23andMe Raw Data: Download, Upload and Interpret It (2026)

If you have taken a DNA test from 23andMe, AncestryDNA, MyHeritage, or FamilyTreeDNA, your raw data file already contains the genetic variants needed for pharmacogenomic analysis. This guide walks you through the entire process — from downloading your raw data to understanding what your results include.

The process takes under 5 minutes. You do not need to order a new test, provide a saliva sample, or wait for lab results. You are reusing data you already have.

What Is a Raw Data File?

When a consumer DNA testing company like 23andMe analyzes your saliva sample, they genotype hundreds of thousands of specific positions (SNPs) in your genome. The raw data file is a text file containing every variant they measured — commonly cited as roughly 600,000 to 700,000 data points for current 23andMe (v5) and AncestryDNA chips, though the exact count varies by chip version.

This file includes the pharmacogenomic variants in genes like CYP2C19, CYP2C9, SLCO1B1, and MTHFR — the same variants used in clinical pharmacogenomic testing. DecodeMyBio extracts this subset and maps it to CPIC clinical guidelines.

It also contains a handful of CYP2D6 SNPs — but not the gene deletions, duplications, or hybrid alleles that actually determine CYP2D6 metabolizer status. Calling that gene needs a specialized structural-variant caller a genotyping array cannot provide, so DecodeMyBio does not report a CYP2D6 result. Rather than guess, we tell you the truth: we will never invent a result the data cannot support.

How to Read Your 23andMe Raw Data File

Open the file in any text editor and you will find a header of comment lines beginning with #, then one row per genotyped position. Each row has four tab-separated columns: the variant identifier (rsID), the chromosome, the position, and your genotype — the two alleles you carry.

That single row is a real pharmacogenomic result. rs4244285 is the variant that defines CYP2C19*2, the most common loss-of-function allele in that gene. A GG genotype is the reference — no *2 allele on either copy. An AA genotype means two copies, which makes you a poor metabolizer for CYP2C19 and changes the CPIC recommendation for clopidogrel and several antidepressants.

Two things to know before you start reading positions manually. First, 23andMe files use genome build 37 (GRCh37), so the coordinates will not match a build 38 reference — rs4244285 sits at 10:96541616 on build 37 but 10:94781859 on build 38. Second, a genotype of -- means the assay failed to read that position, not that you carry a deletion.

Reading a handful of rows by hand is feasible. Turning them into phenotypes is not: a metabolizer status depends on the combination of alleles across a gene (your diplotype), plus a CPIC-standardized activity score. That translation is what an analysis service does for you.

What You Can Actually Do With Your Raw Data

The most clinically actionable use of the file is pharmacogenomics — working out how your genes affect your response to medications you may already be taking. This is not speculative territory: CPIC publishes peer-reviewed, genotype-specific prescribing recommendations, and the FDA includes pharmacogenomic information in the labeling of hundreds of medications.

These are established drug-gene pairs, each with a specific CPIC recommendation attached:

• CYP2C19 poor metabolizer + clopidogrel: clopidogrel may not be effectively activated, reducing its antiplatelet effect. CPIC recommends an alternative agent.
• CYP2C9 / VKORC1 variants + warfarin: certain genotypes need substantially reduced doses to avoid bleeding risk. CPIC provides genotype-guided dosing.
• SLCO1B1 reduced function + simvastatin: raises the risk of statin-related muscle effects. CPIC recommends a lower dose or a different statin.

Other uses exist and are worth knowing about, with the caveat that they rest on weaker evidence: methylation and nutrient-metabolism panels built on MTHFR and related genes, trait and ancestry reports, and broad variant-exploration tools that surface literature associations without clinical guidelines behind them. For a side-by-side comparison of what each type of service actually delivers, see the guide to the best 23andMe raw data analysis tools.

One reason to act on the file now rather than later: keeping your own copy is the only thing that guarantees long-term access to it, independent of what happens to the company that generated it — see what the 23andMe bankruptcy means for your data.

Step 1 — Download Your Raw Data File

Each DNA testing company provides a way to download your raw data. Here are the current steps for each provider (as of March 2026):

23andMe

1. Log in to your 23andMe account at you.23andme.com
2. Navigate to Settings (gear icon, top right)
3. Scroll to 23andMe Data and select Download Raw Data
4. Re-enter your password and confirm via the email verification link
5. Click Download — you will receive a .zip file containing a .txt file

Note: Do not unzip the file. DecodeMyBio accepts the .zip file directly.

AncestryDNA

1. Log in to your Ancestry account at ancestry.com
2. Go to DNA → Settings
3. Under DNA Data, click Download Raw DNA Data
4. Confirm your identity via email or password
5. Download the .zip file

Once you have your AncestryDNA file, see our guide to AncestryDNA pharmacogenetics for the medication-related markers it covers and how they map to CPIC guidelines.

MyHeritage

1. Log in at myheritage.com and go to DNA → Manage DNA kits
2. Click the three-dot menu next to your kit and select Download raw DNA data
3. Confirm and download

FamilyTreeDNA

1. Log in at familytreedna.com
2. Go to Results → Raw Data → Download Raw Data
3. Select Build 37 Autosomal (the standard format)
4. Download the file

Step 2 — Upload to DecodeMyBio

1. Go to the upload page
2. Click Choose File or drag and drop your raw data file (the .zip or .txt file you downloaded)
3. DecodeMyBio automatically detects your DNA testing provider and file format — no manual configuration needed
4. Processing takes approximately 30–60 seconds. You will see a progress indicator while your variants are being extracted and mapped to CPIC guidelines

Step 3 — Review Your Reports

Once processing completes, you can unlock six reports covering 19 genes and 150+ drug-gene interactions:

What You Get

• Decode+ — Covers cardiovascular medications (clopidogrel, warfarin, simvastatin), pain medications (codeine), and more. Each drug-gene interaction includes your metabolizer phenotype, CPIC evidence level, and the clinical recommendation.
• Psychiatric Medication — Covers SSRIs (escitalopram, sertraline, paroxetine), SNRIs (venlafaxine), tricyclics (amitriptyline), antipsychotics (aripiprazole), and ADHD medications (atomoxetine).
• Nutrition & Methylation — Covers MTHFR variants and nutrient metabolism genes relevant to folate, B12, and methylation pathways. For background on what methylation tests measure and their limitations, see the testing methylation pathways guide.
• Cannabis & CBD — Covers how your CYP2C9, CYP2C19, CYP3A5, and CYP1A2 genetics relate to THC metabolism, CBD processing, and drug interactions from smoked cannabis.
• Pain & Anesthesia — Covers opioid receptor sensitivity (OPRM1), pain sensitivity (COMT, BDNF), reward signalling (ANKK1), and scenario-based insights for surgery prep and chronic pain management. CYP2D6, the opioid-activation gene, is not included — it needs a specialized structural-variant caller that a consumer array cannot provide.
• Celiac & Gluten Screening — HLA-DQ2 and HLA-DQ8 tag-SNP screening for celiac disease susceptibility. This is not equivalent to clinical HLA typing and should not, on its own, be used to rule out celiac disease.

What Each Report Includes

• Your metabolizer phenotype for each gene (poor, intermediate, normal, or ultrarapid)
• Your diplotype and activity score — the specific allele combination and the CPIC-standardized score that determines your phenotype
• CPIC clinical recommendations — the specific guideline text for your phenotype-drug combination, with evidence level (A or B)
• A provider-ready summary view — a concise in-app format designed for a doctor or pharmacist to review in under 60 seconds

Supported File Formats

What About Privacy?

Your raw data file is processed locally in your browser session. For details on how your data is handled, see our privacy policy. DecodeMyBio does not sell, share, or retain your genetic data beyond the session needed to generate your results.

Limitations of Consumer DNA Data

Consumer genotyping arrays have known limitations compared to clinical-grade pharmacogenomic testing:

• Structural variants: Consumer arrays cannot detect gene deletions or duplications (e.g., CYP2D6 gene copy number variation). This means some ultrarapid metabolizer or poor metabolizer calls that depend on structural variants may be missed.
• Rare alleles: Only common, well-characterized alleles are covered. Very rare variants may not be on the array.
• Not a clinical diagnostic test: DecodeMyBio provides informational reports — not FDA-cleared diagnostic results. For high-stakes prescribing decisions, a provider-ordered pharmacogenomic test with CLIA-certified laboratory analysis may be appropriate.

For a detailed discussion, see our limitations page and methodology.