Applying fuzzy support vector machine to predict membrane protein interactions with different representations

Membrane protein interactions are involved in the regulation and execution of all biochemical pathways within the cell. It is crucial to identify the binding partners for understanding the function of an uncharacterized protein, and when these partners turn out to have known function, deductions about the potential role of the uncharacterized protein can often be made. However, the analysis of interactions between membrane proteins is difficult because of the hydrophobic nature of these proteins, and conventional biochemical and genetic assays are often of limited use. In this paper, a novel method is presented for the prediction of membrane protein interactions representing the interactions with different features. We adopt fuzzy support vector machine as our classifier. Jackknife tests on the working datasets indicate that the precision, sensitivity and accuracy are improved greatly. The results show that the proposed method is efficient to genetically predict membrane protein interactions.