VPTMdb: a viral post-translational modification database

Yujia Xiang; Quan Zou; Lilin Zhao

Moreover, VPTMpre, a novel feature representative classifier, was developed to predict viral protein serine sites. We compared various feature extraction methods and selected the optimized features using the mRMR algorithm. The feature analysis results showed that 68D was able to distinguish the phosphorylated sites and non-phosphorylated sites in viral proteins. VPTMpre was integrated into the VPTMdb web server to provide an online phosphorylation site prediction service. Users can choose three classifiers (svm, random forest and naïve Bayes) to predict phosphorylated sites of interests. However, because of data limitations, the prediction of VPTMpre is limited to serine sites. With continuous collection of new viral PTM data, we expect that VPTMpre will be extended to predict more types of PTM sites and obtain a better performance.