I observed an inconsistency between these 2 files generated for 2 samples in 2 different runs of GCNV workflow using GATK 18.104.22.168 docker image. I am not sure if my problem is similar to this one but certainly it feels so.
For one of the samples we had a MLPA confirmed heterozygous deletion yet normalized copy ratios txt file shows 1 copy for those intervals but intervals.vcf shows homref therefore CNV call is missed in the segments. This sample was run against a verified normal set in case mode.
A second sample worked in a cohort setting and has a hom deletion with obvious breakpoints in the bam file. copy ratio txt file shows 0 copy for the deleted exon and readcount file shows only 1 read at that location. However intervals vcf shows homref again yet another missed CNV event in the segments file.
First case was rerun in a seperate cohort run and both txt file and intervals vcf showed 1 copy therefore segments vcf showed the proper CNV event called.
I am trying the second case this time using a seperate normal model to see if copy ratio is detected properly this time but I am concerned that something is not right with intervals.vcf creation and inconsistently calls known deletions as homref.
I will post more once I had the data from the second case retrial.
Can you help with this?
Please sign in to leave a comment.