nf-caddsv

Overview

nf-caddsv is a Nextflow DSL2 port of CADD-SV, a framework to score the effect of structural variants in the human genome.

CADD-SV retrieves a wide set of annotations in the range and vicinity of a structural variant (SV), computes summary statistics, and uses trained machine learning models to differentiate deleterious from neutral variants. The method uses human and chimpanzee derived alleles as proxy-neutral and contrasts them with matched simulated variants as proxy-pathogenic.

Note: This is a Nextflow port of the original Snakemake implementation. All computational logic, models, and annotations are from the original CADD-SV authors. This port aims to provide nf-core compatible infrastructure for easier integration into existing Nextflow pipelines.

Original Publication & Credits

CADD-SV was developed by the Kircher Lab at the Berlin Institute of Health.

Original Implementation:

• Repository: https://github.com/kircherlab/CADD-SV
• Web Service: https://cadd-sv.bihealth.org/
• Annotations Download: https://kircherlab.bihealth.org/download/CADD-SV/

Citation: If you use this pipeline, please cite the original CADD-SV publication:

Kleinert P, Kircher M. CADD-SV – a framework to score the effects of structural variants in health and disease. Genome Medicine (2022). https://doi.org/10.1186/s13073-022-01039-1

---

Requirements

• Nextflow >= 25.04.7
• Container engine: Docker or Singularity (recommended for HPC)
• Disk space: ~15GB for annotations (downloaded automatically on first run)

Installation

1. Clone the repository

git clone https://github.com/your-username/nf-caddsv.git
cd nf-caddsv

2. Install Nextflow (if not already installed)

# Using conda
conda install -c bioconda nextflow

# Or using curl
curl -s https://get.nextflow.io | bash

3. Test the installation

nextflow run main.nf --help

Quick Start

Basic Usage

nextflow run main.nf \
    --input /path/to/your_svs.bed \
    --outdir ./results \
    -profile singularity

With Pre-downloaded Annotations

If you already have the CADD-SV annotations downloaded:

nextflow run main.nf \
    --input /path/to/your_svs.bed \
    --annotations_dir /path/to/annotations \
    --models_dir /path/to/models \
    --outdir ./results \
    -profile singularity

Using Docker

nextflow run main.nf \
    --input /path/to/your_svs.bed \
    --outdir ./results \
    -profile docker

Input Format

Input must be a BED file with structural variants in GRCh38/hg38 coordinates.

Required Columns

Column	Description	Example
1	Chromosome (with chr prefix)	chr1
2	Start position (0-based)	821713
3	End position	3928354
4	SV Type	DEL, INS, or DUP

Example Input

chr1    821713    3928354   DUP
chr1    1000000   1010000   DEL
chr2    5000000   5000500   INS
chr7    55000000  55100000  DEL
chr17   7500000   7600000   DUP

Supported SV Types

Type	Description
DEL	Deletion
INS	Insertion
DUP	Duplication

Note: The input file should be coordinate-sorted. The pipeline will sort it automatically, but pre-sorted input is recommended.

Parameters

Required Parameters

Parameter	Description
--input	Path to input BED file with SVs

Optional Parameters

Parameter	Default	Description
--outdir	./results	Output directory
--annotations_dir	null	Path to pre-downloaded annotations
--models_dir	null	Path to pre-downloaded models

Output

The pipeline produces the following outputs in the specified --outdir:

results/
├── scored/
│   └── {sample}_score.bed          # Final scored SVs
├── annotated/
│   ├── {sample}_annotated.tsv      # Main span annotations
│   ├── {sample}_100bpup_annotated.tsv   # Upstream flank annotations
│   └── {sample}_100bpdown_annotated.tsv # Downstream flank annotations
├── caddsv_references/              # Downloaded references (if applicable)
│   ├── annotations/
│   └── models/
└── pipeline_info/
    └── versions.yml

Pipeline Steps

┌─────────────────────────────────────────────────────────────────┐
│                        nf-caddsv Pipeline                       │
└─────────────────────────────────────────────────────────────────┘
                              │
                              ▼
              ┌───────────────────────────────┐
              │  CADDSV_PREPARE_REFERENCES    │  (optional)
              │  Download annotations/models  │
              └───────────────────────────────┘
                              │
                              ▼
              ┌───────────────────────────────┐
              │       CADDSV_PREPBED          │
              │  Sort & prepare BED files     │
              │  - Main span                  │
              │  - 100bp upstream flank       │
              │  - 100bp downstream flank     │
              └───────────────────────────────┘
                              │
          ┌───────────────────┼───────────────────┐
          ▼                   ▼                   ▼
┌─────────────────┐ ┌─────────────────┐ ┌─────────────────┐
│ ANNOTATE_MAIN   │ │ ANNOTATE_UP     │ │ ANNOTATE_DOWN   │
│ Main SV span    │ │ Upstream 100bp  │ │ Downstream 100bp│
└─────────────────┘ └─────────────────┘ └─────────────────┘
          │                   │                   │
          └───────────────────┼───────────────────┘
                              │
                              ▼
              ┌───────────────────────────────┐
              │        CADDSV_SCORE           │
              │  Random Forest prediction     │
              │  PHRED score calculation      │
              └───────────────────────────────┘
                              │
                              ▼
              ┌───────────────────────────────┐
              │         Final Output          │
              │    Scored & sorted BED        │
              └───────────────────────────────┘

Converting Input Formats

VCF to BED

Use SURVIVOR to convert VCF to BED:

# Install SURVIVOR
conda install -c bioconda survivor

# Convert VCF to BED
SURVIVOR vcftobed input.vcf 0 -1 output.bed

# Extract required columns (chr, start, end, type)
awk 'BEGIN{OFS="\t"} {print $1, $2, $6, $11}' output.bed > final_input.bed

Lift-over from hg19 to GRCh38

# Install liftOver
conda install -c bioconda ucsc-liftover

# Download chain file
wget https://hgdownload.cse.ucsc.edu/goldenpath/hg19/liftOver/hg19ToHg38.over.chain.gz

# Perform lift-over
liftOver input_hg19.bed hg19ToHg38.over.chain.gz output_hg38.bed unmapped.bed

Pre-downloading Annotations

For offline use or to avoid repeated downloads:

# Download annotations (~15GB)
wget https://kircherlab.bihealth.org/download/CADD-SV/v1.1/dependencies.tar.gz
tar -xzf dependencies.tar.gz

# Clone models from GitHub
git clone https://github.com/kircherlab/CADD-SV.git
cp -r CADD-SV/models .

# Run with pre-downloaded references
nextflow run main.nf \
    --input your_svs.bed \
    --annotations_dir ./annotations \
    --models_dir ./models \
    --skip_prepare_references \
    -profile singularity

Testing

Run the test dataset:

nextflow run main.nf \
    --input test/data/test_svs.bed \
    --outdir test/results \
    -profile singularity,test

Contributing

Contributions are welcome! Please:

1. Fork the repository
2. Create a feature branch
3. Submit a pull request

License

The original CADD-SV is subject to its own licensing terms. Please refer to the original repository for details.

Support

• Original CADD-SV issues: https://github.com/kircherlab/CADD-SV/issues
• This Nextflow port issues: Open an issue

Acknowledgments

• Original CADD-SV authors: Philip Kleinert and Martin Kircher at the Kircher Lab, Berlin Institute of Health
• nf-core community: For workflow standards and best practices