Genes means stats
Ruth Collier | 22 Jan 08
The launch of the 1000 Genomes Project delivers another reminder that UK scientists often don’t get their fair share of the glory from major international collaborations. As the project is the successor to HapMap, which was itself successor to the Human Genome Project, it’s pretty important stuff. Oxford’s involvement deserves a closer look because it highlights what a big role statisticians play in modern medicine. Here at Oxford we’ve got one of the largest concentrations of statisticians-turned-self-taught-geneticists in the world – and they get involved in all the big gene projects. We quizzed statistician Gil McVean about why the statisticians are important to medicine...
OxSciBlog: So why the need for the maths?
Gil McVean: In these very large projects it’s essential to get the data analysis part right. It’s not enough just to be a good analyst, though – you have to really understand the biology. If you don’t, it’s a recipe for disaster. We’re pretty unique at Oxford in having such a large concentration of people who can do both. I do it, and there’s Peter Donnelly (he was heavily involved in HapMap and chaired the Case Control Consortium), Jonathan Marchini, Simon Myers, and many others.
OSB: Why statistical analysis, though?
GMcV: Well, what are statisticians good at? They’re good at looking for patterns in large, noisy datasets. That’s what a genome is. And you’ve got an even larger, noisier dataset when you’re looking at several genomes. Now, some of that noise is coming from errors in the genomic technologies, which after all are very new. We want to be certain that errors are not introduced by the machines, and statisticians are good at looking out for anomalies in the data and weeding them out. But then are the much more interesting biological questions. The patterns in genetic variation are produced by evolution. If you can analyse those patterns, you can tease out the evolutionary history. It’s a very complex process and requires “number people” to do it.
The other thing is that these genetic projects are driven by medical goals, yet it’s not immediately obvious how you go from these huge datasets to medical science. It requires clever modelling and new statistical thinking to turn the data into something medically relevant and useful. Statisticians, by which I mean in large part the people here at Oxford, have made major inroads into that.
Gil McVean is co-chair of the analysis group in the 1000 Genomes Project and played a major role in HapMap. He and the other researchers mentioned work both in Oxford's Department of Statistics and Oxford's Wellcome Trust Centre for Human Genetics.
Your comments
There are currently no comments on this page.