Until the middle of 2006, the search for new genes that influence disease was pretty much restricted to studies of extended families. Typically, geneticists would identify pedigrees in which a particular type of cancer or heart disease was unusually common and look for parts of the genome that affected individuals have in common. This approach, called linkage mapping, has been the main method for finding single gene disorders, but has had limited success for more complex diseases.
Your parents have between them four copies of every gene. You have two of these, and your children each have a 50-50 chance of receiving each one. Suppose now that you, your father, and two of your three kids have a heart murmur, and both of these kids received the same allele from you, which is also the one you got from your Dad, while their sibling received the other allele. So, four out of seven members of the family have the murmur, each of whom has the same allele of the gene. Something fishy is going on, and you would likely conclude that the level of correspondence between having the allele and having the disease is unlikely to be due just to coincidence. You would suspect that the allele actually causes the murmur.
However, since there are thousands of genes and millions of families with murmurs, that level of coincidence is bound to occur occasionally. But if geneticists find a similar correspondence in dozens of even bigger pedigrees, their confidence that the particular allele of the gene actually causes or at least contributes to the murmur increases. With enough data, the correlation between the gene and the disease does not have to be 100 percent. As a result, it is also possible to detect linkage between regions of the genome and complex diseases where each gene only has a small influence on the disease. On this basis we know, for example, that a dozen or so places in the genome influence type 2 diabetes. For reasons we needn’t concern ourselves with here, those places typically stretch over perhaps a tenth of a chromosome, or hundreds of genes. So they do not pinpoint the problem.
To get around this, the field has now turned to a revolutionary approach called genomewide association mapping, or GWA. Instead of looking in families, geneticists now look at unrelated individuals drawn from an entire population. Two companies, Affymetrix and Illumina, have manufactured little gene chips with up to a million common genetic differences printed on them. These markers stand as proxies for the tens of millions of places in the genome that are different among people. For less than $1,000 a pop, geneticists can now effectively measure what a person’s genetic constitution is, almost as if they were determining the sequence of the person’s entire genome.
For a few million dollars geneticists can go out and compare the genomes of 10,000 people who have a disease, with the genomes of 10,000 people who do not have it. If the frequency of the A at position 102,221,163 on chromosome 11 is 29 percent in people with a heart defect, but only 19 percent in people without the defect, then after appropriate crunching of the numbers we can infer that this site is contributing to the problem. This is a gross oversimplification; all sorts of possible alternative explanations can be made for such a difference. But if another group replicates the result in an independent sample (often from another country), then confidence that the gene is involved in the disease shoots up still higher.
It turns out that this approach is a sufficiently fine genetic scalpel that it actually leads us to one or a few genes involved in the disease. Genomewide association scans for disease will be to human genetics what the microscope was to nineteenth century biology, and what they are telling us is rightly the subject of the remainder of this book.