Quantcast
Channel: haplotypecaller — GATK-Forum
Viewing all articles
Browse latest Browse all 1335

HaplotyeCaller - non-variant block records in gVCF

$
0
0

Hi,

I have generated a gVCF for an exome (with non-variant block records) from a BAM file belonging to the 1000Genomes data.
I am using GATK tools version 3.5-0-g36282e4 and I have run the HaplotypeCaller as follows:

time java -jar $gatk_dir/GenomeAnalysisTK.jar \
-T HaplotypeCaller \
-R $reference \
-I $bamfile \
-ploidy 2 \
-stand_call_conf 20 \
-stand_emit_conf 10 \
-ERC GVCF \
-o output.g.vcf.gz

Within the purpose of the analysis I am performing, from this gVCF I need to be able to know whether the positions are no-called, homozygous reference, variant sites or if the positions were not targeted in the exome sequencing.

However, with the gVCF file I obtained I am not able to do it because there are only variant site records or non-variant block records where the GT tag is always "0/0".

So I have few questions regarding the non-variant block records:

  1. Why the output file does not contain any no-call ("./.") record?

  2. Shouldn't regions where there are no reads have the tag GT equal to "./." instead of "0/0"?

  3. How can regions without reads (not targeted) be distinguished from regions with reads that were not called?

  4. When looking at the bam file with IGV, non-variant blocks displayed in gVCF contain regions with reads. What is the explanation for such behaviour?

Thank you for your attention,

Sofia


Viewing all articles
Browse latest Browse all 1335

Trending Articles