Improved performance of Lysine Glutarylation PTM using Peptide Evolutionary Features

Glutarylation is a new acylated post-translational modification. It is responsible for diverse metabolic functions and biological process. Identification of glutarylation is very significant as it controls different mitochondrial activities. Laboratory methods for identification of glutarylation are costly and take a lot of time. If a computational tool can identify lysine glutarylation sites efficiently, it will both save time and money. In our proposed method, a comparatively new feature extraction technique, peptide evolutionary features, has been used that gave sensitivity of 75.14%, specificity of 99.17%, accuracy of 87.16% and MCC of 0.76 for 5-fold cross-validation and sensitivity of 78.5%, specificity of 99.0%, accuracy of 88.5% and MCC of 0.78 for independent test set. Sequence based feature extraction techniques preserve less information for separating positives and negatives compared to the position specific evolutionary feature extraction technique, proposed in our model. For this reason, the performance of our model outperforms the previously developed tools, MDDGlutar and GlutPred, to a great extent in all the cases.