DV-iSucLys: Decision Voting to Improve Protein Lysine Succinylation Site Identification from Sequence Data

Abstract

Protein Post Translation Modification identification is one of the important steps in conducting disease-associated mutation studies. Though multiple chemical alterations happen in a protein after translation, the addition of succinyl group to lysine residue plays a vital role in regulating cellular metabolism and thus disease. Use of a classification algorithm on some features, driven either from protein structural, physicochemical or even biochemical information becomes a common approach that can yield a satisfactory result up to a certain level. Although, researchers already developed many computational methods to identify whether a lysine residue modified with succinyl group after translation, most of them focused on the improvement either on a single decision using a single method or feature enrichment or even development of a benchmark dataset. Therefore, there still exists scope for further improvement to characterise lysine residues of a protein sequence by considering multiple predictors at a time. In this study, an ensemble based approach called DV-iSucLys has been designed to characterise the lysine residue by adapting three well known and conceptually different classifiers and ensembling their decisions. Also, a benchmark succinylation dataset was extracted from existing benchmark datasets and recently updated succinylation data from UniProt consortium to investigate the performance of the proposed approach as well as contribute to further research. Analysing rigorous cross-validation results show that DV-iSucLys can characterise succinyl lysine residue better than the existing predictors.