Structural haplotypes and recent evolution of the human 17q21.31 region

Structurally complex regions of the human genome are not yet understood. Human 17q21.31 is a particularly interesting such locus: it contains a megabase-long inversion polymorphism1 and many uncharacterized copy-number variations (CNVs); it associates with fertility1, female meiotic recombination1-3, and neurological disease4,5; and it appears to be under selection in Europeans1. We developed a population-genetic approach to uncover complex genome structures and used it to identify nine segregating structural forms of 17q21.31. Five of these structural haplotypes are present primarily in populations with West Eurasian ancestry. Duplications of overlapping segments of the KANSL1 gene have arisen independently on the H1 and H2 forms of the 17q21.31 inversion polymorphism and reached high allele frequencies (26% and 19%) in Europeans. An older H2 form, lacking such a duplication, is present at low frequency in Europeans and in Central African hunter-gatherer populations. Relating such variation to phenotypes will be important; we show that these complex, multi-allelic structural polymorphisms can be analyzed by imputation from SNPs.