Dating admixture with DATES: when did two populations mix?

DATES estimates how many generations ago two ancestries mixed in a population, or in one genome. It measures how ancestry changes along the chromosomes: right after a mixture, ancestry comes in long blocks, and each generation of recombination cuts them shorter. The rate at which the signal decays with genetic distance gives the date.

DATES stands for Distribution of Ancestry Tracts of Evolutionary Signals. Narasimhan et al. 2019 introduced it, and Chintalapati, Patterson and Moorjani 2022 used it to date the main mixtures of the European Holocene.

How it works

1. You name two source populations and a target.
2. At each SNP, DATES weights the target's genotype by how different the two sources are there.
3. It computes how this weighted ancestry signal at one SNP covaries with the signal at SNPs further along the chromosome, by genetic distance.
4. The covariance decays with distance as an exponential curve. Its rate is the number of generations since the mixture.

The calculation needs a genetic map (distances in centimorgans), and the standard error comes from a jackknife over chromosomes.

One genome is enough

Older methods of this family, ROLLOFF and ALDER, need several genomes of the target population. DATES works on a single genome, including a pseudohaploid ancient genome. This makes it possible to date mixture in individuals from a site, not only in whole populations.

From generations to years

DATES returns generations. Multiply by a generation time, usually 28 or 29 years, to get years before the person lived. For an ancient genome, add the genome's own age:

years ago ≈ generations × 28 + age of the genome

For example, 20 generations in a genome dated to 4,500 years ago gives about 20 × 28 + 4,500 = 5,060 years ago.

Assumptions and limits

• One pulse. DATES fits a single mixture event. If ancestry arrived over many generations, the date falls between the start and the end, closer to the recent end.
• Good source proxies. The two sources must differ enough at many SNPs. Very similar sources give a weak signal.
• Old dates lose precision. A very old mixture leaves very short blocks, which are hard to measure, so its date has a large standard error.
• Mixed sources blur the signal. If a source proxy is itself a recent mixture, the curve can carry two decay rates, and a single date fits neither well.

Use DATES together with qpAdm: qpAdm tells you whether a target is a mixture of two sources, and DATES tells you when they mixed.

Running DATES

DATES is free software from Priya Moorjani's laboratory. Nexju runs DATES on the Max plan, on AADR populations or on your own genome, with the two sources you choose.

References

• Narasimhan V. M. et al. 2019. The formation of human populations in South and Central Asia. Science 365. doi:10.1126/science.aat7487
• Chintalapati M., Patterson N., Moorjani P. 2022. The spatiotemporal patterns of major human admixture events during the European Holocene. eLife 11. doi:10.7554/eLife.77625
• Loh P.-R. et al. 2013. Inferring admixture histories of human populations using linkage disequilibrium. Genetics 193. doi:10.1534/genetics.112.147330
• Moorjani P. et al. 2011. The history of African gene flow into Southern Europeans, Levantines, and Jews. PLoS Genetics 7. doi:10.1371/journal.pgen.1001373