Investigate VCF files from the GDC

[agduncan94]

Investigate whether VCF files downloaded from the GDC portal work with the vcf track type.

Note: VCF tracks require the VCF file to be gzipped and an index file to exist.

There are two parts to this. 1) Download a VCF file and generate the index file locally (may need to run bgzip). Using these two files, create a VCF track. Does this work? 2) Find a VCF file with an existing index. Set the URLtemplate path of a VCF track (and index path) to these locations (requires using the authentication token)

Use samtools tabix to create the index file.

Post your results to the investigation as a comment here.

[agduncan94]

The GDC API has the concept of related files, which might be relevant for finding associated index files. https://docs.gdc.cancer.gov/API/Users_Guide/Downloading_Files/#related-files

[agduncan94]

Part 1 - download vcf and generate index locally

• looked at 6abc7d24-74d1-4e62-975c-753aec620201
• No index file, used “tabix -p vcf file.vcf.gz” to generate
• download from UI uses folder structure, would download via API just get the file?
• if vcf file is not .gz, must run “bgzip -c file.vcf > file.vcf.gz” before creating index

I used the following configuration [tracks.vcf] urlTemplate=6abc7d24-74d1-4e62-975c-753aec620201.vep.vcf.gz storeClass=JBrowse/Store/SeqFeature/VCFTabix type=JBrowse/View/Track/HTMLVariants

This assumed the index file was named 6abc7d24-74d1-4e62-975c-753aec620201.vep.vcf.gz.tbi and in the same dir (jbrowse/data) as the vcf file.

[GFJHogue]

Relevant points from #91 investigation:

• 38% of the VCFs are gzipped AND have an index
• 55% of the VCFs are gzipped AND have NO index
• 7% of the VCFs are NOT gzipped AND have NO index

For the remaining part 2 I'll extend the work in #98 to load remote indexed&gzipped VCFs with a GDC token.