Relationship between protein structures and disulfide‐bonding patterns

Abstract

We found that that disulfide-bonding patterns can be used to discriminate structure similarity. Our method, based on the hierarchical clustering scheme, is applicable to proteins with two or more disulfide bonds and is able to detect the structural similarities of proteins of low sequence identities (<25%). Our results show the surprisingly close relationship between disulfide-bonding patterns and proteins structures. Our findings should be useful in protein structure modeling.

Phylogenetic distribution, functional epitopes and evolution of the CSαβ superfamily

“…The importance of disulfide bridges in shaping a protein structure has well been documented [62]. However, their role in the functional evolution of proteins is less well understood.…”

Conformational characterization of disulfide bonds: A tool for protein classification

“…The disulfide threeͲdimensional structure is highly conserved in Nature and has been used for protein clustering (Cheek et al, 2006;Chuang et al, 2003;Harrison and Sternberg, 1996;Thangudu et al, 2007). Different schemes have been introduced to classify the disulfide conformers (Harrison and Sternberg, 1996;Hutchinson and Thornton, 1996;Ozhogina and Bominaar, 2009;Schmidt et al, 2006;Srinivasan et al, 1990) and in this work we adopted the scheme proposed by Schmidt et al (2006).…”

Prediction of disulfide connectivity in proteins with support vector machine

“…The overall Q p of the SVM model on the dataset SP39-PDB, which was also used in previous studies (Baldi et al, 2005;Vullo and Frasconi, 2004), can reach 59%. Moreover, comparing with the CSP method proposed previously (Chuang et al, 2003;van Vlijmen et al, 2004;Zhao et al, 2005), in which the distribution of cysteine residues in a protein sequence was considered as important information for predicting the disulfide connectivity, our SVM model still generates better results by learning from the profile information only.…”

“…Besides encoding information of amino acids as features, some other possible values such as the length of protein sequence can also be used. In this study, we further explored the usage of sequential distances between cysteines, which may contain knowledge for disulfide bonding patterns (Chuang et al, 2003;Harrison and Sternberg, 1994;van Vlijmen et al, 2004;Zhao et al, 2005). Hence, a data entry for our model can be expressed as follows: label : window i ; window j ; distanceði; jÞ ðo ptionalÞ;…”

“…Other than machine learning techniques, a simpler method based on cysteine separation profiles (CSPs) was originally proposed by Chuang et al (2003) and van Vlijmen et al (2004). This method was further developed in our previous work to predict the disulfide connectivity pattern of proteins (Zhao et al, 2005).…”