期刊
ANALYTICAL BIOCHEMISTRY
卷 412, 期 2, 页码 173-182出版社
ACADEMIC PRESS INC ELSEVIER SCIENCE
DOI: 10.1016/j.ab.2011.01.040
关键词
GPCR classification; Multiscale energy; Pseudo-amino-acid composition; Physiochemical properties; Rhodopsin-like receptors
资金
- Higher Education Commission (HEC) of Pakistan [074-1844-PS4-406]
G-protein-coupled receptors (GPCRs) are the largest family of cell surface receptors that, via trimetric guanine nucleotide-binding proteins (G-proteins), initiate some signaling pathways in the eukaryotic cell. Many diseases involve malfunction of GPCRs making their role evident in drug discovery. Thus, the automatic prediction of GPCRs can be very helpful in the pharmaceutical industry. However, prediction of GPCRs, their families, and their subfamilies is a challenging task. In this article, GPCRs are classified into families, subfamilies, and sub-subfamilies using pseudo-amino-acid composition and multiscale energy representation of different physiochemical properties of amino acids. The aim of the current research is to assess different feature extraction strategies and to develop a hybrid feature extraction strategy that can exploit the discrimination capability in both the spatial and transform domains for GPCR classification. Support vector machine, nearest neighbor, and probabilistic neural network are used for classification purposes. The overall performance of each classifier is computed individually for each feature extraction strategy. It is observed that using the jackknife test the proposed GPCR-hybrid method provides the best results reported so far. The GPCR-hybrid web predictor to help researchers working on GPCRs in the field of biochemistry and bioinformatics is available at http://111.68.99.218/GPCR. (C) 2011 Elsevier Inc. All rights reserved.
作者
我是这篇论文的作者
点击您的名字以认领此论文并将其添加到您的个人资料中。
推荐
暂无数据