Skip to main content
. 2017 Apr 27;5:e3261. doi: 10.7717/peerj.3261

Figure 1. Illustration of predicting PTM sites.

Figure 1

(A) Constructing bipartite network between target site and modification. (B) Calculating two kernels, namely the local sequence kernel and the Gaussian interaction profile kernel, and combining these two kernels to train SVM. (C) Ranking all the potential relationships between target site and modification. MK-SVM: multiple kernel SVM.