strange rules from VCF file

Question

[wendy1988816]

Dear all, I divided snps in VCF file into four parts by "QD",then devided each part by "AF" into three groups, and calculate average "GQ" for each groups.

I found some strange rules from my VCF file.

Blockquote

1. Most of snps which had high GQ, they also had high AF.
2. Most of snps which had high GQ, they had low QD.

Does it make sense?

Thanks for your help.

[Brian Bushnell]

GATK's "phred-scaled quality scores" are completely broken. It routinely gives scores of 40000+, meaning there is a 1/1000....0000 (four thousand zeros) of the call being incorrect. The chance of you actually not existing is greater than the chance of their claim being wrong, which is crazy.

So, I recommend not using GATK, which fails at math. But it would be very helpful if you could post the actual commands you used to arrive at the result you are displaying.