Disulfide connectivity prediction based on structural information without a prior knowledge of the bonding state of cysteines

Authors:
Hsuan-Hung Lin;Jiin-Chyr Hsu;Yuan-Nian Hsu;Ren-Hao Pan;Yung-Fu Chen;Lin-Yu Tseng
Affiliations:
-;-;-;-;-;-
Venue:
Computers in Biology and Medicine
Year:
2013

Citing 10
Cited 0

Ensemble Methods in Machine Learning

MCS '00 Proceedings of the First International Workshop on Multiple Classifier Systems
Implementation of algorithms for maximum matching on nonbipartite graphs.

Implementation of algorithms for maximum matching on nonbipartite graphs.
Disulfide connectivity prediction using recursive neural networks and evolutionary information

Bioinformatics
Cysteine separations profiles on protein sequences infer disulfide connectivity

Bioinformatics
Disulfide connectivity prediction using secondary structure information and diresidue frequencies

Bioinformatics
Improving disulfide connectivity prediction with sequential distance between oxidized cysteines

Bioinformatics
Predicting disulfide connectivity from protein sequence using multiple sequence feature vectors and secondary structure

Bioinformatics
Ensemble-based classifiers

Artificial Intelligence Review
LIBSVM: A library for support vector machines

ACM Transactions on Intelligent Systems and Technology (TIST)
Prediction of disulfide bonding pattern based on a support vector machine and multiple trajectory search

Information Sciences: an International Journal

Quantified Score

Hi-index	0.00

Visualization

Abstract

Previous studies predicted the disulfide bonding patterns of cysteines using a prior knowledge of their bonding states. In this study, we propose a method that is based on the ensemble support vector machine (SVM), with the structural features of cysteines extracted without any prior knowledge of their bonding states. This method is useful for improving the predictive performance of disulfide bonding patterns. For comparison, the proposed method was tested with the same dataset SPX that was adopted in previous studies. The experimental results demonstrate that bridge classification and disulfide connectivity predictions achieve 96.5% and 89.2% accuracy, respectively, using the ensemble SVM model, which outperforms the traditional method (51.5% and 51.0%, respectively) and the model that is based on a single-kernel SVM classifier (94.6% and 84.4%, respectively). For protein chain and residue classifications, the sensitivity, specificity, and accuracy of ensemble and single-kernel SVM approaches are better than those of the traditional methods. The predictive performances of the ensemble SVM and single-kernel models are identical, indicating that the ensemble model can converge to the single-kernel model for some applications.