v.66
AADR 1240K v.66
~1240K SNPs
Ancient: 19,119
Modern: 4,131
AdmixLab
Powered by qpAdm
Formal admixture modeling, your hand on the controls.
AdmixLab brings qpAdm — the workhorse of every modern population-genetics paper — into a wizard-driven web app. Pick a reference dataset, choose your targets and sources, fire off thousands of runs against the AADR v.62 and v.66 panels, and explore the fits without ever opening a terminal.
19,119
Ancient Samples
8,635
Modern Samples
~2M
SNPs / Panel
100/day
qpAdm Runs
qpAdm Analysis
A wizard for the wizard's tool.
qpAdm has been the gold standard for testing admixture hypotheses since 2015 — and the gold standard for being miserable to set up. AdmixLab wraps the entire workflow in a five-step wizard so you can model targets in the time it used to take to wrangle a parameter file.
- 5-step guided wizardDataset → Targets → Sources → References → Results. Each step validates the previous so you never run a broken model.
- Model your commercial filesUpload your own raw data files and use them with any datasets. Our non-destructive merging pipelines ensure seamless integration.
- AADR v.66 + v.62 datasets out of the box1240K, Human Origins, and 2M SNP panels — including compatibility variants that level the bias across sequencing methods.
- Everything is identical to the originalDownload full raw outputs for all the statistical details, observe SNP coverges and more through a dynamic terminal.
- Smart presets for sources and referencesSave your favourite right-set + source combinations once, re-use them across every target run. Combine presets to build proximal / distal model families in seconds.
- 100 runs per day, with no queueEach run sits on dedicated compute — kick off a sweep, grab a coffee, and your model fits are waiting.
Genetic Distances
Fst, the way you wish it had always worked.
Compute pairwise Fst between any target and any pool of source populations on the AADR panels. Four steps, no parameter files, ranked results in minutes — same robust statistic the published papers use.
- 4-step wizard, ranked resultsDataset → Targets → Sources → Results. Distances come back sorted closest-first, with the per-population sample sizes and SNP overlap shown inline.
- Multi-target sweeps in one passDrop in a handful of targets and AdmixLab computes their distances against the same source pool.
Reference Datasets
Seven reference panels across two AADR builds.
Every analysis sits on the Allen Ancient DNA Resource. Five v.66 panels — including two Compatibility variants that reduce bias from different sequencing methods — plus two v.62 panels for reproducing older papers.
v.66
AADR Human Origins v.66
~600K SNPs
Ancient: 19,119
Modern: 8,635
v.66
AADR 2M v.66
~2M SNPs
Ancient: 19,119
Modern: 4,131
Recommended
AADR 2M v.66 (Compatibility)
~2M SNPs
Ancient: 19,119
Modern: 4,131
Recommended
AADR Human Origins v.66 (Compatibility)
~600K SNPs
Ancient: 19,119
Modern: 8,635
v.62
AADR 1240K v.62
~1240K SNPs
Ancient: 13,571
Modern: 3,953
v.62
Human Origins v.62
~600K SNPs
Ancient: 13,571
Modern: 7,978
There’s More Inside
Everything else that ships with AdmixLab.
Essential features are built-in for your convenience.
Saved Analyses
Every qpAdm and Fst run can be saved with one click, complete with its full parameter trail for feature reference.
Documentation Ready
Step-by-step guides for proximal / distal modelling, best practices, and the most common pitfalls explained in details to get you started asap.
Common Questions
Q&A
Quick answers to the questions we hear most often about AdmixLab — what it does, what it doesn’t, and what to expect.
What is AdmixLab?
AdmixLab is a browser-based environment for qpAdm (and FST), a popular software widely used in scientific studies for estimating admixture / ancestry proportions of populations.
Will I receive any results after purchasing this service?
AdmixLab is a DIY (Do It Yourself) service — there are no pre-generated results; you'll need to build your own models. We've made the process straightforward with easy-to-follow steps, a user-friendly interface, and detailed documentation on how to use the tool and interpret your results.
However, a basic understanding of population genetics and ancient-DNA literature is still required. At a minimum, you should know which samples to select in order to construct a meaningful model. Because of the nature of the tool, we are unable to provide personalised assistance in designing models tailored to your specific ethno-geographic background.
You should also keep in mind that qpAdm is a relatively hard tool to use efficiently and there is no guarantee that you will achieve your desired results.
Which companies' files are accepted?
You can use the raw-data files from MyHeritage, AncestryDNA, FamilyTreeDNA, 23andMe, Living DNA, or TellMeGen.
The commercial file you want to use has to be an autosomal raw-DNA data file in genotype format. WGS files (VCF, BAM, etc.) cannot be used unless they are first converted into a genotype format.
Which company's file gives the most coverage?
qpAdm operates on the SNPs shared between the samples you select when creating a model. We use the AADR dataset as a base, and the number of SNPs shared between this dataset and commercial raw-DNA files will vary depending on both the company and the chip version used to sequence your DNA.
As a result, there is no definitive answer to how many SNPs will be shared — this can only be determined through extensive testing. However, you can generally expect between 100,000 and 400,000 shared SNPs in total. (This number will be proportionally reduced depending on the specific samples you choose when creating a model, as not all samples in the dataset contain the same number of SNPs.)
Despite this variation, raw-DNA files from major companies typically provide enough coverage to achieve optimal results, especially given qpAdm's robustness with low-coverage data.
How long does it take for an analysis to finish?
A single analysis — either an FST calculation or a qpAdm model — takes around 5 to 10 minutes, mostly depending on the target type (using your own raw-data files as targets increases the time compared to using dataset samples, because of the merging process).
The length of the duration is set by the fact that every analysis in AdmixLab is computed directly from genotype files each time, rather than using pre-processed data like PCA coordinates.
What happens when my subscription ends?
You can still see your saved analyses and download their detailed raw outputs, but you can't create new ones.
Disclaimer
This is a web browser based environment service for ADMIXTOOLS 2 software in compliance to GPL-3 licensing. No modifications are done to the original software.
Ready to put your hand on the controls?
Subscribe once and unlock qpAdm, Fst, saved analyses and every dataset above — including future AADR releases.
Start Now