Inter-chromosomal linkage disequilibrium across rice populations

Anthony Greenberg, Bayesic Research, accounts@bayesicresearch.org
Diane Wang, Dept. of Agronomy, Purdue University, dianewang@purdue.edu

Keywords: population genetics, nucleotide variation, linkage disequilibrium

Levels of linkage disequilibrium (LD) among unlinked sites reflect the interplay of population genetic forces shaping the history of a species. For example, theory suggests that LD is more sensitive to population stratification than sequence divergence at individual sites. However, enumerating pairs of unlinked correlated loci normally requires the examination of all pairwise relationships in a genome. This task becomes computationally prohibitive with modern data sets comprising millions of polymorphisms. We have developed an approximate hash table based method to identify groups of loci in LD across the whole genome that does not require computing associations between uncorrelated pairs. We used this method to study LD among unlinked sites in three major domesticated rice populations: indica, aus, and tropical japonica. These populations vary in the extent of local linkage disequilibrium, with indica the lowest and tropical japonica the highest. We applied our method to accessions from the three populations that were sequenced as part of the 3K genome genome project. We recapitulate the previously-described ranking of these populations by overall levels of LD. We also find more linkage disequilibrium among unlinked sites than expected by chance. We further attempt to tease apart the influences of selection and demography by identifying highly correlated locus pairs that are common among populations.