No abstract

The microfluidic flow is typically laminar due to the dominant viscous effects. At Reynolds numbers far below 1 (Re<<1), the fluid inertia can be neglected. For the steady flow of...

O-GlcNAcylation is a ubiquitous post-translational modification of proteins that is involved in the majority of cellular processes and is associated with many diseases. To reduce the workload and increase the relevance of experimental identification of protein O-GlcNAcylation sites, O-GlcNAcPRED, a support vector machine (SVM)-based model, was developed to capture potential O-GlcNAcylation sites. By virtue of the novel adapted normal distribution bi-profile Bayes (ANBPB) feature extraction method, O-GlcNAcPRED yielded a sensitivity of 80.83%, a specificity of 78.17% and an accuracy of 79.50% in jackknife cross-validation experiments. In an independent test on 38 recently experimentally identified human O-GlcNAcylated proteins with 67 O-GlcNAcylation sites, O-GlcNAcPRED captured 26 proteins and 39 sites, clearly outperforming the existing predictors, YinOYang and O-GlcNAcscan.

Protein S-nitrosylation is a reversible post-translational modification by covalent modification on the thiol group of cysteine residues by nitric oxide. Growing evidence shows that protein S-nitrosylation plays an important role in normal cellular function as well as in various pathophysiologic conditions. Because of the inherent chemical instability of the S-NO bond and the low abundance of endogenous S-nitrosylated proteins, the unambiguous identification of S-nitrosylation sites by commonly used proteomic approaches remains challenging. Therefore, computational prediction of S-nitrosylation sites has been considered as a powerful auxiliary tool. In this work, we mainly adopted an adapted normal distribution bi-profile Bayes (ANBPB) feature extraction model to characterize the distinction of position-specific amino acids in 784 S-nitrosylated and 1568 non-S-nitrosylated peptide sequences. We developed a support vector machine prediction model, iSNO-ANBPB, by incorporating ANBPB with the Chou’s pseudo amino acid composition. In jackknife cross-validation experiments, iSNO-ANBPB yielded an accuracy of 65.39% and a Matthew’s correlation coefficient (MCC) of 0.3014. When tested on an independent dataset, iSNO-ANBPB achieved an accuracy of 63.41% and a MCC of 0.2984, which are much higher than the values achieved by the existing predictors SNOSite, iSNO-PseAAC, the Li et al. algorithm, and iSNO-AAPair. On another training dataset, iSNO-ANBPB also outperformed GPS-SNO and iSNO-PseAAC in the 10-fold crossvalidation test.

No abstract