vartovcf

Convert variants from VarDict/VarDictJava into VCF v4.2 format.

Install with the Conda or Mamba package manager after setting your Bioconda channels:

❯ mamba install vartovcf

Features

• Unlike the Perl script bundled with VarDict, this tool streams record-by-record
• This tool is kept lean on purpose and is solely responsible for fast format conversion
• The output is compliant with the VCF v4.2 and v4.3 specifications
• Output VCF records are unsorted and a call to bcftools sort is recommended
• At this time, only tumor-only mode (var2vcf_valid.pl) is supported

Example Usage

Replace to call to var2vcf_valid.pl with vartovcf in a typical VarDictJava stream like:

❯ vardict-java \
    -b input.bam \
    -G hg38.fa \
    -N dna00001 \
    -c1 -S2 -E3 -g4 -f0.05 \
    --fisher \
    calling-intervals.bed \
  | vartovcf --reference hg38.fa --sample dna00001 \
  | bcftools sort -Oz > variants.vcf.gz

Benchmarks

❯ vartovcf --reference hs38DH.fa --sample dna00001 < test.var > /dev/null
[2025-10-21T01:16:49Z INFO  vartovcf] Input stream: STDIN
[2025-10-21T01:16:49Z INFO  vartovcf] Output stream: STDOUT
[2025-10-21T01:16:49Z INFO  proglog] [main] Processed 60181 variant records

❯ hyperfine --warmup 5 'vartovcf -r hs38DH.fa -s dna00001 < test.var > /dev/null'
Benchmark #1: vartovcf -r /references/hs38DH.fa -s dna00001 < test.var > /dev/null
  Time (mean ± σ):     174.7 ms ±   1.5 ms    [User: 165.7 ms, System: 8.0 ms]
  Range (min … max):   173.2 ms … 178.2 ms    16 runs

❯ hyperfine --warmup 5 'var2vcf_valid.pl -N dna00001 -f 0.0 -E < test.var > /dev/null'
Benchmark #1: var2vcf_valid.pl -N dna00001 -f 0.0 -E < test.var > /dev/null
  Time (mean ± σ):     359.4 ms ±   2.5 ms    [User: 329.2 ms, System: 25.8 ms]
  Range (min … max):   356.1 ms … 363.6 ms    10 runs