Biomolecular Modelling Laboratory

Previous and current research
With numerous organisms now fully sequenced there is a wealth of information about cellular biology at the molecular level. It is now the goal of structural genomics to map this information to the three-dimensional structures of all proteins, particularly those of the human genome. There are too many human proteins (∼ 35,000 protein genes) for the X-ray crystallographers to determine a structure for each. However, efforts, on an international scale, are now under way to experimentally determine a set of unique protein folds from which all other proteins can be modelled.

The main task of the Biomolecular Modelling Laboratory is to generate these atomic models to the highest level of accurately possible. To aid this task a set of computer based algorithms have been developed (program 3D-JIGSAW). These algorithms are continually under development. Recently we improved the three main problem areas of model building: selecting the best parents (X-ray structures) from which to model, increased the accuracy of sequence alignments and made improvements to protein loop modelling. All these improvements are key to assigning and understanding the functions of the proteins modelled. Protein-protein interactions are central to many biological processes such as metabolic pathways, signal transduction and the immune response. Understanding these interactions will aid our understanding of human diseases, such as cancer, and highlight key protein targets for rational drug design.

Work is continuing with rationalising and predicting protein contact interfaces and several computational tools have been developed which will be maintained and improved. New developments include an analysis of the role water molecules play in the mediation of protein-protein interactions. Computer based algorithms developed by the group are constantly tested outside Cancer Research UK, for example in the international blind trials: Critical Assessment of Techniques for Protein Structure Prediction (CASP).

Future projects
Computer based algorithms will be continually developed and refined to gain ever increasing accuracy in the prediction of protein structure and function. The wealth of biological information within databases, such as metabolic pathways, will be linked to our model databases. Predicted protein-protein interactions need to be correlated with experimental data generated for example by yeast two-hybrid and the new microarray technology.