SuccinSite: a computational tool for the prediction of protein succinylation sites by exploiting the amino acid patterns and properties

Abstract

Lysine succinylation is an emerging protein post-translational modification, which plays an important role in regulating the cellular processes in both eukaryotic and prokaryotic cells. However, the succinylation modification site is particularly difficult to detect because the experimental technologies used are often time-consuming and costly. Thus, an accurate computational method for predicting succinylation sites may help researchers towards designing their experiments and to understand the molecular mechanism of succinylation. In this study, a novel computational tool termed SuccinSite has been developed to predict protein succinylation sites by incorporating three sequence encodings, i.e., k-spaced amino acid pairs, binary and amino acid index properties. Then, the random forest classifier was trained with these encodings to build the predictor. The SuccinSite predictor achieves an AUC score of 0.802 in the 5-fold cross-validation set and performs significantly better than existing predictors on a comprehensive independent test set. Furthermore, informative features and predominant rules (i.e. feature combinations) were extracted from the trained random forest model for an improved interpretation of the predictor. Finally, we also compiled a database covering 4411 experimentally verified succinylation proteins with 12 456 lysine succinylation sites. Taken together, these results suggest that SuccinSite would be a helpful computational resource for succinylation sites prediction. The web-server, datasets, source code and database are freely available at http://systbio.cau.edu.cn/SuccinSite/.

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

“…SuccinSite [7] and SuccFind [4]). In this experiment, both Physicochemical and biochemical properties of amino acids were extracted from AAindex database, version 9.1 [12].…”

“…Binary amino acid encoding which featured in [7], calculates the positional information from the corresponding amino acids sequence fragments. In this study, all the 20 standard amino acid residues were considered and non-existing "O" residue was used for gap or padding purpose.…”

“…These metrices Identification from Sequence Data were considered in this study to consistently compare the performance of different PTM identification tools. Formulation of these metrices were also adapted from those state-of-the-art techniques specially from SucPred [3], Succfind [4], iSuc-PseOpt [6], SuccinSite [7] and SucStruct [16].…”

“…In this work, k max = 5 and it produced 21*(k max +1)*21 = 2646 different amino acid pairs which were used in feature vector for each segment sequence. The feature vector was calculated using the following equation which was also used in the design of SuccinSite tool [7]: …”

“…iSuc-PseOpt [6] integrated the sequence-coupling effects into the general pseudo amino acid composition, solved class imbalance problem and applied random forest algorithm for prediction. SuccinSite [7] used a random forest classifier, incorporating three sequence encoding features such as the composition of k-spaced amino acid pairs, binary encoding and amino acid index. pSuc-Lys [8] has incorporated the sequence-coupled information into the general pseudo amino acid composition and used ensemble random forest as a classifier.…”