Equip DNA Info Lab · Revisió editorial · · 8 min de lectura
23andMe vs AncestryDNA: which raw data is better for analysis?
Which is better? For health and pharmacogenomics, 23andMe edges ahead (more drug-response and ClinVar coverage). For ancestry and relative-matching, AncestryDNA wins. For the GWAS research associations it is a tie. Analyse whichever you already have — the two files list different variants, and the report states per area how much of it your file covers.
If you've taken a consumer DNA test you can almost always download the raw data — a plain text file with hundreds of thousands of genetic markers. Once you have it, you can run third-party analyses, contribute to research, or just understand what your sample actually contains. But the file from 23andMe is not the same as the file from AncestryDNA. Here's how they differ and which is better for what.
The chips behind the files
Both companies genotype your sample on a microarray — a glass chip with millions of probes that detect specific positions in the genome. The chip determines which variants appear in your raw file. They are not the same chip:
- 23andMe currently genotypes on a customised Illumina Global Screening Array (GSA): 23andMe describes its v5 platform as around 654,000 standard variants plus some 50,000 of its own. The chip has been refreshed multiple times; newer kits include extra medically-relevant variants the older v3/v4 chips lacked.
- AncestryDNA uses its own customised Illumina chip, which reads over 700,000 positions. Its design biases toward ancestry-relevant markers (population-defining variants).
Overlap and unique coverage
The two current chips share far less than their sizes suggest. A 2020 comparison of 23andMe (v5) and AncestryDNA (v2) files found about 147,000 positions in common — 12.9% of all the positions in either file; earlier chip generations overlapped far more. The non-overlapping portion is where the practical difference lives:
- 23andMe covers more pharmacogenomic variants (drug-response markers) and many of the medically actionable variants flagged by ClinVar.
- AncestryDNA covers more ancestry-informative markers and certain population-specific variants useful for genealogical inference.
File format
Both ship a zipped .txt file. The internal format differs slightly:
- 23andMe: tab-separated, four columns (
rsid · chromosome · position · genotype). The genotype is two letters likeAGwith--for no-calls. - AncestryDNA: tab-separated, five columns (
rsid · chromosome · position · allele1 · allele2). Alleles are one letter each, with0for no-calls.
Both formats are trivially convertible. Any reasonable analysis tool — including ours — accepts either and auto-detects which is which.
Vols veure com és el fitxer per dins? Un exemple anotat de les quatre columnes →
Which is "better" depends on why you're analysing
| Use case | Better source |
|---|---|
| GWAS research associations (T2D, heart, etc.) | Tie — both have enough coverage |
| Pharmacogenomics (drug response) | 23andMe |
| ClinVar pathogenic variants | 23andMe |
| Ancestry / population inference | AncestryDNA |
| Matching with relatives | AncestryDNA (larger database) |
A note on quality
Microarray genotyping is not the same as whole-genome sequencing. At the positions they actually measure the chips are highly accurate: 23andMe reports over 99% accuracy and reproducibility for the variants in its FDA-authorised reports, and where both companies' chips measure the same position their calls agree 99.6% of the time. But a chip reads around 700,000 of the genome's roughly 3 billion nucleotides — about 0.02%. If you want a complete picture, consider a whole-genome sequencing service. For everything our report cross-references, however, microarray data is more than enough.
The practical answer
If you only have one and it's not a primary use case in the table above, analyse what you have. The reports will differ in which variants they list, because each file covers positions the other does not. If you're choosing a new test specifically to do downstream analysis, 23andMe edges ahead for health and AncestryDNA edges ahead for genealogy. The very best option, if budget allows, is to do both and merge the files — with so little overlap, merging adds most of the other chip's positions.
Has de triar una eina per al teu fitxer en brut? Compara eines d’anàlisi d’ADN en brut →
Frequently asked questions
Can I upload either file to the same tool?▼
Yes. The two formats are trivially convertible, and our parser auto-detects the four-column 23andMe layout and the five-column AncestryDNA one. The zipped download works as it comes.
Which file is better for drug-response analysis?▼
23andMe edges ahead: its chip covers more pharmacogenomic variants and more of the medically actionable variants flagged by ClinVar.
Which is better for ancestry?▼
AncestryDNA. Its chip design biases toward ancestry-informative markers and its relative-matching database is larger. Our report does not compute ancestry estimates either way.
Will both files produce the same report?▼
Not identical, and less similar than their sizes suggest: the current chips share only about 13% of their positions, so a variant can be present in one file and absent from the other.
Is microarray data as good as sequencing?▼
At the positions the chip actually measures, yes — 23andMe reports over 99% accuracy for the variants in its authorised reports, and the two companies' chips agree 99.6% of the time where they overlap. But a chip reads only about 0.02% of the genome; whole-genome sequencing is the complete picture.
Should I buy a new test just to analyse it?▼
If you already have a file, analyse what you have — the report states per area how much of it your file covers. If budget allows, doing both and merging the files adds most of the other chip's positions, since the two overlap little.
Already have a raw file?
Upload your 23andMe, AncestryDNA or MyHeritage file — free coverage summary and clinically-reviewed findings, full report 10 €. Our parser auto-detects the format and works with the zipped download.
Related articles
Aquest article té finalitats exclusivament educatives i no constitueix consell mèdic, diagnòstic ni tractament. Consulta sempre un professional sanitari per a decisions sobre la teva salut.