Selective Sweep: Haplotype Wall and Diversity Profile (2D)

A new beneficial allele sweeps through a population and drags its neighbours along: watch the haplotype wall lose diversity around the selected site.

Switch to 3D
Controls

Presets

Parameters

Model checks

The science

A selective sweep. When a new beneficial mutation spreads through a population, it drags along the stretch of chromosome around it: the neutral variants that happened to sit on the same chromosome rise to high frequency with it (genetic hitchhiking, Maynard Smith and Haigh 1974). Diversity next to the selected site collapses and recovers with distance because recombination, shuffling pieces of different chromosomes, lets sites escape. The result is a valley of heterozygosity centred on the selected site — the signature used in genome scans for recent positive selection.

Model. A haploid Wright-Fisher population of N chromosomes with 41 sites. The middle site carries the beneficial allele, fitness 1 + s, starting on one chromosome. Each generation N offspring are drawn from parents weighted by fitness; with probability (L-1)r an offspring is a crossover between two parents. In the deterministic limit the odds of the allele multiply by 1 + s per generation, so the sweep from 10 % to 90 % takes ln 81 / ln(1 + s) generations. A single new copy is usually lost to drift: it survives with probability π solving 1 - π = exp(-(1+s)π), roughly 2s, and the simulation restarts when that happens.

How wide is the valley? A neutral lineage on the beneficial background escapes by recombining onto a wild-type chromosome at rate k r (1 - p(t)) for a site k intervals away; integrating over the logistic sweep gives the chance exp(-k r ln N / ln(1+s)) that it rides along completely. The valley is therefore about s / ln N wide in recombination units: strong selection and weak recombination give wide, deep valleys.

Simplifications. One selected locus, fixed recombination rate with at most one crossover, constant population size, haploid genomes, no mutation during the sweep and a random starting haplotype structure (real populations have linkage disequilibrium before the sweep). The valley formula is an approximation, so that check has a stated tolerance. Checks: logistic odds growth, neutral diversity loss (1 - 1/N)t, linkage disequilibrium decay (1 - r)t, survival probability of a new copy against a one-locus simulation, the sweep time, and the hitchhiking formula.

-
-