search

vgteam / toil-vg

Distributed and cloud computing framework for vg
Apache License 2.0
21 stars 14 forks source link

genotype calling for structral variants #834

Open zqfang opened 3 years ago

zqfang commented 3 years ago

Hi guys, I follow the tutorial command in the wiki page here for structral variants.

when comes to the toil-vg call, the current toil-vg (v1.6.0) did not have these two --chroms and --gams. So I change to comand: 

toil-vg call ./callstore ./gams/test_chr19  \
          --graph ./vg/GRCm38_chr19.vg  \
         --gam ./gams/BTBR_chr19/BTBR_chr19_default.gam  \
          --genotype_vcf ./test.chr6.vcf.gz \
          --realTimeLogging --calling_cores 12 --container None --workDir .

However, the command never work. What should I do? Please help

toil-vg (v1.6.0) vg version ( v1.33.0) toil (v3.24)

zqfang commented 3 years ago

After a lot of anoying testing, I found that added --sample test will make the job run.
But I don't known the why the --alt_path_gam is gone in current toil-vg verison.

adamnovak commented 3 years ago

Hello @zqfang, thanks for reporting this.

We really ought to make sure that the wiki pages stay up to date; we could adopt the txm documentation testing tools that vg is starting to use.

Our most recent work with calling structural variants with vg is at https://github.com/vgteam/giraffe-sv-paper/tree/master/scripts/genotyping but uses a WDL workflow instead of toil-vg.

I don't really know what --alt_path_gam was doing; it might be from when vg call used a different way of enumerating possible alleles? I think @glennhickey might be able to speak to how it is supposed to work now and what the wiki page should say.

glennhickey commented 3 years ago

Yeah, I don't really remember --alt_path_gam. I think it was from a time where xg indexes couldn't store many paths. So in order to get the alt paths (from vg construct -a) through the pipeline (they are required for vg call -v) they came in via xg/GAM pairs. This is no longer necessary as alt paths can be included in the xg with vg index -xL

The calling pipeline has changed a bunch over the years, accumulating cruft in toil-vg. But it's been stable for a while, so I would have hoped the basic toil-vg functionality would be usable.

But toil-vg has become much less necessary for calling now that the vg commands have become much simpler and more efficient. Especially for genotyping: you should be able to do it quickly on a single machine with a very small handful of commands:

https://github.com/vgteam/vg/wiki/SV-genotyping-with-vg https://github.com/vgteam/vg/wiki/Whole-genome-calling-and-genotyping

zqfang commented 3 years ago

Sure, @glennhickey, @adamnovak , Thank you very much for you help. I'll try vg call then