Skip to content

Pharmacogenomics Intelligence Agent

Part of the Precision Intelligence Network — one of 11 specialized agents sharing a common molecular foundation within the HCLS AI Factory.

The Pharmacogenomics Intelligence Agent is a domain-specialized retrieval-augmented generation (RAG) system that translates patient genetic data into actionable drug prescribing recommendations. It searches 15 Milvus vector collections containing pharmacogene references, CPIC/DPWG clinical guidelines, drug-gene interactions, HLA hypersensitivity associations, phenoconversion models, validated dosing algorithms, clinical evidence, and population allele frequency data.

Key Capabilities

Capability Detail
Pharmacogenes 25 genes including CYP2D6, CYP2C19, CYP2C9, DPYD, TPMT, NUDT15, SLCO1B1
Drugs Covered 100+ across 12 therapeutic categories
Dosing Algorithms 9 genotype-guided (IWPC warfarin, tacrolimus, fluoropyrimidine, thiopurine, clopidogrel, simvastatin, SSRI, phenytoin, TCA)
HLA Screening 15 HLA-drug hypersensitivity associations
Phenoconversion 60 CYP inhibitors/inducers across CYP2D6, CYP2C19, CYP2C9, CYP3A4/5
Vector Collections 15 Milvus collections with BGE-small-en-v1.5 embeddings (384-dim)
Clinical Workflows 8 workflows across 10 Streamlit UI tabs
Test Suite 1,001 tests across 16 test files

Architecture

STREAMLIT UI (8507)  ──►  FastAPI REST API (8107)  ──►  PGxRAGEngine
     │                          │                          │
     └── 10 Tabs                └── 16+ Endpoints          └── 15 Collections
         Dashboard                  /health                    pgx_gene_reference
         Drug Check                 /query                     pgx_drug_guidelines
         Medication Review          /v1/pgx/drug-check         pgx_drug_interactions
         Warfarin Dosing            /v1/pgx/warfarin-dose      pgx_hla_hypersensitivity
         Chemo Safety               /v1/pgx/hla-screen         pgx_phenoconversion
         HLA Screening              /v1/pgx/phenoconversion    pgx_dosing_algorithms
         Report Generator           ...                        ...
         Evidence Explorer
         Phenoconversion Modeler
         Population Analytics

Clinical Pipelines

  • Star Allele Calling -- VCF variants to pharmacogene star allele nomenclature
  • Phenotype Translation -- Diplotypes to CPIC-standardized metabolizer phenotypes via activity scores
  • Drug-Gene Matching -- CPIC guideline lookup with alert severity classification
  • Phenoconversion Detection -- CYP inhibitor/inducer phenotype adjustment
  • HLA Screening -- Pre-prescription HLA-drug hypersensitivity screening
  • Genotype-Guided Dosing -- 9 validated dosing algorithms with patient-specific calculations

Documentation

  • Project Bible

    Complete system reference covering architecture, file inventory, collections, clinical pipelines, and configuration.

    Project Bible

  • White Paper

    Technical white paper with clinical validation, performance benchmarks, and comparison with existing PGx solutions.

    White Paper

  • Demo Guide

    Step-by-step walkthrough of all 10 UI tabs with sample inputs, expected outputs, and talking points.

    Demo Guide

  • Deployment Guide

    Docker Compose deployment, manual setup, Milvus tuning, security checklist, and monitoring configuration.

    Deployment Guide

  • Learning Guide -- Foundations

    Introduction to pharmacogenomics: CYP450 enzymes, star alleles, diplotypes, activity scores, CPIC guidelines.

    Learning Guide -- Foundations

  • Learning Guide -- Advanced

    Advanced topics: phenoconversion modeling, multi-gene interactions, IWPC algorithm, HLA pharmacovigilance, RAG architecture.

    Learning Guide -- Advanced

Tech Stack

Component Technology
LLM Claude Sonnet 4.6 (Anthropic)
Vector DB Milvus 2.4 with IVF_FLAT / COSINE
Embeddings BGE-small-en-v1.5 (384-dim)
API FastAPI
UI Streamlit with NVIDIA dark theme
Compute NVIDIA DGX Spark
Export Markdown, JSON, PDF, FHIR R4

Clinical Decision Support Disclaimer

This agent is a clinical decision support research tool. It is not FDA-cleared and is not intended as a standalone diagnostic device. All recommendations should be reviewed by qualified healthcare professionals. Apache 2.0 License.