What you get

The download

One PDF report, typically 15–20 pages

The full analysis: the fitted model, every competing model, the reference sources, and how to read each number.

5 result cards, as images

Shareable one-screen summaries of the headline result, drawn from the same numbers as the report.

Inside the report

The tested sources

The ancient reference populations your genome was fitted against — for example a Steppe pastoralist source, an Indus Valley farmer source and an ancient South Asian hunter-gatherer source — named as populations, not as brands.

A weight and a standard error for every source

Each weight is the share of your autosomal ancestry the model assigns to that source, with one standard error beside it. The error is part of the result, not a footnote.

The model p-value

The probability the data would look this different from the model by chance. At least 0.05 the model is not rejected — which is not the same as proven.

The model grid, one row per model

Every combination of farmer and steppe sources that was tried, with its number of sources, its weights and its p-value, so you can see what else fits and what does not.

The 12 outgroups

The reference populations every model is tested against — ancient and modern genomes far from the sources — listed so the fit can be checked, not taken on trust.

In the grid a model is marked as passing when its p-value is at least 0.05 and every weight lies inside 0–100%. Whether your file receives a report at all is decided only by the second test — weight feasibility — never by the p-value: p measures the power to reject, and it rises as a file's coverage falls, so a p-threshold on its own gets more permissive exactly as an upload gets worse.

What it does and does not tell you

It tells you

• Which published ancient populations, in which proportions, describe your autosomal ancestry — with the uncertainty on each proportion stated.
• Whether that model is feasible, and which competing models also fit.
• Where the reference set runs out: the current sources describe North-West South Asian ancestry best, and the report says so when a fit is weak.

It does not tell you

• It is not a medical, diagnostic, paternity, forensic or legal test.
• It establishes no nationality, ethnicity, community membership, caste, religion or entitlement, and it is not a certificate of origin.
• A passing p-value means the model is not rejected. It does not mean the model is true, and a different reference set can prefer a different model.
• It does not name relatives, health traits or a single "true" percentage. Population history is a decomposition with error bars, and the report leaves the error bars in.

Two outcomes are possible, and which one you get decides the refund. If your file cannot be processed — it is not a usable autosomal genotype — no analysis runs and you are refunded in full. If the file processes and the fitted weights fall outside 0–100% — no combination of the reference sources describes this person — that is a real analytical finding: it is delivered as your report and is not refunded. Details on the refunds page.

A worked example

The headline result, in the layout the report uses. These are illustrative examples, not real customers; the format is what you are looking at.

View all three illustrative examples

Turnaround

Your report arrives by email as a private link within 24–48 hours of payment. Hours rather than minutes, because the model is fitted and checked against the reference set rather than looked up in a table.

Your data

Your file is uploaded over an encrypted connection, stored encrypted, used only to produce your report, and never sold or shared with advertisers or data brokers. How long it is kept is set out in the terms. You can ask for your copy to be deleted at any time by writing to support [at] indoaryan [dot] com .

Read the privacy policy