Gly-PseAAC: Identifying protein lysine glycation through sequences

Similar to the regular enzymatic glycosylation, glycation also attaches a sugar molecule to a peptide, but does not need the help of an enzyme. Glycation may occur both inside and outside the host body, and will compete with the glycosylation procedure for functional regulation of mature protein products. The glycated residues do not show significant patterns, which make both in silico sequence-level predictors and wet-lab validations a major challenge. This study hypothesizes that a better feature set formulated from the glycated flanking peptides may lead to a good glycation prediction program.

We explored the application of sequence order information and position specific amino acid propensity (PSAAP) in the glycation residue prediction problem. The PSAAP demonstrated its ability to discriminate the glycated residues from the background control peptides. A Support Vector Machine (SVM) model was constructed from the training dataset and achieved 68.91% in the overall accuracy. The model also achieves 0.7258 and 0.3198 in the Area under the ROC and Matthew's Correlation Coefficient, respectively. The user-friendly online version of the proposed algorithm may be found on the web server Gly-PseAAC at http://app.aporc.org/Gly-PseAAC/.

The feature set PSAAP was calculated and led to a useful classification of glycation residues.