A population-genetic test of founder effects and implications for Ashkenazi Jewish diseases

Abstract

A founder effect can account for the presence of an allele at an unusually high frequency in an isolated population if the allele is selectively neutral and if all copies are identical by descent with a copy that either was carried by a founder individual or arose by mutation later. Here, a statistical test of both aspects of the founder-effect hypothesis is developed. The test is performed by a modified version of a program that implements the Slatkin-Bertorelle test of neutrality. The test is applied to several disease-associated alleles found predominantly in Ashkenazi Jews. Despite considerable uncertainty about the demographic history of Ashkenazi Jews and their ancestors, available genetic data are consistent with a founder effect resulting from a severe bottleneck in population size between a.d. 1100 and a.d. 1400 and an earlier bottleneck in a.d. 75, at the beginning of the Jewish Diaspora. The relatively high frequency of alleles causing four different lysosomal storage disorders, including Tay-Sachs disease and Gaucher disease, can be accounted for if the disease-associated alleles are recessive in their effects on reproductive fitness.

Figure 1

Illustration of an intra-allelic genealogy for the case with i=4. t₁ is the age of the allele A, t₂ is the time of the most recent common ancestor (MRCA) of copies of A in the sample, and w is the length of branch on which A arose by mutation.

Figure 2

Estimated posterior probability distributions of the number of ancestral lineages, m, carrying N370S at GBA (which causes Gaucher disease) for the nine combinations of N₀ (the population size at a.d. 70) and N₁ (the population size at a.d. 1348) and two different hypothesized times of founder events (a.d. 75 and 1350).

Figure 3

Scatterplot of t₂ (the coalescence time of all copies of A) and j_o (the number of nonrecombinants) generated in 1,000 replicate forward simulations of a neutral allele that was in one copy t₁ generations in the past and was found at t=0 with a frequency between 0.027 and 0.037 (i.e., x=0.032±0.005, which was chosen to match the frequency of N370S at GBA, 0.032). In each replicate, the allele age (t₁) was chosen randomly from a probability distribution on t in which Pr(t) is proportional to N(t), where N(t) is based on table 1 with N₀=600 and N₁=3,000. Parameters for A and the linked marker are the same as for N370S at GBA: i=268, j_o=163, p_N=0.205, and θ=0.007 (cf. table 2).

Figure 4

The probability that an allele present in one copy at t=0 is still segregating and has a frequency of at least x 26.08 generations later. Calculations were made using the theory described in the appendix.