Carboxylator: incorporating solvent-accessible surface area for identifying protein carboxylation sites

Abstract

In proteins, glutamate (Glu) residues are transformed into γ-carboxyglutamate (Gla) residues in a process called carboxylation. The process of protein carboxylation catalyzed by γ-glutamyl carboxylase is deemed to be important due to its involvement in biological processes such as blood clotting cascade and bone growth. There is an increasing interest within the scientific community to identify protein carboxylation sites. However, experimental identification of carboxylation sites via mass spectrometry-based methods is observed to be expensive, time-consuming, and labor-intensive. Thus, we were motivated to design a computational method for identifying protein carboxylation sites. This work aims to investigate the protein carboxylation by considering the composition of amino acids that surround modification sites. With the implication of a modified residue prefers to be accessible on the surface of a protein, the solvent-accessible surface area (ASA) around carboxylation sites is also investigated. Radial basis function network is then employed to build a predictive model using various features for identifying carboxylation sites. Based on a five-fold cross-validation evaluation, a predictive model trained using the combined features of amino acid sequence (AA20D), amino acid composition, and ASA, yields the highest accuracy at 0.874. Furthermore, an independent test done involving data not included in the cross-validation process indicates that in silico identification is a feasible means of preliminary analysis. Additionally, the predictive method presented in this work is implemented as Carboxylator ( http://csb.cse.yzu.edu.tw/Carboxylator/ ), a web-based tool for identifying carboxylated proteins with modification sites in order to help users in investigating γ-glutamyl carboxylation.

MDD-Palm: Identification of protein S-palmitoylation sites with substrate motifs based on maximal dependence decomposition

“…The five validation results were then combined to generate a single estimation. Cross-validation evaluation improves the reliability of evaluation, because it considers all original data, in both the training and testing data sets, in general, and tests each subset only once [57]. To gauge the effective predictive performance of training model, the following measures were used: sensitivity (Sn), specificity (Sp), accuracy (Acc) and Matthews Correlation Coefficient (MCC): where TP, TN, FP and FN represent the numbers of true positives, true negatives, false positives and false negatives, respectively.…”

An Intelligent System for Identifying Acetylated Lysine on Histones and Nonhistone Proteins

“…The validation results were then combined to produce a single estimation. The advantage of performing a cross-validation evaluation is that all original data are regarded as both training set and testing set, and each dataset is used for testing exactly once [36]. …”

ViralPhos: incorporating a recursively statistical method to predict phosphorylation sites on virus proteins

“…The k results are then combined to produce a single estimation. The advantage of k-fold cross-validation is that all original data are regarded as both training set and test set, and each data is used for testing exactly once [31]. For this study, k was set to five.…”