Genome Analysis Toolkit

Variant Discovery in High-Throughput Sequencing Data

GATK process banner

Need Help?

Search our documentation

Community Forum

Hi, How can we help?

Developed in the Data Sciences Platform at the Broad Institute, the toolkit offers a wide variety of tools with a primary focus on variant discovery and genotyping. Its powerful processing engine and high-performance computing features make it capable of taking on projects of any size. Learn more

Allele Depth (AD) is lower than expected Follow


  • Avatar
    Shashank Katiyar

    Thanks, this is a well explained article about a much needed missing information. I wonder if is there any parameter to change the 'difference between the Phred scaled likelihoods' from .2 to the desired number?

    Comment actions Permalink
  • Avatar

    It'll be helpful to report read coverage after filtering for each sample in the output of GenotypeGVCFs

    Comment actions Permalink
  • Avatar
    Burak Alver

    Thank you for this explanation. This clarifies some questions. However, I continue to be confused.

    The "The explanation: uninformative reads" section of the article suggests that there should sometimes be two different DP values.
    1. DP-info: how many reads cover the position.
    2. DP-format: how many reads are informative to distinguish two different alleles.
    Then, the second type of reads should also be reported in AD, and there sum(AD)=DP-format.

    All this sounds reasonable. But the vcf line example does not agree with this explanation. There, both DP values are 10, and sum(AD)=0.

    Is there a difference between DP-info and DP-format?

    Comment actions Permalink
  • Avatar
    Sinem Selvi

    Based on the information, following variant is homozygous based on FORMAT DP or INFO DP?

    When i check on IGV the variant looks heterozygous with following read counts:
    chr1:228346358;total count 150;ref G count 150;ins count 1.
    chr1:228346359;total count 157;ref C count 157;ins count 80.

    chr1    228346358       .       G       GCCGCCGCGGCCCCCCGGCCT   3040.03 .       AC=2;AF=1.00;AN=2;DP=149;ExcessHet=3.0103;FS=0.000;MLEAC=2;MLEAF=1.00;MQ=60.00;QD=30.51;SOR=2.105      GT:AD:DP:GQ:PL  1/1:0,72:72:99:3054,203,0

    So how to be sure the variant is hom alt or het?
    Thank you

    Comment actions Permalink
  • Avatar
    Vee Koo


    I have this problem that I have encountered in regards to HaplotypeCaller's -bamout produced realigned BAM and AD according to VCF. The VCF has been done with 63 samples through the HaplotypeCaller -> CombineGVCF -> GenotypeGVCF pipeline. I have produced IGV pictures from these HaplotypeCaller created BAMs. Below is all samples' FORMAT field in the order they are in the IGV picture (there are many sites like this in other IGV pictures also):




    The position in question in the VCF and IGV pictures is Contig31:12504. So the last sample is of interest as it shows AD of 51,0 and GQ of 99 and despite that it doesn't show the reads. Have you any idea why? -bamout option in HC won't produce uninformative reads but these shouldn't be uninformative, right?

    Comment actions Permalink

Please sign in to leave a comment.

Powered by Zendesk