PharmGKB (Now ClinPGx): The Pharmacogenomics Knowledgebase

ClinPGx is the current name for what most of the pharmacogenomics field still knows as PharmGKB, the Pharmacogenomics Knowledgebase: the NIH-funded resource, hosted at Stanford University since 2000, that curates the evidence linking specific genetic variants to drug response. Per ClinPGx’s own FAQ, “ClinPGx was formerly the PharmGKB, and the ClinPGx website and database replace the previous PharmGKB website and database” — and pharmgkb.org itself now redirects permanently to clinpgx.org. The rename is not just a URL change: ClinPGx describes itself as comprising three components under one platform — PharmGKB’s own curated knowledgebase, the Clinical Pharmacogenetics Implementation Consortium (CPIC) dosing guidelines that the PharmGKB team co-founded, and PharmCAT, a genotype-to-phenotype annotation tool — with CPIC’s separate website and database still being “folded into ClinPGx over time,” so the consolidation is an ongoing process rather than a single completed cutover. For citation, linking, and search purposes, treat PharmGKB and ClinPGx as the same underlying resource under its current name.

What PharmGKB/ClinPGx actually curates

The core of the resource is clinical annotations: curated summaries of the relationship between a specific gene variant (or star allele) and a drug’s efficacy, dosing, toxicity, or metabolism, each backed by cited primary literature and assigned a level of evidence — ranging from strong, clinical-guideline-referenced evidence with replication across multiple studies, down to preliminary or single-study findings. Around that core sit several linked outputs: dosing guidelines (co-developed with CPIC, at St. Jude Children’s Research Hospital and Stanford, for gene-drug pairs with strong enough evidence to guide prescribing directly), drug label pharmacogenomic annotations (tracking which regulators — the FDA, EMA, PMDA, Health Canada, and Swissmedic — reference pharmacogenomic information on a given drug’s label and what that language actually says), pathway diagrams for how a drug is absorbed, metabolized, and acted on, and gene-centric summaries for the field’s most consequential pharmacogenes (CYP2D6, CYP2C19, TPMT, DPYD, and similar). PharmCAT, the third component, takes this curated knowledge a step further into software: it is a tool that translates a patient’s or research sample’s raw genotype data into a predicted phenotype (e.g., “poor metabolizer”) using the same underlying gene definitions.

Why the rename happened, and what actually changed

The practical answer, stated plainly on ClinPGx’s own FAQ, is that PharmGKB’s website and database were replaced by a new platform that also brings CPIC’s guidelines and PharmCAT’s annotation tool into one place — rather than a researcher needing to separately consult pharmgkb.org for clinical annotations and a CPIC guideline PDF for dosing recommendations. What has NOT changed: the host institution (Stanford University, joined for CPIC by St. Jude Children’s Research Hospital and for PharmCAT by the University of Pennsylvania), the NIH funding base (multiple institutes, including NHGRI, NICHD, NIDA, and NCI, have funded the underlying work), and the curated-evidence model itself. What HAS changed, in practice: the public URL (pharmgkb.org now 301-redirects to clinpgx.org), the site name researchers should cite going forward, and — gradually — the degree of integration between the three components as CPIC’s separate site is absorbed into the shared platform over time.

What makes something a PharmGKB/ClinPGx entry, specifically

Not every mention of a gene-drug relationship in the literature belongs in ClinPGx. What qualifies is that the resource’s own curation team has reviewed the underlying evidence, written a structured clinical annotation (gene, variant, drug, phenotype category, and a level of evidence), and published it under a stable, citable ClinPGx identifier — distinct from a primary research paper simply reporting an association, and distinct from a general pharmacology reference that covers drug-drug interactions without a genetic component.

• A clinical pharmacogenomics team preparing a CYP2C19-clopidogrel dosing protocol references ClinPGx’s dosing guideline (co-published with CPIC) rather than re-deriving a recommendation from the primary meta-analyses the guideline itself cites.
• A manuscript author citing a gene-drug relationship that ClinPGx has graded at its strongest evidence tier links directly to the ClinPGx clinical annotation page rather than restating every underlying study individually — the curated annotation is itself the citable artifact, similar in spirit to citing a Research Resource Identifier (RRID) rather than re-describing a reagent from scratch.

By contrast, a general drug-drug interaction lookup — checking whether two prescribed drugs interact with each other, with no genetic variant involved — is not a PharmGKB/ClinPGx use case. The resource’s scope is specifically genotype-to-drug-response relationships, not pharmacology or interaction-checking generally.

Where PharmGKB/ClinPGx fits among research data resources

Within CASRAI’s broader research-data-management vocabulary, ClinPGx functions as a curated, domain-specific knowledgebase rather than a general-purpose data repository: it doesn’t just host deposited datasets, it synthesizes evidence across many primary studies into one graded annotation. That distinguishes it from a controlled-access archive like dbGaP, which hosts the underlying genotype-phenotype study data itself (much of which is exactly the kind of primary evidence a ClinPGx clinical annotation later cites and curates). Researchers working with individual-level genomic data governed by a biobank or clinical cohort, and needing to interpret a specific patient’s or participant’s genotype, will typically consult ClinPGx (and PharmCAT) for the interpretation layer and dbGaP or a similar controlled-access archive for the underlying individual-level data itself. Like other named domain repositories, its FAIR posture rests on stable identifiers for genes, variants, and drugs (PharmVar’s star-allele nomenclature, which ClinPGx has adopted for gene variant naming, is one example) and on the clinical annotations themselves being independently citable and dated.

Frequently asked questions

Is PharmGKB still a separate website from ClinPGx? No. pharmgkb.org 301-redirects to clinpgx.org, and ClinPGx’s own FAQ describes the ClinPGx site and database as replacing the previous PharmGKB site and database, not running alongside it.

Does “ClinPGx” replace CPIC and PharmCAT too, or just PharmGKB? ClinPGx describes itself as comprising all three — PharmGKB, CPIC, and PharmCAT — under one platform, but as of this writing CPIC’s own separate website and database are still being folded in “over time” rather than already fully merged, so expect CPIC-branded content to persist for a while yet.

Should I cite “PharmGKB” or “ClinPGx” in new work? Cite ClinPGx going forward, since that is the current, live resource; if referencing PharmGKB-era publications or identifiers that predate the rename, it’s reasonable to note the former name for continuity, the way current literature increasingly writes “ClinPGx (formerly PharmGKB).”