Resolution of haplotypes and haplotype frequencies from SNP genotypes of pooled samples

Recent efforts to characterize genetic variation indicate that humans share large chromosomal blocks, along which little to no recombination is observable. Thus, on a segment-by-segment basis, only a handful of haplotypes account for most human genotypes. Currently, the challenge of registering haplotypes and their frequencies is met by genotyping individuals one by one, a process which is overall resource intensive. Instead, we propose utilizing the ability of current genotyping technologies to pool DNA samples and output allele frequencies of SNP markers. We enable inference of haplotypes and haplotype frequencies from such pooled data, by novel computational methods. This strategy harnesses the economics of pooling for the task of haplotyping, potentially offering a 10-20-fold saving in genotyping reactions performed.We show that a small number of pools can be used to accurately and cost-effectively reconstruct a haplotype block and determine haplotype frequencies.