DephosSitePred: a high accuracy predictor for protein dephosphorylation sites.

In this study, we developed a novel PTP site prediction model, DephosSitePred, based on bi-profile sequence features. Weight parameters in the support vector machine were applied to improve the low sensitivity due to extremelyimbalanced datasets.DephosSitePred yielded Matthews correlation coefficientsof 0.686 for proteintyrosine phosphatase 1B (PTP1B), 0.668 for Src homology region 2 domain-containing phosphatase (SHP)-1,and 0.748 for SHP-2substrate sites, whichsignificantly outperformed other existing predictors. Moreover, 30 times of 5-fold cross-validations showed that DephosSitePred achieved average area under the curve values of 0.968, 0.968, and 0.982 for PTP1B, SHP-1 and SHP-2, respectively, which were 0.115, 0.105 and 0.105 higher than those of the second best model, MGPS-DEPHOS, respectively. The obvious improvements demonstrated that DephosSitePred might be a powerful and complementary tool for further experimental investigation of PTPs.
PMID: 28031011 [PubMed - as supplied by publisher]

Source: Combinatorial Chemistry and High Throughput Screening - December 27, 2016 Category: Chemistry Authors: Jia C, He W, Zou Q Tags: Comb Chem High Throughput Screen Source Type: research