Oxford Nanopore Methylation and Structural Variant Analysis Pipeline

Overview

This is a Snakemake workflow for analyzing Oxford Nanopore Technologies (ONT) sequencing data, with a focus on methylation calling and structural variant detection. This pipeline processes raw ONT reads (POD5 format) through basecalling, alignment, methylation analysis, and various structural variant calling methods.

Features

• Basecalling and QC: Uses Dorado for basecalling with modified base detection, followed by quality assessment with PycoQC
• Alignment: High-quality read mapping with Minimap2
• Methylation Analysis:
  • Methylation pileup with ONT Modkit
  • Visualization with methylartist
• Structural Variant Detection: Multiple SV callers for comprehensive detection
  • Clair3 for SNV/indel calling
  • ClairS for somatic variant detection in tumor-normal pairs
  • Sniffles2 for germline structural variation detection
  • Severus for germline and somatic structural variation detection

Requirements

• Snakemake (>=7.0)
• Conda or Mamba (for environment management)
• For GPU-accelerated basecalling: CUDA-compatible GPU

Installation

1. Clone this repository:

git clone https://github.com/cnio-bu/myeloma-epi-sv.git
cd myeloma-epi-sv

2. Create your config file:

cp config/config.yaml.example config/config.yaml

3. Edit the configuration file to match your environment, data locations, and analysis parameters.

Configuration

The pipeline is configured through the config/config.yaml file, which includes:

• Reference files: Paths to genome reference, annotations, and other required files
• Tool paths: Locations of external tools like dorado and Clair3 models
• Analysis parameters: Quality thresholds and analysis options
• Samples structure: Hierarchical organization of samples with associated metadata
• Resource specifications: Resource allocation for different workflow steps

See config/config.yaml.example for a detailed example of the configuration structure.

Usage

Basic Execution

Run the full pipeline with:

snakemake --use-conda --cores <N>

Execution with Slurm

For cluster environments using Slurm:

snakemake --use-conda --profile slurm

Container-based rules (ClairS)

Some rules run inside containers rather than Conda environments. In particular, the clairs rule runs in the hkubal/clairs Docker/Apptainer image and refers to its inputs and outputs as absolute paths inside the container (e.g. /results/..., /models/..., /logs/...). For those paths to resolve, the corresponding host directories must be bind-mounted into the container at those locations.

Pass the binds through --apptainer-args (use --singularity-args if you run Singularity instead of Apptainer):

snakemake \
    --sdm conda apptainer \
    --apptainer-args "-B /path/on/host/results/:/results/,/path/on/host/clairs_models/:/models/,/path/on/host/logs/:/logs/" \
    --executor slurm

Replace the /path/on/host/... entries with the actual locations of your results/, ClairS model, and logs/ directories. The format of each bind is <host_path>:<container_path>.

Pipeline Steps

1. Basecalling: Convert raw POD5 files to BAM format with Dorado, including modified base detection
2. Quality Control: Generate QC metrics and reports with PycoQC
3. Quality Filtering: Filter reads based on quality score
4. Alignment: Map reads to reference genome with Minimap2
5. Coverage Analysis: Generate coverage statistics with Mosdepth and samtools
6. Methylation Analysis:
   • Extract methylation information with Modkit
   • Visualize methylation patterns with methylartist
7. Variant Calling:
   • SNV/indel detection with Clair3
   • Somatic variant detection with ClairS (for tumor-normal pairs)
   • Germline structural variant detection with Sniffles2
   • Germline/somatic structural variation detection with Severus

Output Structure

The pipeline generates results in a hierarchical directory structure:

results/
├── basecall_dorado/       # Basecalled reads
├── pycoqc/                # Quality control reports
├── minimap2/              # Aligned reads
├── primary/               # Filtered primary alignments
├── mosdepth/              # Coverage statistics
├── modkit/                # Methylation data
├── methylartist/          # Methylation visualizations
├── clair3/                # Small variant calls
├── sniffles/              # Structural variant calls
├── severus/               # Tandem repeat expansions
└── clairs/                # Somatic variants

Testing

Run the included integration test to verify that the pipeline is properly installed:

export SKIP_BASECALLING=true #skip this line to run the GPU-based basecalling step
bash .tests/integration/get_resources.sh
snakemake --sdm conda --cores 1 --directory .tests/integration