Figure 3.

Flow chart for applying SVMs to predict PPIs from sequences. Four issues must be addressed. First, protein sequences must be represented by vectors. Second, the vector representation of protein pairs must be symmetric. Third, a set of non-interacting protein pairs (negative PPIs) should be provided because SVMs are supervised learning algorithms. Fourth, a proper kernel will facilitate the nonlinear prediction. The focus of this paper is on the first and second issue. Community standard procedures are adopted to address the third and fourth issues in this paper.