Figure 1

Workflow chart of the LDsplit algorithm. For each candidate SNP, LDsplit first divides the haplotype data into two subpopulations by SNP alleles. Then it calls LDhat to estimate the recombination rates for each subpopulation, and calculates the differences (Δρ) of hotspot strengths between the SNP alleles as (ρ₀ − ρ₁)/(ρ₀ + ρ₁), where ρ₀ and ρ₁ denote hotspot strengths in different subpopulations. Finally, the p-value of association is estimated by comparing the observed Δρ with the null distribution of random Δρ simulated by permutation tests.