Sunday, August 15, 2004

How Population Genetics Can Mislead

This story illustrates the way in which the sampling techniques and assumptions used by population geneticists can give rise to an extremely misleading picture of the genetic makeup of populations.

Scientists are to launch a £2 million study to uncover the genetic make-up of the British people. The aim is to find tell-tale pieces of DNA that will reveal the influences - including those of the Vikings, Saxons and Celts - which have shaped regional populations.

Researchers will take blood from thousands of volunteers from Cornwall to Shetland and from Kent to Tyrone. The samples will then be used to isolate key pieces of DNA that best identify how the different races, tribes and invaders have influenced the country's modern population.

This data will not only be used to create a genetic history; it will generate vital information about patterns of diseases such as diabetes, Alzheimer's, heart disease and other conditions in various parts of the country. Such data could also provide researchers with ways of identifying people at risk of succumbing to common illnesses.

The project - to be carried out by Oxford University researchers and funded by the Wellcome Trust - is intended to provide a complete overview of the nation's genetic structure, though researchers say only those living in the countryside will be asked to take part.

'Urban populations are already far too mixed up for us to be able to tease out their genetic roots,' said the project's leader, Sir Walter Bodmer. 'However, in the country there has been much less genetic input from "outsiders" over the past century. People there will give us a far better picture of ancient population patterns.' (emphasis added)
"What's the problem then?" one might ask, and the answer is in that final paragraph. These researchers aren't going to bother sampling urban populations because they don't want to deal with the complexities introduced by the global mixing that's been going on in an age of high international mobility, but the thing is - Britain is a predominantly urban country.

By far the great majority of Britons are urbanites, and if they're excluded, this survey will only provide information that is of primarily historical interest; it won't tell us what the primary disease risks are likely to be for most Britons, and it certainly won't tell us much worth knowing about the genetic makeup of the average Briton, yet someone dared to headline this article with the title "Scientists go in search of the true Brit". Are we to take it then that the majority of Britain's citizens are "false", ersatz or counterfeit Brits?

The survey detailed in this paper won't be the first one to have ignored urban populations for the sake of tractability, as the well-known dataset assembled by Luigi Cavalli-Sforza et al. did much the same. The problem with this sort of data is that those who are most eager to misuse it aren't the sorts of people to carefully read through the accompanying notes to understand the assumptions at play: instead they'll leap straight from looking at data that describes isolated populations that are untypical nearly everywhere to making bold claims about the genetic differences between nations, races, etc. The reality is that if urban populations were taken into account in most surveys, we'd realize that the genetic distinctions between most populations are undergoing a significant decline over a short period of time, so rapid that I expect that the makeup of the average citizen of the world of 2200 will look much the same whichever part of the world one might care to look.