Long-range linkage disequilibrium in the human genome

In the study of genomic variation, the nonrandom association of alleles has been a valuable tool for understanding population history, correlating disease phenotypes, and mapping genetic structural variation. These associations, termed linkage disequilibrium, have been quantified to great extent in humans over short distances. Linkage disequilibrium between genetically distant sites has garnered far less attention. These long-range patterns are quantified using a statistic pD that can distinguish true associations from chance occurrences. Using null distributions generated by randomization we detect statistically signicant long-range linkage disequilibrium on 19 of 22 human chromosomes. Given that recombination tends to break down associations in relatively few generations, this finding indicates that countervailing forces are at work. The factors that could have created the observed long-range linkage disequilibrium are epistasis, recent population admixture, genetic structural variation, and artifacts in the data collection. This analysis is unable to rule on any particular cause, but it seems admixture is the most likely to have produced the following results.