorporate it in a soft computing framework of training and testing to determine the optimum adaptive thresholds. The search space of the trained database is reduced by determining a dynamic range of thresholds using fuzzy logic rules, which makes our approach faster. To test our approach, we applied the proposed algorithm on a particular solution to the exon prediction problem, which uses a threshold on the frequency component at f = 1/3 in the nucleotide sequences. Preliminary experiments on the nucleotide data of Saccharomyces Cerevisiae (Bakers yeast) illustrate the potential of our approach. The adaptive thresholding approach gave suitable thresholds to detect the exons which were otherwise difficult to detect using a traditional static thresholding scheme.