POSTER NO: 218

How many SNPs do we need for whole-genome linkage disequilibrium mapping?

¹Andres Metspalu, ²Maido Remm
¹Institute of Molecular and Cellular Biology, University of Tartu, 23 Riia St. Tartu 51010, Estonia, ²Estonian Biocentre, 23 Riia St. Tartu 51010, Estonia

Whole-genome linkage disequilibrium mapping and haplogroup description are the prerequisites for genome wide association studies. Several population based health database collections together with DNA samples are in progress. The rationale is that by using this new resource and relevant computing tools human genes associated with the common diseases could be discovered. There are many questions still, but one of the most pressing is the number of the SNPs to be typed for LD map. Data from chr 21 and chr 22 LD mapping and modeling are demonstrating that first we have to establish the LD block boundaries with ca 400-500 000 SNPs from at least 100 individuals and then to type large population samples with minimal number well selected SNPs which are required to reveal the major (3-4) LD blocks at least in 80% of cases. We will demonstrate how to calculate the minimal number of the SNPs for each chromosome and present the algorithms. We also propose the two-stage process for the whole genome LD mapping starting from the exons fist and expanding it to intergenic areas later. We see as a major problem the lack of the SNPs for the 15% of the exons in the public databases. The next problem is the low minor allele frequency (less than 10%) for 30% of the SNPs in the Caucasian population which will increase the cost of the assay development.