Thursday 24 February 2011

Computational Biology - and Bioinformatics - at UCL

People not uncommonly mix up the term "bioinformatics" with "computational biology", or use the two terms interchangeably. This is an easy mistake to make as both disciplines involve a computational approach to life sciences, and both terms have evolved over time. The consensus, however, is that there are significant differences: bioinformatics refers to the analysis of large quantities of (generally molecular) biological data, whereas all research fields that involve the development and use of computational approaches to them study of biological problems can be grouped into coomputational biology. Thus, although this is not precisely correct, it is more accurate to think of bioinformatics as a sub-division of computational biology than the other way round.

Birkbeck's much larger neighbour, University College London (UCL) has over twenty research groups working in areas that fall within the remit of computational biology, scattered across a number of departments and locations. Last week the college's whole computational biology community came together in a one-day symposium to make connections between these diverse research groups and to promote and celebrate the wide range of computational biology research that is carried out at UCL. This post gives a very brief overview of some of the work presented there, with links to the research groups involved.

The symposium was chaired by David Jones, director of the Bloomsbury Centre for Bioinformatics. which includes researchers from both UCL and Birkbeck. It started with a keynote lecture from Steve Oliver of the University of Cambridge, describing his group's ambitious project to understand and model completely the metabolism and behaviour of a simple, single celled organism, the yeast Saccharomyces cerevisiae.

The first UCL speaker was Christine Orengo of the Research Department of Structural and Molecular Biology. In the 1990s, she, with Janet Thornton (now director of the European Bioinformatics Institute) developed the CATH protein structure classification database which is very widely used in PPS. Since then, researchers in her group have built more databases of protein structure and function and prediction tools, and have moved on to the analysis of complete genomes and (as she presented here) functional protein networks. Domenico Cozzetto of the Department of Computer Science then described novel methods of predicting protein function from multiple data sources.

The research presented in both these talks clearly falls within the remit of "bioinformatics", in that it is concerned with the analysis of large quantities of molecular data. The next speakers, however, illustrated just how widely the term "computational biology" is being applied. Peter Hammond trained as a mathematician but is now working at UCL's Institute for Child Health, using imaging techniques and mathematical models to determine the subtle effects of genetic differences on human face shape. These models are already being used to aid early diagnosis of developmental disorders, facilitating both early intervention and genetic counselling. His presentation was followed by two more with a medical focus, by Angus Silver, a neuroscientist who develops mathematical models of neuronal signalling in the cortex, and a physician, Malcolm Finlay from the Heart Hospital, who described the computer simulations that his group has developed to predict the electrochemical responses of individual patients' heart muscles during periods of abnormal heart rhythm (arrhythmia).

A later talk by Sally Price of UCL's Department of Chemistry illustrated the value of computational biology to the pharmaceutical industry. She described the use of inter- and intra- molecular forces (to be covered in PPS section 9) to determine which crystalline structures of chemicals, including prescription drugs, would be most likely to form. For me, however, one of the highlights of the day was a talk by Mark Girolami of the Department of Statistical Science that linked computational biology, biostatistics and genetics to archaeology and anthropology. Mark described how a genetic mutation that allows some Europeans to digest milk as adults spread through the continent. His models traced the emergence of the mutation (which would not have remained in the population if it had not conferred significant evolutionary advantage) to a time and a place - about 7-8,000 years ago and in what is now central / Eastern Europe - when cattle replaced sheep and goats as the main domesticated animals.

I hope that this partial overview of a fascinating day's science will give you some idea of the breadth of computational biology research, and the depth of its coverage at UCL.