WEIGHTED SOFT SET APPROACH FOR MINING FREQUENT AMINO ACID ASSOCIATIONS IN PEPTIDE SEQUENCES OF SWINE INFLUENZA VIRUS

Life Science -Bioinformatics

Abstract

The amino acid associations present in molecular sequences of viruses have correlation with the molecular mechanisms of the disease and other molecular process of an organism. The knowledge of these amino acid associations is crucial for understanding these molecular mechanisms and process. The major challenge in exploring amino acid association patterns in molecular sequences is the presence of uncertainty. In this paper weighted soft set approach is proposed to mine amino acid associations in molecular sequences of swine influenza virus. Soft set has been employed to incorporate the relationship of amino acid associations with the parameters. The parameter like length has been incorporated by assigning weights to the different length ranges. The dataset of 82611 sequences of swine influenza virus is taken from NCBI and filtered to obtain 36434 non redundant sequences. The proposed approach is employed to explore amino acid associations and results have been compared with ordinary soft set approach. It is observed that the weighted soft set approach is able to prune the over estimation of support by ordinary soft set approach. Further the weighted soft set approach is able to reduce the deviation in association patterns thereby improving the consistency of results. It is also observed that weighted soft approach is able to address the 50% to 85% of uncertainity left out by soft set approach which is leading to improvement in the results. Thus weighted soft set approach is superior to ordinary soft set approach for mining amino acid associations in swine influenza virus.