GWAS here, last year

摘要

Often, at the start of a new year, a journal editor tends to prelude on what this year might bring. In this case, however, I would like to reflect on the past year. Indeed, a major development in the field of Human Genetics has been is the arrival of the Genome-Wide Association Studies or GWAS. Many previous years have seen reports of GWAS, and a frequently heard complaint has been, with some exceptions, that the findings were met with irregular or downright controversial replication results. If anything, the year 2007 has been the year of the publication of a large amount of rapidly and widely replicated GWAS. Notably, this has taken place in the field of diabetes, with now over 20 loci identified in total. The largest harvest for 1 year reaped by type 2 diabetes with now in total 11 loci confidently identified, six of which last year by GWAS; type 1 coming in second, with five loci last year on a total of 11 as well, and additional loci for prostate cancer and rheumatoid arthritis, the latter replicated in parallel by candidate gene analysis. And with bonuses coming along: dissection for body mass index has yielded the first obesity gene, FTO, the prostate cancer risk allele is protective for type 2 diabetes, and some type 2 diabetes risk genes when highly expressed, looks like predisposing for colon cancer when poorly expressed. These findings - and I count on correspondence telling us that we overlooked other examples - supports the concept, of connectivity: our biology does not respect classical clinical discipline boundaries, and the key switches are rather involved in maintaining balance, with quite different pathology following dearth or excess. The main accelerator of these breakthroughshas been the development of the HapMap resource and the Perlegen SNP set, heirs of the early days SNP consortium. All in the wake of the Human Genome Project, resulting, within half a decade, in robust, very high-density SNP detection platforms of Affymetrix and Illumina. And as each of these is derived fromthe same HapMap set, tools have already been developed to jointly analyse large study samples typed by either platform by imputation, computing missing values by interpolation (eg, Servin and Stephens).