Global25 (G25) ancestry analysis

This analysis starts from a Global25 coordinate row you already have — paste it at checkout and the report is computed automatically, for €29.99. Coordinates come from the independent Eurogenes Global25 service; raw DNA files are not accepted for this product.

This is our Global25 product, which is a different method from qpAdm — and qpAdm is the one that takes a raw DNA export. It merges your export against the Allen Ancient DNA Resource and returns a formally tested ancestry model with p-values and standard errors, which Global25 does not produce.

Price

€29.99, one-time. No subscription.

Input

Your own Global25 coordinate row. Raw DNA files are not accepted for this product

Reference set

30,386 individual samples across 1,535 curated populations

Eras

Six — Late Bronze Age (3000–1200 BC), Pre-Classical Iron Age (1200–0 BC), Imperial Antiquity (0–600 AD), Middle Ages (600–1400 AD), Early Modern (1400–2000 AD), Modern Era (2000 AD on)

Distances

Euclidean across all 25 dimensions; the 25 closest populations per era

Admixture

nMonte-style Monte-Carlo mixture modelling, reported with a fit distance

PCA

Individual samples — not population averages — across 12 regional groupings

Also included

Server-rendered cinematic videos for each of the three analyses

Data location

European Union (Germany and Finland), under GDPR

Global25 places every sample as 25 numbers in a shared space, which makes it fast and enormously flexible: distances, mixture models and projections all become arithmetic on coordinates. That is why it is the workhorse of amateur population genetics.

It is worth being clear about its limit. A mixture solver returns the combination of sources that lands closest to your point, and some combination always lands closest — the method has no way to say that a model is incompatible with the data. That is what fit distance is for, and it is why we show it rather than presenting the percentages alone. If you want a result that can be statistically rejected, that is qpAdm.

Distance is a single operation repeated across a panel. For each reference population, the difference from your coordinate is taken in each of the 25 dimensions, squared, summed and square-rooted — plain Euclidean distance — and the list is sorted closest first. It is deterministic: the same row against the same era always produces the same ranking. Because a reference population is an average of individuals, and individuals scatter widely around their group's centre, a close distance is evidence about populations rather than about any person who lived.

Admixture asks a different question: which weighted mixture of source populations lands nearest your point? The search is a Monte-Carlo fit in the nMonte tradition, and the leftover gap between the mixture and your coordinate is reported as the fit distance. Each era's model is fitted against a curated source panel, and the report lists every population that panel offered — including the ones the model was offered and did not use. Those rejected sources appear nowhere else, and showing them is what stops a breakdown reading as the only possible answer.

PCA projects the coordinate space down to a readable scatter and places you among individual samples, not population averages — which is the distinction that makes the plot informative, because it shows the spread real people occupy rather than a tidy row of group centroids. The axes are not ancestries, and closeness on a two-dimensional plot can hide separation in the dimensions the projection discarded.

Three habits keep a coordinate report useful. First, read groupings rather than individual lines: sources that sit close together in the reference space trade weight almost freely, so the total two related sources share is far more stable than their split. Second, treat rank order as orientation, not significance — populations separated by a thousandth in distance are, for practical purposes, tied. Third, compare across eras rather than fixing on one: your position relative to a Bronze Age panel and a modern one are different measurements, and the movement between them is the interesting part.

The report also states its own scope on every era — the calculator used, your declared ethnicity, the regions included and the size of the source panel — because a percentage means nothing without the panel it was fitted against. Sources under 3% that the chart folds into “Other” appear in that provenance list with their real values.

Most of the tooling is public and free. The G25 distance calculator ranks your closest populations across six eras, the admixture calculator models mixtures against curated or pasted panels, Average G25 builds a population average from individual rows, and the authenticity check flags simulated, edited or precision-lost coordinate rows. All four run in your browser — nothing is uploaded — and none of them needs an account.

If you are new to the method, Understanding Global25 explains what the numbers mean, how to get your coordinates covers obtaining a row from any testing company, and qpAdm vs Global25 sets out when a coordinate fit is the wrong instrument. Terms are defined in the glossary.

• 01

  Do you generate Global25 coordinates from my raw DNA?

  No. A G25 order takes the official Global25 coordinate row you already have — coordinates come from the independent Eurogenes Global25 service, and you paste that row at checkout. If you only have a raw DNA export, our qpAdm analysis is the product that takes one.

• 02

  Can I see which populations the model could have used?

  Yes. Each admixture era lists every population in the calculator it was fitted against, split into two labelled groups: the ones the model used, and the ones it was offered and did not use. The rejected populations appear nowhere else in the product, and showing them is what stops a breakdown looking like the only possible answer. The report also states the scope — the era, your declared ethnicity, the regions, and the size of the source panel — and a source below 3% that the chart folds into Other appears there with its own real percentage.

• 03

  What do I get for my coordinates?

  Three analyses across six eras: distances to the closest reference populations, an admixture model of your coordinate as a mixture of source populations, and a principal-component scatter placing you among individual samples. Each comes with a downloadable cinematic video.

• 04

  How large is the reference set?

  30,386 individual samples and 1,535 curated populations. Distances are computed across six eras — Late Bronze Age, Pre-Classical Iron Age, Imperial Antiquity, Middle Ages, Early Modern and Modern Era — and the 25 closest populations are returned for each.

• 05

  Does the PCA plot populations or individuals?

  Individuals. The scatter places you among individual ancient and modern samples rather than among population averages, across 12 regional groupings, so you can see how much internal spread a population actually has instead of a single point standing in for thousands of people.

• 06

  How are the distances calculated?

  Euclidean distance across all 25 dimensions of the Global25 coordinate space, against every reference population in the selected era, ranked closest first.

• 07

  How is admixture modelled?

  With a Monte-Carlo solver in the nMonte tradition: your coordinate is fitted as a weighted mixture of curated source populations, small components are pruned, and the model is re-solved. Every result is reported with its fit distance so you can see how well the mixture actually reproduces your coordinate.

• 08

  Are these percentages the same thing as qpAdm proportions?

  No, and the difference is worth understanding. A Global25 admixture percentage is a coordinate-fitting estimate: the solver returns the mixture that lands closest to your point, and some mixture always exists. qpAdm works from allele-frequency statistics and reports a p-value, so it can reject a model as incompatible with the data. Global25 is faster and broader; qpAdm is the one that carries statistical confidence.

• 09

  How long does it take?

  Results are computed automatically — there is a holding period by default, which the optional fast-compute add-on removes so your report is built as soon as you have paid. We email you as soon as it is ready.

• 10

  Where is my data stored?

  On EU infrastructure, in Germany and Finland, under GDPR. You can delete your data at any time, and full account deletion and data export are self-service.