Study On The Structure And Function Of Membrane Protein Receptors Based On The Integrated Intelligence

Proteins are essential parts of organisms and participate in virtually every process within cells. It is highly desired to develop some powerful computational tools and effective methods and solve the problem with the massive growth of the protein sequence after human genome project. Hence, many intelligent algorithms and Web servers are widely used on proteomics and bioinformatics. However, the computation methods are just developed in membrane proteins, due to the complexity of the structure and function of the protein. Membrane receptor, which is the largest class of membrane protein, plays an important role in biology signal transduction. It is meaningful for studying the structure and function of the membrane receptor and the proteins related. Thisthesis mainly focuses on several important issues in prediction of membrane protein receptor structure, the coupling between membrane protein receptors, the drug and membrane protein receptors interaction network, apoptosis protein subcellular localization and transcriptional factor co-regulation network. We aim to develop the integrated approaches to predict membrane protein receptors function from its sequence. The main contributions in the thesis are listed as follows:On the study of the structure class of membrane protein receptors, G-Protein Coupled Receptors (GPCRs) are adopted as our objects. According to the concept of Dipeptide composition model and Binary Particle Swarm Optimization (BPSO), we propose a novel approach for protein sequence feature selection. An ensemble classifier is proposed, in which the basic classifier is Fuzzy K Nearest Neighbors (FKNN) algorithm. The training dataset is from Web server GPCRDB. Finally we build a web server for the researchers'online study.On the study of the coupling of membrane protein receptors, we use GPCR-G-protein coupling as objects. A novel approach of improved LogitBoost Classifier (ILC) with Self Adaptive Immune (SAI) algorithm for optimizing the parameter of ensemble method is introduced on the prediction. The dataset is from Web server GRIFFN. The Jackknife tests indicate that the proposed approach is practical.There are few studies on the drug and membrane protein receptors interaction network. Based on the framework of semi-supervised learning and integrated intelligence, a SemiBoost method is used for reconstruction of the network. Then, we discuss the network from the view of complex network theory. The results are validated in the latest version of Web server KEGG and SuperTarget.Membrane protein receptors play central role in the process of apoptosis protein subcellular localization and transcriptional factor co-regulation. In apoptosis protein subcellular localization, we propose an approach of Improved Pseudo Amino Acids Composition (IPseAA) in which the weight factors of the model are optimized by immune genetic algorithm. Based on the approach of IPseAA, a novel vector is developed to represent the feature of protein which incorporates the concept of Approximate Entropy (ApEn) and hydrobolicity pattern. An alogrithm of ensemble of AdaBoost classifier is adopted for prediction. The better results compared with the prior works indicate that the proposed approach is more effective.Meanwhile, we explore the linking between co-regulation and co-expression impact by membrane protein receptors, develop a co-regulaiton simulation algorithm, build the co-regulation protein network based on Partial Correlation coefficient with Information Theory (PCIT) and Regulatory Impact Factor (RIF), and analyze the network from the theory of complex network and entropy of network ensembles. The proposed dataset is from Commonwealth Scientific and Research Organization (CSIRO). It is the first test to explore the co-regulation in non-model organism.At the end, we summarize of content, advantage and the deficiency of the thesis, and narrate further development of the study.