Prediction of DNA-binding protein based on statistical and geometric features and support vector machines

Abstract

BackgroundPrevious studies on protein-DNA interaction mostly focused on the bound structure of DNA-binding proteins but few paid enough attention to the unbound structures. As more new proteins are discovered, it is useful and imperative to develop algorithms for the functional prediction of unbound proteins. In our work, we apply an alpha shape model to represent the surface structure of the protein-DNA complex and extract useful statistical and geometric features, and use structural alignment and support vector machines for the prediction of unbound DNA-binding proteins.ResultsThe performance of our method is evaluated by discriminating a set of 104 DNA-binding proteins from 401 non-DNA-binding proteins. In the same test, the proposed method outperforms the other method using conditional probability. The results achieved by our proposed method for; precision, 83.33%; accuracy, 86.53%; and MCC, 0.5368 demonstrate its good performance.ConclusionsIn this study we develop an effective method for the prediction of protein-DNA interactions based on statistical and geometric features and support vector machines. Our results show that interface surface features play an important role in protein-DNA interaction. Our technique is able to predict unbound DNA-binding protein and discriminatory DNA-binding proteins from proteins that bind with other molecules.

Identification of DNA-Binding and Protein-Binding Proteins Using Enhanced Graph Wavelet Features

“…The enhanced graph wavelet features (EGWF) outperform the other three features, "Z" (Zhou and Yan features [46], [47]), "Z"þ"FA" (improved features by the way of additional new features), and "Z"þ"GW" (improved feature using graph wavelet alone), which indicates that the combination of two enhanced strategies most effectively improve Zhou et al's "Z" feature. Particularly, the comparison results between the weight matrix (WM) method that is used in Fang et al's [39] paper for predicting protein-protein interactions (PPI) via the topology of a PPI network, DNAbinder [35], and Zhou et al's method show that the graph wavelet is a useful tool to enhance the discrimination of the features.…”

“…Eighty-six RNA-binding proteins are obtained from the datasets by Zhou and Yan [30] (http://www.hy8.com/main.php) 186 protein binding proteins are obtained from Murakami and Mizuguchi [5], and 106 ligand-binding proteins are randomly chosen from PDB to serve as a control set. Testing data: 104 DNA-binding proteins and 401 nonDNA-binding proteins obtained in Zhou and Yan [46]. Template library: The 199 protein-DNA complexes are used to form the template library for structure alignment.…”

“…The graph wavelet measurement GW j;v k is applied to enhance the extracted feature. Thus, the correlation between adjacent atoms is considered, which compensates the limitation of the Zhou et al's methods [46], [47]. The remainder of this paper is organized as follows: Section 2 introduces the new features for the interface of protein related complex, which contains feature representation and classification.…”

“…In view of the advantages of machine learning methods, Zhou and Yan [46], [47] designed a feature system by using hybrid information including sequence, structural, and physicochemical properties to represent the interface surfaces of protein-DNA and protein-protein interaction complexes by using SVM classifier for classification. Although, these methods achieve good results, they still have some limitations.…”

“…This work proposes a new method of feature representation for a biomolecular interface based on two strategies: one is to improve Zhou and Yan [46], [47] method based on alpha shape features, and the other is to enhance the extracted features by taking into account the influence of neighborhood via graph wavelet theory. The enhanced graph wavelet features (EGWF) outperform the other three features, "Z" (Zhou and Yan features [46], [47]), "Z"þ"FA" (improved features by the way of additional new features), and "Z"þ"GW" (improved feature using graph wavelet alone), which indicates that the combination of two enhanced strategies most effectively improve Zhou et al's "Z" feature.…”