Cysteine separations profiles on protein sequences infer disulfide connectivity

Abstract

cykao@csie.ntu.edu.tw.

The Disulfide Connectivity Prediction with Support Vector Machine and Behavior Knowledge Space

“…Eventually, the prediction accuracy of our hybrid method with SVM, BKS and CSP reaches 69.1%, a great improvement compared with the previous results. Zhao et al (East Zhao et al, 2005). b Proposed by Wang et al (Chong-Jie Wang et al, 2012).…”

“…Instead of large amount of features used by the SVM method, Zhao et al (East Zhao et al, 2005) adopted only one feature, CSP (cysteine separations profile), to achieve nearly 50% accuracy in the insufficient dataset. The CSP of protein x with 2n oxidized cysteines (n disulfide bonds) is defined as…”

“…The divergence (D) of two CSPs for two proteins x and y is defined (East Zhao et al, 2005) as follows:…”

“…The features and cysteinepair representation we adopt are inspired by Wang et al (Chong-Jie Wang et al, 2012) and Lu et al (ChihHao Lu et al, 2007). In addition, we also combine the CSP method (East Zhao et al, 2005) to our hybrid method.…”

“…Ferre and P. Clote, 2005) were devoted to the DCP problem, but most of their accuracies are less than 50%. In 2005, Zhao et al (East Zhao et al, 2005) utilized the global information in a pro-tein, called cysteine separation profile (CSP), which is the separations among all oxidized cysteines on a protein sequence.…”