Homozygous GT value while IGV shows otherwise about deepvariant HOT 9 CLOSED

from deepvariant.

Hi @asalimih ,
For WES, DeepVariant runs a realignment step.

You can read this section about how realignment works: https://github.com/google/deepvariant/blob/r1.6.1/docs/FAQ.md#what-is-the-realigner-and-how-does-it-work

To understand your cases better, it'll be useful to generate the realigned reads in the region, which might explain why DeepVariant makes the call differently.

You can take a look at #691 (comment) for a more detailed example.

Once you generate an example, please share the realigned reads in an IGV and we can check what it looks like? Thank you!

from deepvariant.

Hi @asalimih , For WES, DeepVariant runs a realignment step.

You can read this section about how realignment works: https://github.com/google/deepvariant/blob/r1.6.1/docs/FAQ.md#what-is-the-realigner-and-how-does-it-work

To understand your cases better, it'll be useful to generate the realigned reads in the region, which might explain why DeepVariant makes the call differently.

You can take a look at #691 (comment) for a more detailed example.

Once you generate an example, please share the realigned reads in an IGV and we can check what it looks like? Thank you!

@pichuan
Thanks for the reply. here is the realigned bam in IGV for chr1-1668449-A-G:

The read counts matches what the AD field says in the vcf. however the GT value is 1/1 !

The weird thing is another variant just 76 bp to the left has the correct genotype in the vcf file:

chr1    1668373 .       C       T       6.4     PASS    .       GT:GQ:DP:AD:VAF:PL      0/1:4:139:84,55:0.395683:4,0,5

from deepvariant.

Hi @asalimih ,

can you downsample your bam for this specific variant and run it again?

samtools view -s 0.5 -b -@16 YOU_BAM.bam > YOUR_BAM.downsampled.bam

And run it one more time to see if this specifically has a different genotype?

from deepvariant.

Hi @asalimih ,

can you downsample your bam for this specific variant and run it again?

samtools view -s 0.5 -b -@16 YOU_BAM.bam > YOUR_BAM.downsampled.bam

And run it one more time to see if this specifically has a different genotype?

sure, no change in GT (using v1.6.1):

chr1    1668449 .       A       G       42.1    PASS    .       GT:GQ:DP:AD:VAF:PL      1/1:27:77:36,41:0.532468:42,26,0

I tried with samtools view -s 0.25 too:

chr1    1668449 .       A       G       35.5    PASS    .       GT:GQ:DP:AD:VAF:PL      1/1:16:44:22,22:0.5:35,16,0

from deepvariant.

@kishwarshafin @pichuan
I can provide the bam file for some of the variants if you give me an email. I would really appreciate any help

from deepvariant.

Hi @asalimih ,
You can send to [email protected]. Please start with a small BAM file and the command where you can reproduce the call on that BAM file.

from deepvariant.

Thank you @pichuan, I sent the bam file and commands to your email.

from deepvariant.

I'll close this bug for now. Later on I can plan to post a summarized update.

from deepvariant.