Articles
72 articles on guide, ancient-DNA research written up for readers who want the evidence, not just the headline.
Step-by-step guides to the practical half of ancestry analysis: downloading raw data from each testing company, obtaining Global25 coordinates, reading a qpAdm report, running the free tools, and knowing which method answers which question. Every guide states costs and waiting times honestly, including for the steps we have no hand in.
Every service that sells or offers a Global25 (G25) analysis in 2026, side by side: what each computes, whether the coordinates are official, whether the source panels are published, whether anyone builds a calculator around your own row, and the price model as each site states it.
Every place that sells, obtains or simulates Global25 (G25) coordinates in 2026, side by side: the official portal, the done-for-you route, derived rows and free simulations, with prices as each site states them.
A structured learning path through everything qpAdm — what to read in what order, from the f4-statistics underneath to running models in R, choosing outgroups, reading results and knowing the method's measured limits.
The definitive working checklist for qpAdm in 2026 — temporal stratification, right-set construction, lowest-rank-first search, composite feasibility and reporting standards — each rule carrying its measured justification from the 2021–2025 audit literature.
The Middle East is where qpAdm's sources crowd closest together: Natufian, Anatolian, Iranian and Caucasus ancestries all interrelated. The working recipe, the right set that splits them, failure modes and a worked reading.
South Asia is qpAdm's hardest standard fixture: one ancestral stream has no ancient sample at all. The working recipe — Indus Periphery, steppe MLBA, the Onge-as-AASI-proxy problem — with right sets, failure modes and a worked reading.
The three-source model that rebuilt European prehistory — WHG, Anatolian farmers, steppe pastoralists — as a working qpAdm recipe: exact source and right-set choices, regional adjustments, failure modes and a worked reading.
Same model, different p-value: the real differences between original qpAdm and ADMIXTOOLS 2 — allsnps semantics, fudge_twice, f2 precomputation — and the settings that reproduce classic behaviour when you need to.
The suffixes on AADR population labels encode how each genome was produced — capture, shotgun, diploid, array — and mixing them carelessly biases f-statistics. What each label means and the selection rules that keep models honest.
The settings tutorials skip and reviewers ask about: extract_f2 versus genotype input, allsnps, maxmiss, fudge and fudge_twice, blgsize, boot, afprod and constrained — what each does, with the measured stakes.
The undocumented half of every qpAdm analysis: file formats, genome builds, strand hygiene, convertf and Poseidon, and the merge arithmetic that decides your standard errors before any model runs.
Looking for an IllustrativeDNA alternative? The real decision points — coordinate reports versus hand-checked qpAdm, subscription versus one-time, WGS input, published statistics — and which services deliver each.
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.
The complete Global25 price list: what official coordinates cost, what is genuinely free, what a worked analysis and its add-ons cost, and the traps that make 'free G25' expensive.
From genotype files to a tested model: f2 extraction, qpadm() and its output tables, the arguments that silently change results (allsnps, fudge_twice, constrained), and the protocol discipline the code will not enforce for you.
Negative weights, SE 9.99, every model rejected, every model passing, infeasible popdrop rows, allsnps confusion — the standard failure gallery of qpAdm runs and what each symptom actually indicates.
The canonical nine outgroups from Lazaridis 2016, the o9a and o9aamcn extensions, the arithmetic floor and the ~30-population ceiling, and the published example where one added outgroup cut standard errors threefold.
The fit distance measures how far a mixture still sits from your coordinate — not whether the model is right. Calibrated bands for reading one, why lower stops being better, and the checks that actually validate a model.
Every Global25 row exists in two forms, and mixing them is the hobby's most common silent error. What scaling does mathematically, which form each tool expects, and the tell-tale signs of a mixed comparison.
Global25 is a projection, not a test — so 'accuracy' has layers: the coordinate's own fidelity, the references it is compared against, and the models built on top. An honest audit of all three.
What a Global25 PCA projection shows, what the axes are and are not, why plot distance lies, the difference between projecting onto a fixed view and fitting your own — and the honest uses of both.
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.
Everything you can do with a Global25 row without paying anyone: distance rankings, admixture fits, PCA plots, averaging, authenticity checks — Vahaduo, nMonte and the Ancestrify Lab, honestly compared.
Dante Labs, Nebula, Sequencing.com and other WGS files can become official Global25 coordinates — via the portal's own conversion surcharge or a free WGSExtract conversion. What each route costs and where the traps are.
What a Global25 distance ranking actually measures, why era choice changes everything, how close is 'close', and the four misreadings that turn a good tool into a wrong conclusion.
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.
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.
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.
Every place that sells or offers qpAdm in 2026: hand-checked analyses, DIY tools and subscriptions, with inputs, reference data and price models side by side.
What a qpAdm analysis costs in 2026: four one-time tiers from 29.99 EUR, the add-ons, what does not change the price, and what a DIY subscription costs instead.
What Davidski's Eurogenes G25 Requests portal is, what you upload, the scaled and unscaled rows you get back, their stated fee, and what to do with the row next.
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.
How to choose qpAdm left and right populations: the classic outgroups, the drift rule, the rank test, temporal logic, proxy labels, rotation and a worked example.
What a qpAdm p-value below 0.05 does and does not mean, the five usual causes of a rejection, negative weights, low Z-scores, what to try next and when to stop.
An illustrative qpAdm report for one era, read from the model line through the weights, p-value, right set, nested models and rank test to the explanation.
How to download your 23andMe raw data, what the v3, v4 and v5 chips mean for coverage after the AADR merge, and what a qpAdm order does with the file.
How to download your AncestryDNA raw file, what the v1 and v2 chips mean for coverage after the AADR merge, the allele quirk, and what a qpAdm order does.
How to download your MyHeritage raw CSV, what its chip means for coverage after the AADR merge, the low-pass WGS note, and what a qpAdm order does.
What ancient genomes actually say about Jewish ancestry, why a consumer 'Ashkenazi Jewish' percentage answers a different question, and what a formal qpAdm model with a p-value shows for a Jewish genome.
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.
Global25 coordinates explained from scratch: what the 25 numbers are, where they come from, scaled versus unscaled, what you can compute from them, and the honest limits of a coordinate-based ancestry analysis.
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.
What a 23andMe or AncestryDNA ethnicity estimate measures, what an ancient-DNA ancestry test measures instead, why the two disagree by design, and how to run the second one on the raw file you already have.
Who the Western Steppe Herders were, how steppe ancestry spread across Europe and Asia after 3000 BC, what a 'steppe percentage' actually measures, and how to estimate yours with qpAdm or Global25 from a raw DNA file.
Who Europe's Mesolithic hunter-gatherers were — Western, Eastern and Caucasus — how their ancestry survived farming and the steppe migrations, and how to measure your hunter-gatherer share from a raw DNA file.
Who the Anatolian Neolithic farmers were, how their ancestry spread across Europe from 6500 BC and became the largest component in most southern Europeans, what 'early European farmer' means in a model, and how to measure your share.
How to run an ancient-DNA ancestry analysis from a whole-genome sequencing VCF — tellmeGen, Dante Labs, Nebula and similar — what happens to the file at upload, why it is converted to the panel's markers, and what changes versus a chip export.
What a maternal (mtDNA) haplogroup is, how the tree is built, what H, U5, K, T2 and the other common branches tell you, why most chip kits resolve to a broad branch, and what a haplogroup can and cannot say about ancestry.
How to run qpAdm on your own genome after a published report — choosing sources and outgroups, reading a rejection, testing nested models, 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.
What the Allen Ancient DNA Resource is, who curates it, what a version like v66 contains, the difference between the 1240K and Human Origins panels, and how the dataset becomes the reference behind an ancient-DNA ancestry analysis.
What it means to share a stretch of DNA with a person who died thousands of years ago, why such matches are identity by state rather than identity by descent, what the centimorgan figures mean, and why a shared segment is never proof of descent.
What a Y-DNA haplogroup actually is, how to export the raw file from 23andMe, AncestryDNA or MyHeritage, how to get a free and honest paternal haplogroup call from it, and where to explore your branch afterwards.
What mitochondrial DNA records, how to get a free maternal haplogroup call from a 23andMe, AncestryDNA or MyHeritage export, why most chip kits resolve to a broad branch, and how to explore the maternal tree afterwards.
23andMe does not produce G25 coordinates — but its raw data file is a good starting point. How to download it, what the v5 chip actually covers, and the exact route from file to coordinate row.
AncestryDNA does not produce G25 coordinates — but its raw data download is the input that becomes one. The emailed-confirmation download flow, what the file covers, and the route from file to coordinate row.
MyHeritage does not produce G25 coordinates — but its raw data export is the input that becomes one. The download flow, the low-pass WGS caveat that costs people their haplogroups, and the route to a coordinate row.
FamilyTreeDNA does not produce G25 coordinates — but a Family Finder raw data export is the input that becomes one. Which download to pick, what the file covers, and the route to a coordinate row.
LivingDNA does not produce G25 coordinates — but its raw data download is the input that becomes one. Where the download lives, what the file covers, and the route to a coordinate row.
Where Global25 coordinates come from, how to get them whether you tested with 23andMe, AncestryDNA, MyHeritage, FamilyTreeDNA or LivingDNA, and how to tell a real coordinate row from a simulated one.