Methodology deep-dives, population spotlights, and engineering notes from the Ancestrify team, qpAdm modelling, ancient DNA panels, and how genetic ancestry analysis really works.
31 stories
Looking for a DNAGENICS alternative? The reasons people look (a G25-derived row instead of the official one, pack pricing, no formal test) and which services answer each, with every DNAGENICS fact read on dnagenics.com on 2026-09-06.
Have Global25 coordinates? Contribute your row to the public modern G25 dataset and get a free report at once: admixture per era, closest populations and notable matches. No payment.
Looking for a MyTrueAncestry alternative? What people actually want from one (individual ancient matches, published statistics, one-time pricing) and which services deliver each, honestly compared.
ADMIXTURE describes structure with K inferred components; qpAdm tests an explicit historical model and can reject it. What each estimates, why papers use both, and which answers which of your questions.
A Global25 model and a testing company's ethnicity estimate answer different questions with different references and different time depths. What each is actually computing, and which to trust for which claim.
Every admixture calculator — GEDmatch's classics, Global25 fits, testing-company estimates — is an optimiser that cannot say no. How the three families work, what the percentages mean, and the questions a calculator can and cannot answer.
Which GEDmatch admixture project to run for your background, what each calculator's components mean, why the projects disagree with each other, and what has aged since 2012–2016.
There is no single best admixture calculator — there is a best one per question. An honest decision guide across GEDmatch's classics, Global25 tools and formal methods, from someone who builds one of them.
Accuracy has three different meanings for an ancestry calculator, and the tools do well on exactly one of them. Where percentages are trustworthy, where they are noise, and how to tell which regime you are in.
The same genome scores 44% North European in one calculator and 51% in another. Component anchoring, reference panels, the calculator effect and K explain the spread — and none of the numbers is 'the real one'.
Oracle ranks reference populations by how closely their calculator percentages match yours — not by shared DNA. How the distance is computed, what single and mixed modes tell you, and the misreadings to avoid.
The thirteen K13 components, what North Atlantic, East Med and West Asian are anchored to, who the calculator works best for, how to read the Oracle — and the caveats that come with a 2012-era tool.
What the 36-component Eurogenes calculator can genuinely show, why its regional labels invite over-reading, what a K36 heat map is actually plotting, and when a distance ranking answers the question better.
The original admixture project's most-used calculator: what the twelve K12b components are anchored to, what Gedrosia and Caucasus mean in modern ancient-DNA terms, and how to read a K12b breakdown.
What HarappaWorld's S-Indian, Baloch and NE-Euro components are anchored to, why it remains the reference calculator for South Asian ancestry, and how its 2012 categories map onto what ancient DNA later proved.
The program behind the bar plots in population-genetics papers: how ADMIXTURE estimates K components jointly from all samples, supervised versus unsupervised runs, choosing K, and why hobbyist calculators are its frozen shadows.
A calculator against modern populations answers 'who do I resemble today'; one against dated ancient samples answers 'which deep ancestries formed me'. Mixing up the two produces the classic misreadings.
GEDmatch's admixture calculators froze around 2012–2016. The free alternatives that run on current ancient reference panels — era-scoped calculators, distance rankings, PCA — and what each replaces.
A step-by-step qpAdm tutorial: check your raw file, merge it into AADR v66, choose sources and outgroups, run it in the browser or in R, and read the result.
What an admixture calculator optimises, why it always answers, what a qpAdm model adds, when each is the right tool, and when qpAdm is not worth paying for.
What consumer ancestry percentages measure for Jewish customers, why an 'Ashkenazi 99%' result is circular, what a formal qpAdm model shows instead, what the four tiers buy, and the honest limits: endogamy, drift and coverage.
What the full qpAdm model record in an Ancestrify report means — chi-square, degrees of freedom, f4 rank, SNP counts, jackknife blocks, 95% confidence intervals, the nested-model table and the rank test — and how to read the plain-text download.
What a qpAdm ancestry test actually does with your raw DNA file, what the report contains, what the four tiers buy and what they don't, and how to tell a formal model from a percentage generator.
A plain reading guide to the three numbers in every qpAdm result — what the p-value tests, what a standard error bounds, what a Z-score rules out — with worked examples and the mistakes that make a passing model wrong.
A step-by-step Vahaduo tutorial: where to paste your Global25 coordinates, how the Distance, Single, Multi and PCA tabs work, how to build a source panel, and how to read a fit distance without over-reading it.
How to run qpAdm on your own genome in the Ancestry Workbench: preparing a kit, choosing sources and outgroups, reading a rejection, running a rotation as a batch of real fits, scanning Fst distance, and downloading the EIGENSTRAT bundle to reproduce everything on your own machine.
Why a standard Global25 calculator can describe a population well and one person badly, how an analyst hand-builds a source panel around your own coordinates, and what changes in the report when a personalised calculator is published beside the standard one.
Global25 fits your coordinate to a mixture and always returns percentages. qpAdm tests a model against allele-frequency statistics and can reject it. A practical comparison of when each one is the right instrument.
What the 25 numbers mean, why scaled and unscaled forms must never be mixed, how a closest-population ranking is computed, what a fit distance does and does not tell you, and the failure modes that make confident results wrong.
Ten genomes from a Himalayan cave reveal a population that was half Tibetan-related and half North Indian-related, mixing from about 2800 years ago.
The method, for readers who want the maths: what qpAdm computes, what the p-value, standard error and z-score each mean, why outgroup choice decides whether a model is worth anything, and how to read a rejection — with a worked example.