Raw data file formats: four providers compared
All four providers give you a marker file — but they look different. Here are the differences that matter when you analyze them.
Format overview
| Provider | File you get | Internal format | Approx. markers | Direct into DNAHelper? |
|---|---|---|---|---|
| MyHeritage | .zip | Comma-separated .csv | ~700,000 | Yes — the zip goes straight in |
| 23andMe | .zip (genome_…) | Tab-separated .txt inside the zip | ~600,000–950,000 | Unpack the zip → drop the .txt |
| AncestryDNA | AncestryDNA.txt (possibly zipped) | Tab-separated .txt, alleles in two columns | ~650,000–700,000 | Unpack the zip → drop the .txt |
| FamilyTreeDNA | .csv.gz (Build 37) | Comma-separated CSV, gzipped | ~700,000 | Unpack the .gz first → .csv |
What DNAHelper accepts
.txt, .csv and .zip (zips are unpacked automatically — DNAHelper finds the .csv or .txt inside, so drop the package as it is). Not .gz — FamilyTreeDNA's export must be unpacked to .csv first (double-click in Finder/Explorer, or gunzip). Everything is read locally in the browser.
Same structure, different headers
All four contain the same core data — rsID, chromosome, position, genotype — but the layout differs: AncestryDNA stores the alleles in two separate columns, and headers, comment lines and separators vary. DNAHelper's parser identifies the provider automatically from the file's structure, so you do not need to prepare anything — just pick the right file.
No-calls and quality
Rows where the measurement failed are marked -- or 00 and filtered during analysis. The frequency is low but somewhat higher in older exports. Read why results can differ between files and what Build 37 means for FTDNA's export.
Also: Compatibility check — measure your file's coverage · What is a raw DNA file? · Download wait times · English overview