gnomad exome issue

[matthdsm]

gnomad exome issue

[pfpjs]

Hi @matthdsm, @chapmanb, @naumenko-sa,

I'm getting the same truncated files as @matthdsm, but only for GRCh37 and hg38, while hg19 completed successfully.

Looking at the ggd recipes, hg19 uses wget to fetch the gnomad VCFs (gnomad_exome and gnomad_genome), while GRCh37 and hg38 both use the URL directly as an argument to vt decompose.

I've tried rerunning the GRCh37 and hg38 recipes three times now, and they always fail at roughly (but not exactly) the same spot - leaving me with a ~400Mb gnomad_exome.vcf.gz file and a ~340Mb gnomad_genome.vcf.gz file.

For reference, here are the sizes of the hg19 version of these two files:

gnomad_exome.vcf.gz: 5391108579 bytes
gnomad_genome.vcf.gz: 26405145616 bytes

-- Paulo

[naumenko-sa]

Hi Matthias @matthdsm and Paulo @pfpjs !

Thanks for reporting the issue and pointing out how to fix it!
I indeed see the same effect on my system.

I pushed a fix in my fork https://github.com/naumenko-sa/cloudbiolinux
and created a pull request for Brad @chapmanb.

Please let me know if this resolved the issue!
Sergey

[chapmanb]

Thanks Sergey -- this got pulled and is available now. Thank you all for testing and reporting and let us know if you hit any other problems.

[matthdsm]

Hi guys,

I was doing some digging and turns out we're making things too hard on ourselves. The annotation works just as well without decomposing/normalizing the gnomad vcf. What are your opinions on this?

Cheers
M

[pfpjs]

Hi @naumenko-sa,

GRCh37 gnomad_exome.vcf.gz completed sucessfully, thank you!

However, gnomad_genome.vcf.gz also needs the wget "treatment". Thanks!