Application Of Support Vector Machine On Bioinformatics

With the arrival of the post-genome era, researchers begin to develop various tools on analyzing biological data in order to turn it into knowledge.The method of support vector machine are applied in bioinformatics to discover useful biological information .G-protein Coupled Receptors (GPCRs) belong to one of the largest superfamilies of membrane proteins and are important targets for drug design. The features of single amino acid frequency, dipeptide composition of proteins and the codon usage are extracted from protein and mRNA sequences, then GPCRs are recogonized and classified into five super families using support vector machine with high accuracy. In this paper, the information of nucleotide sequence is firstly used to recognize the family of GPCRs, which leads to a high prediction accuracy.Recombination hotspots and coldspots have come to widely attention in research of mechanism of meiotic recombination. A novel method is developed for predicting hotspots and coldspots by extracting sequence features using dinucleotide abundance and codon usage combined with support vector machine. The result indicates that coldspots and hotspots can be classified with high accuracy in this method.Horizontal Gene Transfer (HGT) can be regarded as one of the most important factor in the evolution. We use codon usage feature to classifying horizontal genes and other genes combined with support vector machine. Compared with other methods to discover horizontal genes, it has been testified that codon usage feature can be applied as a useful standard to detect HGT on bacterial genome sequences. We develop a novel method to predict the activity of siRNA by extracting the dinucleotide features of siRNA sequences combined with support vector machine. This algorithm achieves a better performance than several previous published methods.