[BioC] ensemblVEP: passing on coverage / frequency information

[Valerie Obenchain]

Perfect. Thanks for sending.

There were a couple of things going on. In the case of no INFO, 
readVcf() was storing a DataFrame of 0 rows and 0 columns and writeVcf() 
was writing out a vector of '.' the length of ncol(info(vcf)). This 
resulted in no '.' space holders in the newly written vcf.

readVcf() now creates a DataFrame with N rows (number of records) and 0 
columns when no INFO are present. This new behavior would allow 
writeVcf() to get the correct length from ncol(info(vcf)). Instead, to 
ensure backwards compatibility, I'm using the number of records to 
determine the length of '.' to write out.

Patched versions are 1.7.6 in devel and 1.6.2 in release. Let me know if 
you run into problems. Sorry I didn't identify this last week.

Valerie



On 04/15/2013 09:00 AM, Thomas Sandmann wrote:
> Hi Valerie,
>
> thanks a lot for looking following up on this issue.
> Yes, I generated the input for ensemblVEP with a call to writeVcf.
>
> Here's my workflow:
>
> 1. Variant calling with the '/callVariants/' function from the
> '/VariantTools/' package: returns the*'variants.gr <http://variants.gr>'
> GRanges object*
> 2. Generate a Vcf object from the 'variant' GRanges with a call to
> '/variantGR2Vcf'/: returns *variants.vcf VCF object*
> 3. write Vcf object to disk as a vcf file with '/writeVcf/': generates
> variants.vcf file
>
> You can find the variants.gr <http://variants.gr> and variants.vcf R
> objects here:
> https://dl.dropboxusercontent.com/u/126180/variants.object.Rdata
> and the exported vcf file from the same data here:
> https://dl.dropboxusercontent.com/u/126180/variants.vcf
>
> Thanks a lot !
> Thomas