Genomes from a medieval mass burial show Ashkenazi-associated hereditary diseases pre-date the 12th century

Abstract

We report genome sequence data from six individuals excavated from the base of a medieval well at a site in Norwich, UK. A revised radiocarbon analysis of the assemblage is consistent with these individuals being part of a historically attested episode of antisemitic violence on 6 February 1190 CE. We find that four of these individuals were closely related and all six have strong genetic affinities with modern Ashkenazi Jews. We identify four alleles associated with genetic disease in Ashkenazi Jewish populations and infer variation in pigmentation traits, including the presence of red hair. Simulations indicate that Ashkenazi-associated genetic disease alleles were already at appreciable frequencies, centuries earlier than previously hypothesized. These findings provide new insights into a significant historical crime, into Ashkenazi population history, and into the origins of genetic diseases associated with modern Jewish populations.

Keywords: Ashkenazi; Britain; Jewish; Medieval; ancient DNA; genetic disease.

Figure 1

Details of the archaeological and radiocarbon information from the Chapelfield well deposit (A) Location of the site in Norwich, UK. (B) West-facing vertical section drawing of the Chapelfield well shaft. (C) Probability distribution for the date of deposition of the human remains based on 5 radiocarbon dates modeled as a single event using the Combine function in OxCal 4.4 and the IntCal20 curve. 1165–1179 CE (31.4% probability), 1190–1207 (36.8% probability), and 1161–1216 CE (95.4% probability). χ² test, df = 4, T = 4.882 (5% 9.488); agreement, n = 5; Acomb = 75.8% (An = 31.6%),^, from isotopic data in Table S1. See also Method details: Radiocarbon dating.

Figure 2

Runs of homozygosity inferred among the Chapelfield individuals using hapROH The stacked plot on the right shows total lengths of RoHs for different length categories in the six sequenced Chapelfield individuals. Plots on the right show expected sum of RoH lengths for close kin (“Recent Loops”) and under small population size scenarios. Detailed plots of RoH length distributions and autosome locations for SB604, SB605, and SB696 are provided in Figure S3, with comparisons indicated for certain recent inbreeding scenarios.

Figure 3

Principal component analysis (PCA) exploring the genetic affinities of the six Chapelfield individuals We projected the six Chapelfield genomes on a PCA defined by variation among modern western Eurasian population samples, including modern Jewish individuals. All six Chapelfield individuals project well away from present-day British samples, as well as northern Europeans more generally. Instead, they partially overlap with Southern Europeans, close to Cypriots, modern Ashkenazi, Turkish, and North African Jews. These results are consistent with the Chapelfield individuals having Jewish ancestry (cf. Kopelman et al.³⁸).

Figure 4

Pairwise coalescence rates between European, Middle Eastern, modern Ashkenazi, and Chapelfield individuals (A) Multidimensional scaling (MDS) of pairwise coalescence rates. Rates between the chromosomes of a single individual are not included. (B and C) Heatmaps of average pairwise coalescence rates (apCRs) between and within groups of individuals. Color code for inter-group comparisons goes from light yellow (low apCR) to dark red (high apCR); for intra-goup apCR from light gray (low apCR) to black (high apCR). years ago, ya; Middle Eastern, ME; European, Eur; modern Ashkenazi Jews, AJ; Chapelfield, Ch.

Figure 5

Heatmap of simulation results for genetic disorders Comparison of test statistic (number of disease alleles with a read depth > 0) of observed Chapelfield data with 22,500 simulations under the assumption of modern Ashkenazi Jewish population allele frequencies, and modern European (non-Finnish) population allele frequencies, across a range of plausible read error rates. Methods underlying this analysis are described in Method details: Analysis of Ashkenazi-associated Mendelian disorders, and observed read depth per loci in the Chapelfield samples are shown in Figure S4.

Figure 6

Determining the probability of observed genetic disorder reads in the Chapelfield sample, given different population origins Log likelihoods compare the probability that the observed data were generated under a model of either the modern Ashkenazi population or the modern European (non-Finnish) population, using respective population allele frequencies in the gnomAD database. Each population model has its own parameter α, which determines the allele read error probability and is assumed constant across all loci. The maximum likelihood ratio indicates the data are 4,615 times more probable under the Ashkenazi model than the European model.