Pharmacogenetic Testing
Which DNA Test Should I Get for Pharmacogenetics?
If you already have results from 23andMe, AncestryDNA or MyHeritage, use them. If you're buying a test with your medications in mind, here is what we found when we tested each option against reference samples.
Reviewed by the Gene2Rx team. Updated 2026-10-16.
People ask us this a lot, so we measured it. We took reference samples whose pharmacogenes have been characterized by clinical labs,[1] trimmed their whole genome data down to the positions each consumer test reads, and ran Gene2Rx on every version. Short answer: the consumer tests are close to each other, whole genome sequencing is better, and the biggest difference is one gene, CYP2D6.
92.2–93.9% of metabolizer results Gene2Rx got right from 23andMe, AncestryDNA, MyHeritage and FamilyTreeDNA data, compared with 96.6% from whole genome data
| Input | Gene2Rx |
|---|---|
| Whole genome | 96.6% (95% CI 94.9 to 98.1) |
| 23andMe | 92.2% (95% CI 89.8 to 94.3) |
| AncestryDNA | 93.4% (95% CI 91.2 to 95.3) |
| MyHeritage | 93.2% (95% CI 91.0 to 95.1) |
| FamilyTreeDNA | 93.9% (95% CI 91.7 to 95.8) |
Share of metabolizer results Gene2Rx got right from each kind of DNA data, measured on GeT-RM reference samples.
- Scored:
- The gene result that drives medication advice matches the reference: metabolizer status (for example "intermediate metabolizer") or, for SLCO1B1, transporter function.
- Genes:
- Gene2Rx: 14 genes (CYP1A2, CYP2A6, CYP2B6, CYP2C19, CYP2C9, CYP2D6, CYP3A4, CYP3A5, CYP4F2, DPYD, NUDT15, SLCO1B1, TPMT, UGT1A1).
- Results:
- 590 gene results from 95 reference samples for Gene2Rx; a missing answer counts as wrong. The line around each marker is the 95% confidence range.
- Not scored:
- 20 genes with no usable reference result (ABCG2, ADH1B, AKT1, ALDH2, BCHE, CACNA1S, CFTR, CHRNA5, COMT, F2, F5, FAAH, G6PD, GIPR, GLP1R, IFNL3, NAT2, RYR1, SULT1A1, VKORC1); 36 results that depend on a whole-gene deletion or duplication (1 CYP2A6, 35 CYP2D6), which none of these inputs can show.
- Consumer tests:
- Simulated from the whole genome data, keeping only the positions each test reads.
What we found
The big consumer tests perform about the same
From 23andMe, AncestryDNA, MyHeritage and FamilyTreeDNA data, Gene2Rx got metabolizer status right between 92.2% and 93.9% of the time. Those differences are within the margin of error, so no single company is clearly best for pharmacogenetics. Pick on price, on the ancestry and health features you want, or on which test a relative already took. If you already have one of these tests, you don't need another.
Whole genome sequencing is the most accurate input
From whole genome data, Gene2Rx got metabolizer status right 96.6% of the time. Sequencing reads every position in your pharmacogenes instead of a fixed list, so it catches rarer variants that consumer tests never look for. It costs more than a consumer test, and Gene2Rx reports from whole genome files cost $20 extra. If you're choosing a test mainly for medication reasons, it's the better buy.
CYP2D6 is where consumer tests fall short
CYP2D6 handles about a quarter of commonly prescribed drugs, including many antidepressants, antipsychotics and opioids such as codeine and tramadol. From consumer test data, Gene2Rx got CYP2D6 right 60 to 72% of the time, against 95% from whole genome data. Consumer tests read only some of the positions that define CYP2D6 variants. Some people also carry an extra copy of the gene or are missing one, and neither consumer tests nor a whole genome VCF show that. If a CYP2D6 medication matters to you, ask your clinician about a clinical CYP2D6 test.[2]
A normal result from a consumer test is less certain
When a consumer test doesn't read a rare variant, the result for that gene usually comes out as normal. Most of the errors we saw from consumer test data were of this kind. A result that flags a problem is more reliable than one that says everything is typical, so treat a normal result for an important medication as a good sign, not a guarantee.
Already have 23andMe, AncestryDNA or MyHeritage results? Use them. Buying a test for your medications? Whole genome sequencing is the better choice.
How your genetics can play a role
How much the test matters depends on the gene. For most of the genes behind common prescriptions, every option did well.
| Gene | What it affects |
|---|---|
| CYP2C19 | Affects clopidogrel, proton pump inhibitors and several SSRIs such as citalopram, escitalopram and sertraline. Every consumer test read it almost as well as whole genome data. |
| CYP2C9 | Affects warfarin, phenytoin and many anti-inflammatory painkillers. Consumer tests read it well. |
| SLCO1B1 | Affects statin muscle side effects, especially with simvastatin. Results were the same from consumer tests and whole genome data. |
| DPYD | Affects the chemotherapy drugs fluorouracil and capecitabine, where a missed variant can cause severe toxicity. Gene2Rx read it well from every input we tested, but the rarest DPYD variants are only visible to sequencing, so oncologists still order a dedicated test before treatment. |
| CYP2D6 | The gene where the test matters most, as described above. Whole genome data is far better than any consumer test, and a clinical test is better still because it can count gene copies. |
Have raw data from 23andMe, AncestryDNA or MyHeritage? You don't need another test.
Upload the file you already have and get guidance on up to 120 medications in minutes.
How to choose
Start from what you have. If you or a relative already tested with a consumer company, download the raw data and upload it to Gene2Rx; there's no need to buy a second consumer test. If you're buying a test and medications are the main reason, choose whole genome sequencing from a provider that gives you a VCF file. If your doctor needs results to change a prescription, a clinical pharmacogenetic test ordered through them is the standard, and you can enter those results into Gene2Rx afterwards.
What you can do next
- Already tested with 23andMe, AncestryDNA or MyHeritage? Download your raw data file from your account settings.
- Buying a new test for medication reasons? Choose a whole genome sequencing provider that lets you download a VCF. Gene2Rx reads GRCh37 and GRCh38 files.
- Upload the file to Gene2Rx. Reports start at $5, and the full report covers 120 medications.
- Take the report to your prescriber. Consumer data is a starting point, and they may confirm an important result with a clinical test.
Related medications
See how genetics may affect your response to Sertraline, Escitalopram, Clopidogrel, Codeine, Tramadol, Fluorouracil, Capecitabine, Simvastatin.
Related guides
- Whole Genome Sequencing for Pharmacogenetics: A Complete Guide
- Nebula Genomics Drug Response: What You'll Actually See in a Report From Your WGS Data
- Dante Labs Pharmacogenetics: Turn Your Dante WGS Data Into a PGx Report
- Sequencing.com Pharmacogenetics: Turn Your WGS Data Into a PGx Report
- 23andMe Pharmacogenetics: How to Get a Drug Response Report From Your Existing Data
- AncestryDNA for Drug Testing: Get Pharmacogenetics From Your Ancestry Data
Frequently asked questions
Is 23andMe or AncestryDNA better for pharmacogenetics?
In our testing they were very close: Gene2Rx got metabolizer status right 92.2 to 93.9% of the time across 23andMe, AncestryDNA, MyHeritage and FamilyTreeDNA data, a gap within the margin of error. Either works.
Is whole genome sequencing worth it for pharmacogenetics?
If you're buying a test with your medications in mind, yes. Gene2Rx got 96.6% of metabolizer results right from whole genome data, and CYP2D6 in particular is much better. If you already have consumer test results, start with those.
Does 23andMe's own drug response report cover the same thing?
No. 23andMe's pharmacogenetics reports cover a few genes. The raw data behind them can tell you about many more, which is what Gene2Rx reads.
How did you test this?
We used reference samples whose pharmacogenes were characterized by clinical labs (the CDC's GeT-RM program), cut each sample's whole genome data down to the positions each consumer test reads, and checked Gene2Rx's results against the reference. Real files have occasional read errors, so real-world results may be slightly lower.
References
- Pratt VM et al., Journal of Molecular Diagnostics. Characterization of 137 Genomic DNA Reference Materials for 28 Pharmacogenetic Genes: A GeT-RM Collaborative Project (2016). pubmed.ncbi.nlm.nih.gov
- Pratt VM et al., Journal of Molecular Diagnostics. Recommendations for Clinical CYP2D6 Genotyping Allele Selection (2021). pmc.ncbi.nlm.nih.gov
- Clinical Pharmacogenetics Implementation Consortium (CPIC). CPIC Guidelines. cpicpgx.org
Already have DNA data? See what it says about your medications.
Disclaimer: This content is for educational purposes only and is not a substitute for professional medical advice, diagnosis, or treatment. Always consult your healthcare provider before making changes to your medication. Never stop or change a medication without medical supervision.