Informed by preliminary analyses of these data, the 1000 Genomes Project is studying the collection of samples from five populations within each continental region. Our data suggest that a strategy of identifying polymorphic SNPs and CNPs followed by imputation in densely genotyped samples can provide information even for lower-frequency alleles. Necessary components of such a reference panel include accurate genotyping and characterization of the haplotype background for the alleles (which included here the use of pedigree information to inform phasing), and a broad range of reference populations to capture geographically local variants. The ultimate utility of such a strategy (as compared to a more complete approach using exome or whole genome sequencing) will depend on the as yet poorly characterized distribution of causal alleles across traits, across exons as compared to non-coding regions, and the relative cost and accuracy of sequencing as compared to genotyping followed by imputation. The development of a robust reference panel will be a necessary step in the evaluation of these different strategies across a wide variety of diseases.
