CYP2C19 in your raw data: how you metabolize certain drugs

Genetic marker guide · Updated June 2026

23andMe & raw DNA

What CYP2C19 does

CYP2C19 is one of the liver's cytochrome P450 enzymes — a family of proteins that does much of the body's chemical housekeeping, including breaking down (and sometimes switching on) the drugs we take. Think of these enzymes as the metabolic assembly line that decides how long a medication stays active in your bloodstream. When that line runs faster or slower than average, the amount of active drug circulating in your body shifts, and so can the medication's effect.

CYP2C19 in particular handles a handful of widely-used drugs, including:

• Clopidogrel (Plavix) — an antiplatelet medication that actually needs CYP2C19 to activate it
• Proton-pump inhibitors (PPIs) — for acid reflux and ulcers (omeprazole and similar)
• Some antidepressants — certain SSRIs such as citalopram and escitalopram are examples that CYP2C19 helps clear
• Voriconazole — an antifungal medication where CYP2C19 status is well documented

The field that studies this — how your genes influence your response to medications — is called pharmacogenomics. It is one of the more mature corners of consumer genetics: the biology of CYP2C19 is well established, and clinicians already use it in real prescribing decisions. That maturity is exactly why the caution here is so important. Because the mechanisms are real and repeatable, it is tempting to draw conclusions and act on them. But the leap from "my raw data shows a variant" to "this drug will or won't work for me" is a clinical judgment that depends on validated testing, the specific drug, the dose, and everything else going on in your health — not on a lookup in a text file.

The metabolizer types

Pharmacogenomics describes CYP2C19 activity as a spectrum of metabolizer phenotypes. You inherit one copy of the gene from each parent, and the combination of those two copies — written using the star-allele system (*1, *2, *17, and others) — predicts roughly how much working enzyme you produce.

A few things to hold in mind about this table. First, the *1 allele is simply the reference "normal" version — it is defined by the absence of the known variants rather than by a marker of its own, which is part of why arrays can't always be certain about it. Second, loss-of-function alleles like *2 reduce or remove activity, while increased-function alleles like *17 push it the other way; a person can even carry one of each, which is where a simple lookup gets genuinely tricky to interpret. Third, these are predicted phenotypes — labels the field assigns based on genotype — not measurements of how your liver is actually behaving today. Real-world enzyme activity is also shaped by other medications you take, liver health, and interacting drugs.

Prodrugs versus active drugs: the direction matters

The single most important idea in CYP2C19 pharmacogenomics is that the consequence of low activity depends entirely on the drug. There is no universal "fast is good, slow is bad" rule.

Consider two cases:

• A prodrug is swallowed in an inactive form and must be converted by the body into its working form. Clopidogrel is the textbook example: CYP2C19 helps turn it into its active antiplatelet compound. For a prodrug, low CYP2C19 activity can mean less of the active drug is produced, so a poor metabolizer may get a reduced antiplatelet effect. This is well documented enough that clinicians sometimes order CYP2C19 testing before relying on clopidogrel in certain situations.
• A drug that CYP2C19 clears is already active when you take it, and the enzyme's job is to break it down and remove it. Here the logic flips: low activity can mean the drug is cleared more slowly and builds up, while very high activity can clear it so fast that standard doses do less. Several PPIs and antidepressants fall broadly into this category.

That reversal is the crux of why raw data is dangerous to act on alone. Seeing "reduced CYP2C19 activity" tells you nothing actionable until you know which medication is in question and whether it is a prodrug or an active drug — and even then, the actual decision involves clinical context a genotype cannot supply. This is precisely the kind of reasoning a doctor or pharmacist is trained to do, and precisely the kind of reasoning that goes wrong when people try to self-interpret.

Why raw data is incomplete and non-clinical

Consumer DNA files from 23andMe and AncestryDNA are generated by genotyping arrays — chips that read a predetermined set of positions in your genome rather than sequencing it fully. For CYP2C19 that has three practical limits:

1. Partial coverage of star alleles. Arrays typically report only some CYP2C19 star alleles. If a less common loss-of-function or increased-function allele isn't on the chip, your file simply won't mention it — which can make the picture look tidier than it is.
2. Single-position calls can be wrong. Arrays occasionally miscall an individual marker. In casual traits that's a curiosity; for a gene that touches medication response, a single miscalled position is a reason to distrust a raw-data readout on its own.
3. No clinical validation or quality control. Consumer arrays are not run, interpreted, or reported under the standards used for clinical pharmacogenetic testing. 23andMe's own materials are explicit that raw data is not for medical use, and that is the correct framing.

The takeaway isn't that raw data is useless — it can be a genuinely interesting prompt for a conversation. It's that raw data is a hint, not a result. A clinical pharmacogenetic test is a different instrument entirely: it targets the relevant alleles deliberately, runs under quality controls, and produces a report a prescriber can actually rely on.

The role of CPIC and clinical testing

When a clinician does have a validated CYP2C19 result in hand, they don't have to improvise what it means. The Clinical Pharmacogenetics Implementation Consortium (CPIC) publishes peer-reviewed, freely available guidelines that translate a known genotype into concrete prescribing guidance — for clopidogrel, certain antidepressants, voriconazole, and more.

Two features of CPIC are worth understanding as a curious reader:

• CPIC guidance is written for clinicians and is designed to be used with clinical-grade test results, not consumer raw data. It assumes the genotype feeding it has been established by validated testing.
• It is applied by a prescriber who is weighing the genotype alongside your diagnosis, other medications, and everything else in your chart. The guideline is one input into a professional decision, not a self-service instruction sheet.

In other words, the infrastructure to use CYP2C19 responsibly already exists — it just lives in the clinic, keyed to proper testing, and operated by professionals. That's the pathway your raw-data curiosity should feed into.

How to find CYP2C19 in your raw data

If you want to explore what your file contains — as a starting point for a conversation, not a conclusion — here's the responsible way to look:

1. Download your raw data (or from AncestryDNA / MyHeritage).
2. Look for the CYP2C19 star alleles your provider reports — commonly the *2 loss-of-function allele and the *17 increased-function allele.
3. Or use our free DNA explorer to read the file privately in your browser, with nothing uploaded.

Because CYP2C19 carries real clinical weight and arrays can miscall a single position or miss less common alleles, never treat the raw-data readout as final — and never let it drive a medication decision.

What to do with the result

If you take — or are about to take — clopidogrel, a long-term PPI, an affected antidepressant, or voriconazole, your possible CYP2C19 status is a genuinely reasonable thing to raise with your doctor or pharmacist. They can decide whether validated pharmacogenomic testing is warranted, interpret it against CPIC guidance, and factor in the rest of your health. Acting on it is their job, not a raw-data lookup's.

The healthiest way to hold a result like this is as a question, not an answer: "My raw data suggested I might be a poor metabolizer for CYP2C19 — is that worth confirming for the medication I'm on?" That single sentence, brought to a professional, is exactly what consumer genetics is good for. Changing a dose on your own is exactly what it is not.

For the rest of what your file holds, see our complete guide to analyzing 23andMe raw data, or browse the rest of the Quanome blog.

Frequently asked questions

What does CYP2C19 do?

CYP2C19 is a liver enzyme in the cytochrome P450 family that metabolizes several important drugs, including the antiplatelet drug clopidogrel (Plavix), proton-pump inhibitors (PPIs) for acid reflux, some antidepressants, and the antifungal voriconazole. Variants change how quickly you process these medications, which can shift their effect.

What are the CYP2C19 metabolizer types?

From slowest to fastest: poor, intermediate, normal, rapid, and ultrarapid metabolizers. Loss-of-function variants (such as the *2 allele) reduce or remove enzyme activity; increased-function variants (such as *17) raise it. Your combination of inherited alleles determines your predicted phenotype.

How do I find CYP2C19 in my 23andMe data?

You can search your raw file for the star alleles it reports, commonly *2 and *17. But 23andMe and AncestryDNA arrays only cover some CYP2C19 star alleles, and single-position genotyping can occasionally miscall a marker. Treat any raw-data readout as a starting point for a conversation, never as a clinical result.

Can I change my medication based on raw data?

No. Never start, stop, adjust, or dose any medication based on consumer raw data. That includes deciding whether a drug will 'work' for you. Pharmacogenomic decisions require clinical-grade testing and a doctor or pharmacist. This page is educational only.

Is 23andMe raw data the same as a clinical pharmacogenomic test?

No. Consumer arrays are not validated for clinical decision-making. They genotype a limited set of positions, may not capture the full star-allele picture, and are not run under the same quality controls as a clinical pharmacogenetic test. A clinician can order proper testing when it is warranted.

What is a prodrug, and why does it matter for CYP2C19?

A prodrug is a medication that is inactive as swallowed and must be converted by the body into its active form. Clopidogrel is the clearest example — CYP2C19 helps activate it. For a prodrug, low enzyme activity can mean less active drug; for a drug that CYP2C19 clears, low activity can mean the drug lingers. The direction depends on the specific medication.

What is CPIC?

The Clinical Pharmacogenetics Implementation Consortium (CPIC) publishes peer-reviewed, freely available guidelines that translate a known genotype into prescribing guidance for clinicians. CPIC guidelines are used with clinical-grade test results — not consumer raw data — and are applied by a prescriber, not the patient.

Should I get pharmacogenomic testing?

That is a decision for you and your clinician, usually most relevant if you take or are about to start a medication that CYP2C19 affects. If your raw data hints at a poor or ultrarapid metabolizer pattern, that is a reasonable thing to raise with a doctor or pharmacist, who can decide whether validated testing is appropriate.