raxtik-etl Documentation

raxtik-etl is an enterprise-grade pharmacovigilance data engine, canonical schema platform, and reproducible pipeline for FDA FAERS and AEMS adverse event safety databases.

---

Getting Started

• Quickstart Guide
  • 1. Using the Python SDK (FAERS)
  • 2. Command Line Interface (CLI)
• CLI Reference
  • Commands Overview
  • Detailed Command Specifications

Architecture & Design

• System Architecture
  • 1. Decoupled Pipeline Design
  • 2. The 7 Canonical Entities
  • 3. Schema Evolution Signatures
• Cross-Quarter Case Deduplication Engine
  • 1. Deduplication Principles
  • 2. 2004–2006 Empirical Cohort Benchmark

Guides & Schema Specifications

• 2004Q1 Dataset Integration & Usage Guide
  • 📦 Package Summary & Baseline
  • 🚀 Quickstart: Consuming via Python SDK
  • 🧬 Data Generation Recipe for Downstream Safety Analytics
  • 📖 Embedded Reference & Dimension Tables
  • 🛠 Rebuilding or Verifying the Package
• Official FDA FAERS Data Structure & Technical Reference Guide
  • 1. Overview of the 7 Core FAERS Tables
  • 2. Exhaustive Data Dictionaries by Table
  • 3. Historical Schema Evolutions & Gotchas
• FDA FAERS/AEMS Historical Schema Audit & Evolution Guide (2004 – 2026+)
  • 1. Dual Historical Timelines
  • 2. Table-by-Table Verified Historical Analysis
  • 3. Transformation & Evolution Categories (8-Category Formal System)
• FAERS/AEMS Coded Values & Code-List History Reference
  • 1. Drug Role Codes (ROLE_COD / role_code)
  • 2. Patient Outcome / Seriousness Codes (OUTC_COD / outcome_code)
  • 3. Patient Sex (SEX / GNDR_COD / reported_sex_code)
  • 4. Age Units (AGE_COD / reported_age_code) & Normalization Multipliers
  • 5. Weight Units (WT_COD / reported_weight_code) & Normalization Multipliers
  • 6. Reporter Occupation (OCCP_COD / reporter_occupation_code)
  • 7. Report Sources (RPSR_COD / report_source_code)
  • 8. Dechallenge / Rechallenge Codes (DECHAL / RECHAL)
• Phase 0 Empirical Schema Fingerprint Comparison Report
  • 1. Verified Release Signatures & Fingerprints
  • 2. Table-by-Table Column Evolution

API Reference

• Python API Reference
  • SDK Client
  • Longitudinal & Deduplication Engine
  • Package Builder
  • Distribution Subsystem

Community & Contributing

• Contributing to raxtik-etl
  • 🛠️ Development Setup
  • 🧪 Testing
  • 🧹 Code Quality & Formatting
  • 🔍 Type Hint Checking
  • 📚 Building Documentation
  • 🤖 All Contributors Bot
• Contributors
• Changelog
  • [0.1.0.dev0] - 2026-10-25

---

🌟 Core Features

• Decoupled Architecture: Source-agnostic canonical data engine separated from source-specific connectors (raxtik-source-faers).

• Signature Detection (S1 → S2 → S3): Fail-closed schema evolution tracking legacy ASCII (2004–2012), migration era (2012–2014), and modern FAERS/AEMS (2014–present).

• Dual Presentation Packaging: Produces self-describing standalone SQLite, DuckDB, and zero-copy Apache Parquet formats.

• Cross-Quarter Case Deduplication Engine: Deterministic longitudinal case resolution resolving follow-up submissions across quarters and years while preserving raw physical provenance.

• Zero-Copy Python SDK: Query canonical cases, suspect drugs, reactions, and outcomes directly as Polars DataFrames.

• Hugging Face Hub Remote Distribution: Instant public streaming download of precompiled packages without requiring local compilation or API tokens.

---

🚀 Quick Installation

Requires Python >= 3.11.

# Core package
pip install raxtik-etl

# With developer & Hugging Face upload tooling
pip install "raxtik-etl[dev]"

# With documentation tooling
pip install "raxtik-etl[docs]"

---

📄 License

• Source Code: GNU Affero General Public License v3 or later (AGPL-3.0-or-later).

• Raw FDA Data: United States Federal Government Work / Public Domain (17 U.S.C. § 105).

• Derived Safety Packages: Open Data Commons Attribution License (ODC-By 1.0).