Proteins, nucleic acids and other biological macromolecules take part virtually in all processes within living organisms. Knowledge of their 3-dimensional structures is essential for understanding how they work. X-ray crystallography is a powerful experimental technique that gives 3D structures with high accuracy. According to the Protein Data Bank more than 85% of all structures have been analysed using this technique.

Our research is centred on the development of efficient mathematical, statistical, computational algorithms for Macromolecular X-ray crystal (MX) structure analysis. We implement the developed algorithms in the software tools and distribute them to the structural biology community. Most of our software is distributed to the community via a UK based crystallographic software initiative - CCP4.

We use the Bayesian method, which combines prior structural and chemical knowledge with the MX data and allows extraction of biologically relevant information from noisy diffraction data. Bayesian statistics has two components: likelihood through which new information from the experimental data to the model transferred and prior probability distribution that ensures models' consistency with the knowledge available about them.

Two examples of software developed in our group are: a maximum likelihood refinement program - REFMAC and an automatic structure solution pipeline - BALBES.