Calculate the proportion of missing data per sample from VCF file

[LeonDK]

I have a VCFv4.2 (Sanger-imputation-service genotype data) file and I need to calculate the proportion of missing data per sample.

Each sample is encoded like so:

GT:ADS:DS:GP
0|0:0.25,0.15:0.4:0.6375,0.325,0.0375

I am uncertain as to how 'missing' is identified - Any suggestions?

[dpryan]

Missing data will have ".|." or "./." as the genotype. However, since this is imputed I'm not sure how much there will be in the way of missing values (they should have been largely imputed).

[LeonDK]

Missing data will have ".|." or "./." as the genotype. However, since this is imputed I'm not sure how much there will be in the way of missing values (they should have been largely imputed).

Running the following on the vcf files

for f in `ls -1`; do gunzip -c $f | grep -c '\.|\.'; done

Yields nothing but zeros, so it looks like you are correct.