Genome-wide association scans (GWASs) have proven to be a successful strategy for detecting common variants exerting modest effects on complex disease risk. Currently available commercial platforms focus on common variants and capture the majority of HapMap1 SNPs with minor allele frequency (MAF) >0.05 in European populations.2 Several large-scale consortia have been formed in order to carry out GWAS meta-analyses for various phenotypes, with successful outcome (eg, Zeggini et al,3 Prokopenko et al,4 Franke et al,5 Barret et al6 and Soranzo et al7). To enable the combination of data across studies carried out on different platforms, and to enable in silico fine mapping of association signals, imputation approaches were proposed a few years ago8 as a means of statistically inferring genotypes at untyped loci using a reference set, for example, the HapMap (∼2 500 000 SNPs).
