LDhat

LDhat is a software package for estimating historical recombination rates from population genetic data. It implements a coalescent-based, linkage-disequilibrium (LD) informed method to infer the population recombination rate, ρ = 4N_e r, along a contiguous genomic region. The approach uses patterns of LD in a sample of haplotypes or genotypes to construct a likelihood surface for ρ under a neutral model with recombination.

LDhat consists of several programs, most notably interval, which estimates ρ across a window by comparing observed haplotype configurations to those expected under a hidden genealogical process. The method relies on a precomputed lookup table of likelihoods for combinations of θ (the population mutation rate) and ρ, enabling relatively fast evaluation across many windows. Users provide aligned sequence data; the output includes point estimates of ρ for each window and, optionally, a recombination map showing rate variation along the region. The package is designed for Unix-like systems and is implemented in C; it requires phased or haplotype data and user-specified θ or an estimate thereof.

Limitations include reliance on a composite likelihood that approximates the full data likelihood, sensitivity to sample size and SNP density, and assumptions of neutral evolution without strong selection or population structure. It can be computationally intensive for large regions, and modern alternatives offer improvements in speed and modeling.

LDhat has been widely cited in population-genetics studies and remains a reference point for recombination-rate inference from LD. It is commonly used for small to moderate genomic regions and as a teaching example of coalescent-based LD methods.