A web server for identifying S-glutathionylation sites. GSHSite is based on a statistical method for identifying S-glutathionylation sites and potential consensus motifs by maximal dependence decomposition (MDD). With the application of MDD, a large group of aligned sequences can be moderated into subgroups that capture the most significant dependencies between positions. By further evaluation using five-fold cross-validation, the support vector machine (SVM) models trained with MDD-clustered subgroups could improve predictive accuracy when compared to the model without MDD clustering. Moreover, the experimental S-glutathionylation data from published database (independent set) are used to test the effectiveness of the models in cross-validation.
Identifies S-glutathionylated sites from primary protein sequences. PGluS can mine the protein information by using multiple features. In order to reduce the computational time, 5-fold cross validation test instead of jack-knife test were used. The tool was evaluated on an independent testing dataset resulting in an accuracy of 71.25%. It can be useful to assist the discovery of S-glutathionylated sites.