recombination & selection

O'Reilly PF, Birney E & Balding DJ 2008 Confounding between recombination and selection, and the Ped/Pop method for detecting selection. Genome Res 18:1304-1313.

• we propose a novel genome-wide method for detecting selection("Ped/Pop") that identifies genomic regions with a markedly lower population-based than pedigree-based estimate of recombination rate

• actual estimates of s in humans typically range up to 0.1, the approximate intensity of selection reported at LCT
• estimates of s up to 0.3 have been reported at other loci
• the total branch length of the coalescent tree, which is approximately proportional to the number of detectable recombinations, at a locus subject to selection of strength s = 0.01 in a population of size 10⁴ is 49% of that under neutrality

• in some cases the low rate of recombination could have contributed to a spurious signal of positive selection
• a statistic sensitive to recombination rate will incorrectly estimate the selection coefficient at every locus due to dependence on the local recombination rate
• Weir et al. (2005) apply several strategies to exploit the standard measure of population differentiation, F_ST, to detect selection
• they used mean F_ST over 5-Mb windows
• the confounding effect of recombination rate, which is absent at a single SNP, is introduced by this averaging due to the correlation between neighboring SNPs
• 5-Mb windows ensure that variability in recombination rates is reduced
• the impact that selection has on such large regions is likewise reduced

• there are several limitations to the LRH approach
• it assumes that recombination rates are not sequence-specific
• there is evidence that different haplotypes can have different recombination rates