Merge VCF file into a reference

This python script is designed to merge VCF files with that of a reference. This is useful for pre-processing multiple VCF files that may have been genotyped/sequenced on different platforms. This can result in distinct annotation of variants, for example the strand may be flipped.

Merging the VCF files involves:

• removing variants that are not present in the reference
• removing SNVs that are strand ambiguous
• flipping the strand
• flipping the reference and alternate alleles

The stand and alleles must be unambiguous in order for them to be flipped. For example, an A/T SNP is ambiguous. A/T can specify reference/alternate or forward/reverse strand.

Usage

The usage is as follows:

python MergeVcf2Panel.py \
           --ref 1KG/ALL.chr1_GRCh38_sites.20170504.vcf.gz \
           --vcfin my_geno.vcf \
           --vcfout my_geno.chr1.1KG.vcf

This takes the reference file from 1KG/ALL.chr1_GRCh38_sites.20170504.vcf.gz , the first chromosome, and merges the variants from my_geno.vcf into my_geno.chr1.1KG.vcf.

Reference

The reference file is a VCF where only the first 5 columns are used (corresponding to location, ID, and alleles). A popular reference panel is the 1,000 Genomes. The download_1KG.sh script downloads the variant sites in VCF format into the folder 1KG. These are variants lifted over into GRCh38 coordinates. These files can be used as the reference input for the MergeVcf2Panel.py script. To merge the chromosomal files and remove the columns after the fifth, run get_htslib.sh to get the bgip tools and then run format_1KG.sh. This places all the variants into 1KG/ALL.GRCh38_sites.20170504.vcf, which can be used with the python script.

If you want to run all chromosomes in a single serial run, merge VCF with concat_chrm_1KG.sh.