Best way to approach DELLY BCFs

[lydiayliu]

I know the BCF format is in binary and meant to be storage-saving as compared to VCFs. However, I don't know how to preview a BCF file in the terminal and hence don't know how to use it for anything. My first step is always converting the BCF file into VCF using bcftools, which is not even a module available on the cluster.

I was wondering what the best way to approach BCF outputs of DELLY would be and what the design intention was in having it as the final output of the pipeline. If all users must convert it into VCF before analysis, it would seem natural for the pipeline to include such a step.

[tyamaguchi-ucla]

@lydiayliu bcftools/1.15.1 is available as an env module on the cluster. Please let me know if there is an issue with the module. Here is the usage: bcftools view [options] <in.vcf.gz> and no need to convert the BCF to a new VCF file.

Regarding the pipeline design, it was inherited from the legacy Perl SV pipeline and we're considering using vcf.gz instead of .bcf as you suggested.

[Faizal-Eeman]

Yup, I agree. There is an issue already here - #98. @Faizal-Eeman I believe this change should be relatively easy to add but can you comment on this?

Yes, we can perhaps make this change in 6.0.0-rc.2. I can prioritize it this week, so that the changes from current rc.1 and rc.2 could be added to meta-pipeline together.

[tyamaguchi-ucla]

@lydiayliu Hmm, it seems like there was some miscommunication. I was thinking the other way around as you mentioned. #97

[tyamaguchi-ucla]

@Faizal-Eeman @lydiayliu Also, if I recall, call-gSV has the same issue.

[lydiayliu]

@tyamaguchi-ucla Hmmm now I'm confused as well XD For the record, I vote for standardizing pipeline outputs in the vcf.gz format :) Thanks!

[Faizal-Eeman]

Discussion #97 and Issue #98 are old, and will be updated based on the best approach that's convenient for users.

So, we will go with a standardized vcf.gz output.