A Network Approach to Discover Motifs of Protein-Ligand Interaction for Docking and Binding Site Predictions
| Autor: | Zhong-Ru Xie, 謝忠儒 |
|---|---|
| Rok vydání: | 2011 |
| Druh dokumentu: | 學位論文 ; thesis |
| Popis: | 99 For many proteins, interaction with ligand mediates their biological function. Detection and identification of where the interactions take place and how the interactions are carried out – generally known as the protein docking problem – using computational strategies have been a major focus of structure-based bioinformatics research. In this thesis, we developed a novel approach incorporating concepts of network analysis to address this problem of structural bioinformatics. Specifically, we created a new, network-based docking scoring function, called MotifScore, which is conceptually distinct from existing scoring functions. It is non-energy-based, and docking is, instead, scored by counting the occurrences of motifs of protein-ligand interaction networks constructed from a large number of protein-ligand complex structures. In addition, during the course of developing MotifScore, we discovered a set of protein atom triplets, called protein triangles, which are preferentially present at the interface of protein-ligand binding. These binding-site-enriched protein triangles were then employed to develop a new method, called TriPred, for predicting protein-ligand binding sites. Results of MotifScore and TriPred on benchmark test datasets showed that they both performed as well as, or better than, several widely used existing methods. Therefore, the network-based approach developed here represents a novel and potentially very useful method for further studies on molecular docking. |
| Databáze: | Networked Digital Library of Theses & Dissertations |
| Externí odkaz: |
|