Bio Variant Calling Structural Variant Calling

Version Compatibility

Reference examples tested with: bcftools 1.19+, samtools 1.19+

Before using code patterns, verify installed versions match. If versions differ:

• CLI: <tool> --version then <tool> --help to confirm flags

If code throws ImportError, AttributeError, or TypeError, introspect the installed package and adapt the example to match the actual API rather than retrying.

Structural Variant Calling (Short Reads)

"Call structural variants from my WGS data" → Detect large genomic rearrangements (deletions, insertions, inversions, duplications, translocations) using split-read and discordant-pair evidence.

• CLI: configManta.py (Manta), delly call, lumpyexpress/smoove call

Manta (Recommended)

# Configure Manta run (creates runWorkflow.py)
configManta.py \
    --bam sample.bam \
    --referenceFasta reference.fa \
    --runDir manta_run

# Execute
manta_run/runWorkflow.py -j 8

# Output: manta_run/results/variants/
# - diploidSV.vcf.gz (germline SVs)
# - candidateSV.vcf.gz (all candidates)
# - candidateSmallIndels.vcf.gz (small indels)

Manta Tumor-Normal Mode

# Somatic SV calling
configManta.py \
    --tumorBam tumor.bam \
    --normalBam normal.bam \
    --referenceFasta reference.fa \
    --runDir manta_somatic

manta_somatic/runWorkflow.py -j 8

# Output includes:
# - somaticSV.vcf.gz (somatic SVs)
# - diploidSV.vcf.gz (germline SVs)

Manta Options

# WES mode (for exome data)
configManta.py \
    --bam sample.bam \
    --referenceFasta reference.fa \
    --exome \                          # Use exome settings
    --callRegions regions.bed.gz \     # Restrict to regions
    --runDir manta_exome

# RNA-seq mode
configManta.py \
    --bam rnaseq.bam \
    --referenceFasta reference.fa \
    --rna \                            # RNA-seq mode
    --runDir manta_rna

Delly

# Call SVs
delly call \
    -g reference.fa \
    -o sv_calls.bcf \
    sample.bam

# Convert to VCF
bcftools view sv_calls.bcf > sv_calls.vcf

# Multiple samples (joint calling)
delly call \
    -g reference.fa \
    -o joint_svs.bcf \
    sample1.bam sample2.bam sample3.bam

Delly Somatic Mode

# Call with tumor-normal
delly call \
    -g reference.fa \
    -o svs.bcf \
    tumor.bam normal.bam

# Create sample file
echo -e "tumor\ttumor\nnormal\tcontrol" > samples.tsv

# Filter for somatic
delly filter \
    -f somatic \
    -o somatic_svs.bcf \
    -s samples.tsv \
    svs.bcf

Delly SV Types

# Call specific SV type
delly call -t DEL -g ref.fa -o deletions.bcf sample.bam
delly call -t DUP -g ref.fa -o duplications.bcf sample.bam
delly call -t INV -g ref.fa -o inversions.bcf sample.bam
delly call -t BND -g ref.fa -o translocations.bcf sample.bam
delly call -t INS -g ref.fa -o insertions.bcf sample.bam

LUMPY

# Extract split reads and discordant pairs
samtools view -b -F 1294 sample.bam > discordant.bam
samtools view -h sample.bam | \
    /path/to/lumpy-sv/scripts/extractSplitReads_BwaMem -i stdin | \
    samtools view -Sb - > splitters.bam

# Run LUMPY
lumpyexpress \
    -B sample.bam \
    -S splitters.bam \
    -D discordant.bam \
    -o lumpy_svs.vcf

Smoove (LUMPY Wrapper)

# Simplified LUMPY workflow
smoove call \
    --name sample \
    --fasta reference.fa \
    --outdir smoove_output \
    -p 8 \
    sample.bam

# Output: smoove_output/sample-smoove.genotyped.vcf.gz

Merge Multiple Callers

Goal: Increase confidence in SV calls by requiring support from multiple callers.

Approach: Run 2-3 callers independently, then merge callsets with SURVIVOR requiring agreement on breakpoint proximity and SV type.

# Use SURVIVOR to merge callsets
# Create file listing VCFs
ls manta_svs.vcf delly_svs.vcf lumpy_svs.vcf > vcf_list.txt

# Merge with parameters
SURVIVOR merge vcf_list.txt 1000 2 1 1 0 50 merged_svs.vcf

# Parameters: max_dist min_callers type_agree strand_agree estimate_dist min_size

Filter SV Calls

# Filter by quality
bcftools view -i 'QUAL >= 20' svs.vcf > svs.filtered.vcf

# Filter by size
bcftools view -i 'ABS(SVLEN) >= 50' svs.vcf > svs.min50.vcf

# Filter by SV type
bcftools view -i 'SVTYPE="DEL"' svs.vcf > deletions.vcf
bcftools view -i 'SVTYPE="INS"' svs.vcf > insertions.vcf
bcftools view -i 'SVTYPE="INV"' svs.vcf > inversions.vcf
bcftools view -i 'SVTYPE="DUP"' svs.vcf > duplications.vcf
bcftools view -i 'SVTYPE="BND"' svs.vcf > translocations.vcf

# Keep only PASS
bcftools view -f PASS svs.vcf > svs.pass.vcf

Annotate SVs

# AnnotSV annotation
AnnotSV \
    -SVinputFile svs.vcf \
    -genomeBuild GRCh38 \
    -outputFile annotated_svs

# Output includes: genes, DGV, gnomAD-SV, ClinVar

SV Types

Comparison: Manta vs Delly vs LUMPY

Coverage Guidelines

Long-Read SV Callers

For long-read data (ONT/PacBio HiFi), use specialized callers with higher sensitivity:

See long-read-sequencing/structural-variants for long-read SV workflows.

Related Skills

• long-read-sequencing/structural-variants - Long-read SV calling
• copy-number/cnvkit-analysis - Copy number variants
• variant-calling/filtering-best-practices - Filter VCF files
• alignment-files/alignment-filtering - Prepare BAM files