Alpha shape based methods for analysis and prediction of biomolecular interactions

Abstract

Biomolecular interactions such as protein-DNA, protein-protein and protein-ligand interactions play significant roles in almost every living organism. Studying these interactions is important in understanding the formation of diseases and the discovery of new drugs. However, the underlying principles of molecular recognition processes are still largely unknown. In recent years, more three dimensional biomolecular structures have become available, which has made the studies of biomolecular interactions much easier. There is a strong demand for geometric models to study biomolecular interactions. In this dissertation, we introduce the alpha shape which is a powerful tool to represent biomolecular structures and has advantages in characterizing the surface curvature and atom contacts. Based on the alpha shape, we develop novel methods to analyze and predict protein-DNA, protein-protein and protein-ligand interactions. The first part of this dissertation presents the prediction of protein-DNA interaction based on alpha shape modeling. Alpha shape models are applied to reconstruct the interfaces of protein-DNA complexes. Features such as atom type, residue type and surface curvature are extracted from the interfaces to characterize these protein-DNA complexes. Based on the interface features, we develop a scoring function to estimate the probability of the correctness of a protein-DNA complex using conditional probability theory. The second part of the dissertation presents the application of the alpha shape in recognition of protein-binding and DNA-binding proteins. In this part, structural alignment is employed to obtain the potential binding mode of protein-protein or protein-DNA complexes. Alpha shapes are used to reconstruct the interfaces of potential structures and we extract a number of structural and physiochemical features to characterize the interfaces. A support vector machine is applied to build classifiers to recognize protein-binding and DNA-protein binding proteins. The third part of the dissertation focuses on the analysis of hydrogen bonding in protein-ligand interactions. Although previous studies have demonstrated the importance of hydrogen bonding in protein-ligand interactions, the physical role that a hydrogen bond plays and its geometric characteristics are still not well understood. We apply the alpha shape to reconstruct the interface in a protein-ligand structure and use solid angles to analyze the geometric properties of the hydrogen bonds. Geometric patterns are found between atoms in the formation of hydrogen bonds. The fourth part of the dissertation aims to analyze the relationship between periodic dinucleotides and the spatial structure of nucleosomes. In this part, an alpha shape is used to model the nucleosome structure, and we show that the periodic DNA dinucleotides AA, TT and GC occupy special positions in nucleosome structure with one nucleotide inside and the other outside the nucleosome surface. This structural feature and other dinucleotide characteristics can provide useful information for the study of nucleosome positioning. In conclusion, this dissertation contributes to untangling the principles of biomolecular interactions and provides practical tools for the analysis and prediction of protein-DNA, protein-protein and protein-ligand interactions.

Date of Award	2 Oct 2013
Original language	English
Awarding Institution	City University of Hong Kong
Supervisor	Hong YAN (Supervisor)