Mapping human variation and disease

In some cases, having one letter rather than another at a particular position is known to be associated with an increased risk of, for example, heart disease. Medical researchers are carrying out genome-wide disease association studies to build more of these links. HapMap is a crucial tool in these studies, because instead of going to the time and expense of looking for all the SNPs in a particular patient's DNA, researchers can simply look for the key SNPs that label particular haplotypes.

'A huge change in our understanding of the genetics of common diseases has come about over the past two years because of the possibility of measuring a large number of positions', says Professor Donnelly. 'That's given us the ability to look in an unbiased way across the whole genome. We have a much better chance of finding something – and we've learned that in the previous generation of experiments, where we were trying to pick a few specific places to look that we thought might be important, we were really bad at choosing the right places.'

Professor Donnelly chaired one of the largest genome-wide association studies, which involved 200 researchers at 50 British institutions. The Wellcome Trust Case Control Consortium looked at half a million SNPs in each of 17,000 people: 2,000 with each of seven common diseases, including coronary heart disease, depression and diabetes, and 3,000 healthy controls. The study turned up many new variants not previously linked to disease. Publishing its findings in June 2007, the Case Control Consortium has since been widely honoured, and has been named Scientific American's research leader of the year.

There are two reasons for doing this kind of study, Professor Donnelly points out. 'If you take schizophrenia, for example, we don't know many of the details about what triggers the disease or what goes wrong. If we find a difference in gene action between two variants, that might give us insights into the disease process as well as identifying the genes involved.' The second reason for doing such studies is that a patient's genetic code could potentially inform the kind of advice a doctor might give about diet or smoking, for example, or even the choice of drugs to prescribe. 'Will that improve outcomes?' asks Donnelly. 'We don't know yet. My own view is that over a medium time frame some aspects of this information will be routine parts of clinical practice.'

The huge amount of data derived from genotyping has provided the incentive to develop new statistical approaches. 'You've got a lot of information in a small sample in HapMap, and less information in a much larger sample in a disease study,' says Professor Donnelly, 'and you want to use the HapMap data to extrapolate to the things you haven't measured. That's an area, for example where my colleague Dr Jonathan Marchini has developed methods that are routinely used in disease studies.' Professor McVean, Dr Simon Myers, Lecturer in Bioinformatics, and Professor Donnelly have also developed methods that use the HapMap data to track the process known as recombination, which mixes up the genes from each parent in the next generation. 'We've known since early this decade that recombination tends to happen in hotspots', says Professor McVean. 'Our new methods have revealed combinations of letters that are associated with hotspots. It's pretty exciting.'

HapMap was succeeded in 2008 by the 1000 Genomes Project, with funding mainly from the Wellcome Trust in the UK, the National Institutes of Health in the USA, and the Beijing Genomics Institute in China. It will examine human variation in great detail by fully sequencing the genomes of 1,000 individuals. The Oxford researchers are centrally involved, with Professor McVean leading the project's analysis group. 'HapMap was aimed to capture common variants', he says. 'Rare variants are bound to be important too, and the sequencing technology will give us a chance to find those.'