nf-core/scge Pipeline

The Problem

The NIH Common Fund’s Somatic Cell Genome Editing (SCGE) program required a reproducible, scalable tumor-normal variant calling pipeline capable of handling petabyte-scale sequencing datasets. Existing workflows lacked the cloud elasticity and hardware acceleration needed for high-depth clinical-grade analysis under tight nf-core reproducibility standards.

The Solution

Co-developed a cloud-native Nextflow DSL2 pipeline adhering to full nf-core community conventions — enabling automated, reproducible somatic variant calling at NIH scale.

Architecture

• Nextflow DSL2 for modular, composable workflow components with strict versioning
• Illumina DRAGEN hardware acceleration on AWS Batch for high-depth tumor-normal variant calling — dramatically reducing compute time vs. software-only approaches
• nf-core conventions for reproducibility, automated testing, CI/CD, and community contribution standards
• Docker/Singularity dual containerization for portability across HPC and cloud environments
• AWS Batch elastic compute scaling — spin up and tear down resources automatically based on input queue depth

Key Results

• Minimized runtime for high-depth tumor-normal variant calling through DRAGEN acceleration
• Full compliance with nf-core community standards, enabling peer review and external validation
• Handles petabyte-scale sequencing datasets with automated ingestion and provenance tracking
• Production-deployed for active NIH SCGE program research cohorts