What have I done?

VCF comparison script:

• Simultaneously reads and compares two vcf files from the same sample
• Required packages:
  • PyVCF (pip install PyVCF)
  • numpy (pip install numpy)
• Input: two vcf files to be compared (should work for zipped and unzipped)
  • If the sample name is different than the vcf sample name, provide it as a third argument.
• Returns:
  • List of vcf entries that have different DP values
  • Summary of how many records in the vcf have different DP values and different records all together.
  • Histogram of DP difference variance
  • List of different vcf records (record = [CHROM,POS,REF,ALT])
• Runtime: can take up to 5 hours for genome vcfs depending on the size of the vcf.
  • If so the vcfs can be split by chromosome

How to - step by step

Run the script

$ python vcf_compare.py original_vcf.gz development_vcf.py > output_filename.txt

The tool gets the sample name from the filename, it has to be the same sample name as it is in the vcf, if it isn't, provide a sample name as a third argument.

$ python vcf_compare.py original_vcf.gz development_vcf.py NA12878 > output_filename.txt

For big files:

1. Index your vcf files using tabix ()

$ tabix -p vcf your_vcf_file.vcf.gz

2. Create a file listing all chromosomes set it as a bash variable.

$ chromosomes=/path/to/file/chromosomes.txt

3. Split the vcfs by chromosome

$ for chrom in $(cat $chromosomes); do echo ${chrom} splitting; tabix -h NA12878.truth.g.vcf.gz ${chrom} > NA12878.dev.${chrom}.g.vcf; done

4. Optional : recompress

$ gzip *.vcf

5. Run script on individual components

$ for chrom in $(cat $chromosomes); do echo $chrom running; python ${path}/bin/vcf_compare.py NA12878.truth.${chrom}.g.vcf.gz NA12878.dev.${chrom}.g.vcf.gz > NA12878.${chrom}.comparison.txt ; done