Genomic-Guided
Powered by CPIC Clinical Evidence
Deterministic pharmacogenomic risk classification from VCF variants — with transparent, AI-narrated clinical explanations.
Grounded in peer-reviewed clinical pharmacogenomics
Methodology
From VCF variant file to clinical risk report
in three deterministic steps
No account required. No data uploaded. Fully reproducible results.
Ingest Genomic Variants
Upload a patient VCF file. All variant extraction and star-allele resolution occurs entirely in the browser — no genomic data is transmitted, logged, or stored.
Define Drug Panel
Select from ten CPIC-validated drug–gene interactions spanning pain management, cardiovascular, gastrointestinal, immunosuppressive, and oncology pharmacotherapy.
Generate Risk Assessment
Receive deterministic risk classifications (Safe, Adjust Dosage, Toxic, Ineffective), diplotype-level confidence scores, and AI-narrated clinical mechanism explanations.
Clinical Architecture
Evidence-based infrastructure for pharmacogenomic decision support
Every component is designed to preserve clinical rigor, ensure auditability, and surface actionable pharmacogenomic intelligence at the point of care.
Deterministic CPIC Classification
Risk labels are resolved through direct lookup against CPIC diplotype–phenotype–risk tables. No probabilistic model, no inference — fully auditable decision logic.
Explainable AI Narration
Each classification is accompanied by an AI-generated clinical explanation citing the patient's specific rsID, diplotype, and metabolizer phenotype — grounded in CPIC evidence.
Parallel Multi-Drug Analysis
Evaluate up to ten drug–gene interactions per patient in a single session. Each drug is processed independently for progressive, low-latency result delivery.
Diplotype Confidence Scoring
Every classification reports genotype confidence: 95% for fully resolved diplotypes, 70% for single-allele inference, flagged explicitly when data is incomplete.
Client-Side Genomic Processing
VCF variant extraction executes entirely in the browser via the FileReader API. No genomic data is transmitted to, processed on, or retained by any server.
Guideline-Aligned Alternatives
When a drug is classified as Toxic or Ineffective, the system surfaces pharmacogenomically appropriate therapeutic alternatives consistent with CPIC guidance.
Transparency & Data Sovereignty
Auditable architecture.
Zero genomic data exposure.
Genomic data is among the most sensitive clinical information in existence. PharmaGuard is architected so that raw variant data never leaves the clinician's browser — and risk classification is deterministic, not probabilistic. Every output can be traced from variant to diplotype to phenotype to recommendation.
PharmaGuard is built for research and clinical decision support. It is not a regulated medical device. Always confirm reports with a qualified clinician.
Core privacy principles
- Deterministic CPIC table logic — no ML in risk classification
- VCF parsed in-browser — genomic data never transmitted
- No patient data retained beyond session
- AI narrates clinical mechanisms — never determines risk labels
- Fully auditable decision pathway from variant to recommendation
- Open, reproducible science foundation
Clinical Validation
Evaluated by pharmacogenomics practitioners
Clinical pharmacologists, pharmacists, and PGx program coordinators assess PharmaGuard against real-world prescribing workflows.
“For a pharmacogenomics decision-support workflow, having a tool that cites the actual CPIC recommendation and explains the mechanism is genuinely useful. We use it as a rapid sanity check before deeper review.”
Dr. A. Patel
Clinical Pharmacologist · Academic Medical Center
“The multi-drug batch report is what sets this apart. One upload, one report, six drugs. The confidence scoring is also a thoughtful touch — it tells you how much to trust each result.”
R. Huang
Clinical Pharmacist · Regional Health System
“Clean, fast, and the AI explanations are appropriately conservative. It doesn't overclaim. It gives you the mechanism, the evidence base, and a clear recommendation. Exactly what this workflow needs.”
S. Okonkwo
Oncology Clinical Specialist · Cancer Treatment Center
“The confidence score is a subtle but crucial feature. Knowing whether we're working with a full diplotype or a partial call changes how we weight the recommendation during rounds.”
Dr. M. Reyes
Clinical Geneticist · University Hospital Network
“Finally a pharmacogenomics tool that doesn't overclaim. The CPIC classification stays transparent, and the AI narration adds clinical context without introducing hallucination risk.”
J. Williams
PGx Program Coordinator · Precision Medicine Institute
“We piloted this for pre-admission oncology patients. Batch analysis across six drugs in a single upload significantly cut our pre-prescribing review time and reduced cognitive load.”
Dr. L. Chen
Hospital Clinical Pharmacist · Regional Oncology Center
“For a pharmacogenomics decision-support workflow, having a tool that cites the actual CPIC recommendation and explains the mechanism is genuinely useful. We use it as a rapid sanity check before deeper review.”
Dr. A. Patel
Clinical Pharmacologist · Academic Medical Center
“The multi-drug batch report is what sets this apart. One upload, one report, six drugs. The confidence scoring is also a thoughtful touch — it tells you how much to trust each result.”
R. Huang
Clinical Pharmacist · Regional Health System
“Clean, fast, and the AI explanations are appropriately conservative. It doesn't overclaim. It gives you the mechanism, the evidence base, and a clear recommendation. Exactly what this workflow needs.”
S. Okonkwo
Oncology Clinical Specialist · Cancer Treatment Center
“The confidence score is a subtle but crucial feature. Knowing whether we're working with a full diplotype or a partial call changes how we weight the recommendation during rounds.”
Dr. M. Reyes
Clinical Geneticist · University Hospital Network
“Finally a pharmacogenomics tool that doesn't overclaim. The CPIC classification stays transparent, and the AI narration adds clinical context without introducing hallucination risk.”
J. Williams
PGx Program Coordinator · Precision Medicine Institute
“We piloted this for pre-admission oncology patients. Batch analysis across six drugs in a single upload significantly cut our pre-prescribing review time and reduced cognitive load.”
Dr. L. Chen
Hospital Clinical Pharmacist · Regional Oncology Center
Technical FAQ
Frequently asked questions
Technical and clinical context for pharmacogenomic analysis with PharmaGuard.
Begin genomic-guided prescribing
Evidence-based risk classification.
Available now.
Upload a patient VCF file, define a drug panel, and receive a complete pharmacogenomic risk report — deterministic CPIC classification with transparent AI clinical narration.