Cloud-based, Evidence-based Rigorous Functional Assay Calibration
A bioinformatics pipeline for validating functional assay predictions against clinical observational data using WDL workflows on Terra or locally.
- New to CERFAC? Read CERFAC_documentation.md for the full step-by-step guide
- Setting up locally? See CROMWELL_SETUP.md for workflow execution
- Want to understand the architecture? Check CLAUDE.md for detailed technical information
CERFAC combines three data sources to calibrate functional assay predictions:
- ClinVar: Clinical variant classification from NCBI
- gnomAD: Population variant frequency data
- Your assay scores: Functional predictions you provide
The pipeline normalizes variants across formats, merges the data, and performs statistical calibration (OR calculations) via Jupyter notebook.
See CERFAC_documentation.md — complete step-by-step guide for cloning a workspace and running workflows.
See CROMWELL_SETUP.md for workflow execution or CLAUDE.md for technical details.
ClinVar Query → ┐
├→ Merge & Normalize → Statistical Analysis
gnomAD Query → ┤ (VRS API) (R Jupyter notebook)
Your Data ────→ ┘
Three Docker images handle different stages:
cerfac-clinvar— NCBI EDirect extractioncerfac-merge— Data merging and API callscerfac-gnomad— Population frequency lookup via Hail
| Document | Purpose |
|---|---|
| CERFAC_documentation.md | User guide for Terra |
| CLAUDE.md | Developer reference |
| CROMWELL_SETUP.md | Local workflow execution |
| docker/README.md | Docker image building |
- Variant normalization: Handles HGVS coding, HGVS genomic, and gnomAD VCF formats
- Scalable: Runs on Terra cloud or local machine via Cromwell
- Reproducible: WDL workflows with containerized dependencies
- Flexible: Accepts TSV, CSV, or TXT input files
- Docker (for containerized task execution)
- Java Runtime (for Cromwell workflow engine)
- Sufficient disk space (~2 GB for Docker images)
Refer to the documentation files above or check CLAUDE.md for technical details.