Protein modelling using an object-oriented database
This thesis describes extensions to an object-oriented database, and the use and benefits of the resulting system in protein structure analysis and homology modelling. An object-oriented database based on a semantic data model (the functional data model) is used to store protein structure data. The logic programming language Prolog, together with the object-oriented database, is useful in protein structure analysis applications that require a combination of data retrieval and calculation. This is demonstrated by the implementation of a procedure for identifying clusters of interacting residues in hydrophobic protein cores. Knowledge of the three-dimensional structure of a protein is important to a full understanding of its function. Experimental determination of a protein's structure is a slow and difficult process, so there is a demand to be able to generate hypothetical computer models of proteins. Homology modelling is the process of constructing a model of a protein based on predicted similarity to a protein with a known structure. A homology modelling system is described in which the object-oriented database is used to manage working data generated during the modelling process. The organisation of this working data is described. Both the database system and the Daplex query language have been extended to support parameterised action methods. These are pieces of code, typically for generating a display or performing database updates, that are stored in the database with the data. Tasks performed in homology modelling are expressed as action methods, and can be invoked using Daplex. The utility of the homology modelling system is demonstrated by the construction of models of domains II and VI of the complement proteins Cls and Clr.