Improving Disulfide Connectivity Prediction with Sequential Distance between Oxidized Cysteines
Resource
Bioinformatics 21 (24): 4416-4419
Journal
Bioinformatics
Pages
4416-4419
Date Issued
2005
Date
2005
Author(s)
Tsai, Chi-Hung
Chen, Bo-Juen
Chan, Chen-hsiung
Liu, Hsuan-Liang
Kao, Cheng-Yan
Abstract
SUMMARY: Predicting disulfide connectivity precisely helps towards the solution of protein structure prediction. In this study, a descriptor derived from the sequential distance between oxidized cysteines (denoted as DOC) is proposed. An approach using support vector machine (SVM) method based on weighted graph matching was further developed to predict the disulfide connectivity pattern in proteins. When DOC was applied, prediction accuracy of 63% for our SVM models could be achieved, which is significantly higher than those obtained from previous approaches. The results show that using the non-local descriptor DOC coupled with local sequence profiles significantly improves the prediction accuracy. These improvements demonstrate that DOC, with a proper scaling scheme, is an effective feature for the prediction of disulfide connectivity. The method developed in this work is available at the web server PreCys (prediction of cys-cys linkages of proteins).
Type
journal article
File(s)![Thumbnail Image]()
Loading...
Name
17.pdf
Size
181.03 KB
Format
Adobe PDF
Checksum
(MD5):0f68b8242d4f4f1adf7b974c1556d8c1