期刊名称：International Journal of Computer Science and Information Technologies
电子版ISSN：0975-9646
出版年度：2011
卷号：2
期号：4
页码：1400-1402
出版社：TechScience Publications
摘要：Proteins are involved in many essential processes within cell. Uncovering the diverse function of proteins and their interactions within the cell may improve our understanding of protein functions. Several high-throughput techniques employed to decipher PPI are erroneous and are limited by the lack of coverage. Computational techniques are therefore sought to predict genome-wide PPI. In this paper, domain structure is used as a feature for computational prediction of PPI and support vector machines (SVM) as a learning system. We have used both, existing method and frequency count (FC) method for feature representation of protein domains and carried our experiment using SVM with different kernels. Both the methods achieved accuracy of about 78% for RBF kernel. But frequency count method reduced the storage requirement by half. These results indicate that PPI can be predicted from domain structure using frequency count method with reliable accuracy and reduced storage requirement.terban
关键词：Protein-protein interactions; Support Vector;Machines; Domain structure; Frequency Count Method.
