How Accurate Is 23andMe for Psychiatric Medications?

If you have taken a 23andMe test, your raw data file contains genetic variants that influence how your body processes many common antidepressants. 23andMe's own health reports touch on a handful of pharmacogenomic markers, but the raw data itself covers significantly more — including variants in the CYP2D6 gene and the CYP2C19 gene, the two genes most relevant to antidepressant metabolism.

This article explains which antidepressant-relevant genes are in your 23andMe data, how accurate genetic testing for psychiatric medications really is, what pharmacogenomics can and cannot predict, and how to turn raw data into a usable pharmacogenomic report. For the full picture, see our complete guide to pharmacogenomics from raw DNA data.

What 23andMe Tests, What It Doesn't, and Where Raw Data Fills the Gap

23andMe uses genotyping arrays that test hundreds of thousands of specific genetic positions (SNPs) across your genome. Your raw data file includes results for all of these positions — far more than what 23andMe displays in its consumer-facing reports.

For antidepressant pharmacogenomics, the key variants sit in two genes: CYP2D6 and CYP2C19. 23andMe's arrays include many of the SNPs that define common star alleles for both genes. For CYP2C19, that is enough — the information is already in your raw data and simply needs a dedicated analysis to extract it and map it to clinical guidelines.

For CYP2D6 it is not enough, and that limitation is decisive: consumer genotyping arrays cannot detect CYP2D6 gene deletions, duplications, or CYP2D6/CYP2D7 hybrid alleles, and those structural variants are what separate a poor metabolizer from an ultrarapid one. A diplotype assigned from SNPs alone would be a guess. DecodeMyBio therefore does not call CYP2D6 from raw data at all — we will never invent a result the data cannot support. For more on what your raw data contains, see our guide on what to do with your 23andMe raw data.

Which Antidepressant-Relevant Genes Are in 23andMe Data?

Two genes dominate the pharmacogenomics of antidepressants. Only one of them can be called from consumer DNA data:

CYP2D6 metabolizes approximately 25% of all prescribed medications, including many antidepressants — paroxetine, fluoxetine, fluvoxamine, venlafaxine, nortriptyline, and atomoxetine. Your 23andMe raw data includes some CYP2D6 SNPs, but not the gene deletions and duplications that also determine metabolizer status, so no reliable phenotype can be assigned from it — DecodeMyBio does not report a CYP2D6 result. For variant details (star alleles, activity scores, structural variant limitations), see the CYP2D6 gene guide; for the full list of affected antidepressants, see CYP2D6 and antidepressants.

CYP2C19 is the primary metabolic pathway for escitalopram (Lexapro), sertraline (Zoloft), citalopram (Celexa), and several tricyclic antidepressants. Your raw data includes the CYP2C19 *2, *3, and *17 variants that define the most common metabolizer phenotypes. For details on how CYP2C19 affects SSRIs specifically, see our CYP2C19 and SSRI metabolism guide.

What Pharmacogenomics Can and Cannot Predict About Antidepressants

Pharmacogenomics shows how your body metabolizes specific drugs. It does not predict whether a particular antidepressant will effectively treat your depression, anxiety, or other condition. This is an important distinction that is sometimes lost in marketing materials for pharmacogenomic testing.

What pharmacogenomics can tell you:

• Whether you metabolize a specific antidepressant faster or slower than the general population
• Whether standard dosing may result in drug levels that are higher or lower than intended
• Whether clinical guidelines suggest dose adjustments or alternative drug considerations for your genotype
• Your metabolizer status (poor, intermediate, normal, rapid, or ultrarapid) for CYP2D6 and CYP2C19

What pharmacogenomics cannot tell you:

• Which antidepressant will be most effective for your condition
• Whether you will experience side effects on a specific medication
• How long a medication will take to start working
• What dose will be optimal for you — dosing involves many non-genetic factors

Antidepressant response is influenced by many factors beyond genetics: the specific condition being treated, co-occurring conditions, other medications, lifestyle factors, and individual neurobiology. Pharmacogenomics provides one piece of this puzzle — the metabolic piece — and is most useful when interpreted alongside clinical context by your healthcare provider.

CPIC-Backed Drug-Gene Interactions for Common Antidepressants

The Clinical Pharmacogenetics Implementation Consortium (CPIC) publishes evidence-based guidelines for drug-gene interactions. For antidepressants, several CPIC Level A guidelines exist — the strongest evidence category with specific, actionable recommendations:

• Escitalopram (Lexapro) — CYP2C19: Poor metabolizers may have higher drug levels. CPIC recommends considering a 50% dose reduction or an alternative SSRI not primarily metabolized by CYP2C19.
• Sertraline (Zoloft) — CYP2C19: Poor metabolizers may need dose reduction. Ultrarapid metabolizers may have reduced drug levels.
• Paroxetine (Paxil) — CYP2D6: Ultrarapid metabolizers may have reduced levels. Poor metabolizers may have increased levels and higher side-effect risk. CPIC provides dosing guidance by phenotype.
• Venlafaxine (Effexor) — CYP2D6: Metabolizer status affects the ratio of parent drug to active metabolite. CPIC provides phenotype-specific recommendations.
• Amitriptyline — Both CYP2D6 and CYP2C19: CPIC guidelines recommend dose adjustments or alternative therapy based on metabolizer status for either gene.
• Aripiprazole (Abilify) — CYP2D6: Often co-prescribed with antidepressants. Poor metabolizers may need dose reductions per DPWG (Dutch Pharmacogenetics Working Group) guidelines.
• Atomoxetine (Strattera) — CYP2D6: While primarily an ADHD medication, atomoxetine is sometimes used alongside antidepressants. Poor metabolizers can have up to 10-fold higher drug levels.

These are summaries — not prescribing recommendations. The full CPIC guidelines contain specific dosing tables and clinical context. Your prescriber can use pharmacogenomic results alongside these guidelines to inform medication decisions.

How Consumer DNA Data Compares to Clinical PGx Testing

Clinical pharmacogenomic tests like GeneSight and others use targeted sequencing or genotyping with clinical-grade validation. They are ordered by clinicians, typically require a new sample (cheek swab or blood draw), and may use proprietary algorithms to generate their reports.

Consumer DNA data from 23andMe or AncestryDNA uses genotyping arrays that cover many of the same key variants but were designed primarily for ancestry and trait reporting, not pharmacogenomics. The raw data can be re-analyzed for pharmacogenomic variants, but with the important caveat that structural variants (gene deletions and duplications) are not reliably detected.

For many people, a raw data analysis provides clinically meaningful information at a fraction of the cost. For others — particularly those with clinical concerns about CYP2D6 gene deletions — a clinical-grade test may be more appropriate. See our detailed comparison of testing options for a full breakdown of coverage, cost, and methodology differences.

How to Get a Pharmacogenomic Analysis From Your 23andMe Data

If you already have 23andMe raw data, getting an antidepressant pharmacogenomic report involves three steps:

1. Download your raw data from your 23andMe account (Settings → 23andMe Data → Download Raw Data).
2. Upload to DecodeMyBio — our pipeline extracts CYP2C19 variants, assigns star alleles and metabolizer phenotypes, and maps results to CPIC guidelines for each CYP2C19-dependent antidepressant. CYP2D6 is not part of the pipeline: it requires a structural-variant caller a genotyping array cannot support.
3. Review your results with your healthcare provider. Your psychiatric medication results lay out your metabolizer status and CPIC guidance for quick provider review.

The same raw data file also works for broader medication results covering non-psychiatric medications, or you can view a sample to see what the output looks like before uploading.

Frequently Asked Questions

Does 23andMe test for the genes that affect antidepressant response?

23andMe's genotyping array includes variants in CYP2D6 and CYP2C19, the two genes most relevant to antidepressant metabolism, and 23andMe's standard health reports do not provide a comprehensive antidepressant pharmacogenomic analysis. A dedicated raw data analysis extracts the CYP2C19 variants and maps them to CPIC clinical guidelines. CYP2D6 is a different case: the array carries some of its SNPs but not the gene deletions and duplications that define metabolizer status, so DecodeMyBio does not call CYP2D6 rather than report a result the data cannot support.

Can a pharmacogenomic test tell me which antidepressant will work best for my depression?

No. Pharmacogenomics shows how your body metabolizes specific drugs — not which drug will be most effective for your condition. It helps guide dosing decisions and flag potential metabolism issues, not predict therapeutic response. Antidepressant effectiveness depends on many factors beyond drug metabolism.

What is the difference between 23andMe's health reports and a pharmacogenomic analysis of raw data?

23andMe's health reports cover a limited set of pharmacogenomic markers (such as CYP2C19 metabolizer status for a few drugs). A dedicated raw data analysis covers substantially more — extracting all available CYP2C19 variants, assigning diplotypes and activity scores, and mapping results to CPIC guidelines for each specific antidepressant, alongside the other pharmacogenes an array can support. It does not add a CYP2D6 result: no analysis of array data can produce one reliably, because the deciding structural variants are not in the file.

Is consumer DNA data accurate enough for pharmacogenomic decisions?

Consumer genotyping arrays are generally reliable for the specific variants they test. However, they cannot detect all possible variants, and structural changes like CYP2D6 gene deletions or duplications are not reliably captured. Results from consumer data should be discussed with your healthcare provider and may be confirmed with clinical testing if needed.

Should I get clinical pharmacogenomic testing instead of using my 23andMe data?

It depends on your situation. Consumer raw data analysis provides clinically meaningful information for most people at a lower cost. Clinical tests offer broader coverage (including structural variants) and come with clinician involvement. See our comparison page for a detailed breakdown. Many people start with a raw data analysis and pursue clinical testing only if results suggest it would add value.