Free G25 tools in 2026: the complete toolbox for your Global25 coordinates

A Global25 row is deliberately portable — plain text, 25 numbers — and an entire free ecosystem grew around the format. This is the current toolbox: what each free tool does, which one answers which question, and the few places where paying for something changes the answer rather than the packaging.

One honest note up front: the row itself is the one thing that is not free — official coordinates come from the independent Eurogenes service at €15 per kit (how to get them). Everything below assumes you have one.

Distances: "who am I closest to?"#

The assumption-free starting point — what the ranking means and how close is close.

• Ancestrify G25 distance — paste one row, get ranked distances against 1,535 curated populations across six dated eras, in the browser, no account. The era scoping is the point: "closest today" and "closest in the Iron Age" are different questions answered separately.
• VahaduoJS distance mode — the community stalwart (our tutorial). Bring-your-own reference sheets: maximum flexibility, no curation, you manage the datasheets and their scaled/unscaled discipline yourself.

Admixture: "what mixture explains me?"#

• Ancestrify G25 admixture — nMonte-style Monte-Carlo fitting against curated, era-scoped calculators (or sources you pick), with the fit distance stated. Browser-only; pasted rows are not uploaded.
• Vahaduo multi mode — the same family of arithmetic, bring-your-own sources. The freedom is real and so is the responsibility: source selection decides the answer, and nothing in any free tool warns you when a panel is wrong.
• nMonte in R — the original script the whole tradition descends from, for people who want the dials (penalty terms, batch runs) and reproducible scripts.

Seeing the space: PCA#

• Ancestrify G25 PCA viewer — project your row onto curated era views beside published ancient and modern samples, or fit a fresh PCA over rows you paste. How to read one without fooling yourself.
• Vahaduo Global25 Views — the classic gallery of preset plots.

Utilities nobody advertises and everybody needs#

• Average G25 — build a population average from individual rows (that is what every reference "population" is).
• G25 authenticity check — reads a row's numeric fingerprint; catches simulated and doctored rows before you build on them.
• Raw file check — pre-flight for the file you would send for coordinates: format, chip generation, usable markers per chromosome.

Tool by tool: what each one answers#

The list above is organised by question. This section walks the same tools in the order most people use them, with what each one computes and the limit each one states.

/lab/admixture: "what mixture of sources explains my row?"#

The Global25 admixture calculator treats your row as a point in the 25-dimensional space and searches for the weighted mixture of reference populations whose combined point sits closest to it, a Monte-Carlo fit in the nMonte tradition. You pick a curated, era-scoped calculator or paste your own sources. The tool ranks your closest single populations by plain Euclidean distance first, so you can see which sources are near you before asking for a mixture of them; it prunes components under 1.5% and re-solves rather than reporting decoration; and it states the leftover gap as a fit distance. On a curated panel of three to eight sources, a scaled fit at or under about 0.020 is excellent, 0.020 to 0.035 is the ordinary range for real individuals, and past roughly 0.045 a relevant source is probably missing. What it cannot do: reject a model. It always returns percentages, and the fit distance is the only hint that a panel is wrong.

/lab/g25-distance: "who am I closest to, in which era?"#

The G25 distance calculator is one arithmetic operation repeated across a panel: Euclidean distance over all 25 dimensions, sorted closest first, against curated reference populations in any of six eras from the Late Bronze Age to today. It is the assumption-free starting point because nothing is fitted. The scale to read it by: a typical individual sits around 0.011 from their own country's average, 0.02 to 0.04 reads as same broad region, and beyond about 0.05 to 0.06 you are outside the coordinate's neighbourhood. Two diagnostics come free with it: if the nearest population on Earth sits past roughly 0.08, suspect an unscaled or corrupted paste; and a top cluster of near-tied populations is the finding, not the exact winner. Closeness is resemblance, not descent, and no ranking can test whether a population belongs in your ancestry.

/lab/g25-pca: "where do I fall among the samples?"#

The G25 PCA viewer works in two modes. An era view projects your row onto a reference space computed once over that era's published samples, so your point lands among ancient and modern individuals without moving any of them and is comparable across sessions. A custom view fits a fresh PCA in your browser over exactly the rows you paste, which is right for comparing a handful of coordinates against each other and wrong for anything else, because adding or removing one row can rotate every axis. The limit shared by both modes: a chart shows 2 of 25 dimensions, chosen for spread, not meaning. Any time two points look close, measure them with the distance tool.

/lab/average-g25: "what is the centre of these rows?"#

Average G25 takes the arithmetic mean of each of the 25 dimensions independently across the rows you paste and returns a copy-ready row. That is how every reference population is built from its member individuals, and it is why a four-person site average makes a steadier distance reference than any single member: an average of N members carries roughly 1/√N of an individual's coordinate noise. The tool does the spreadsheet job with the spreadsheet failure modes removed: full precision, every dimension, nothing uploaded. What it will never do is decide membership. Average two genuinely distinct groups and you get a coordinate for a population that never existed, and it will still return confident distances to everything.

/lab/g25-authenticity: "is this row real?"#

The G25 authenticity check inspects a pasted row for the numeric signature of how it was produced: precision that has been lost, values outside the range a real projection produces, digit patterns consistent with hand editing, and the fingerprints of a simulated rather than measured coordinate. The three rows people actually paste are the forwarded row of forgotten provenance, the row converted from another calculator's output, and their own row checked before building on it, which is the cheapest habit in the hobby. It audits the numbers, not the provenance: it cannot prove a row is yours or certify which service produced it, and it says "looks" rather than "is" on purpose.

/lab/mapper: "how do I turn the numbers into a figure?"#

The Mapper is a figure composer, not an analysis. You place pie and donut charts, keys, labels and images onto a shaded-relief basemap, anchored to the map rather than the screen so a chart placed over a region stays over it when the view moves or the export renders at another size, and export the result as a high-resolution image or an animated MP4. Compositions are saved locally in your browser. Its honest caveat is the one every map figure carries: a chart on a country is a claim about where a population was, and the map lends that claim more precision than the data has. Label what each chart represents and for which period, and place charts near the sites the samples came from.

Ancestrify Lab and Vahaduo, side by side#

Vahaduo is the community's standard and it earns the place; our tutorial walks it. The two toolkits run the same arithmetic with opposite philosophies about sources.

Neither is the "real" tool with the other a demo. Vahaduo is the right choice when you want to test a hypothesis with sources of your own choosing; the Lab is the right choice when you want a defensible era panel without building one, and want the input row checked first. Many people sensibly use both, and the cell-by-cell comparison goes deeper.

What the free stack cannot do#

Three limits are structural, not feature gaps. No free coordinate tool can reject a model — fits always return percentages, and only the fit distance hints when something is off (and it under-hints). None provides curation with a rationale — the difference between a source panel and a defensible source panel is analyst work, which is what the paid Global25 analysis and its personalized calculator add on top of the same arithmetic. And none leaves coordinate space — claims that need allele-frequency-level testing belong to qpAdm, a different method entirely.

For everything else — which is most of the hobby — the free stack above is not a demo of the real thing. It is the real thing, and the sensible order is: distances first, one era-scoped fit second, PCA alongside, and money spent only where a question has outgrown the tools.

Frequently asked questions#

Are Ancestrify's Global25 tools really free?#

Yes. The in-browser tools listed here need no account and no payment; the coordinates you paste are not uploaded and nothing is stored. The only thing that is not free is the row itself, which comes from the independent Eurogenes service.

Which tool should I use first?#

Distances. The distance calculator has no assumptions, so it tells you the neighbourhood any model must explain. Then one era-scoped fit in the admixture calculator, with the PCA viewer alongside, and the authenticity check on the row before any of it if its provenance is uncertain.

Do I need Vahaduo if I use the Lab?#

Not for the standard questions, and not the other way round either. The arithmetic is the same. Use Vahaduo when the source panel you want to test does not exist as a curated calculator; use the Lab when you want the panel curated for you and the row audited first.

Terms used here are defined in the glossary.