Some recent questions from a user

[kmoad]

Several interesting questions were raised by a user to our support email. They relate to issues that may affect other users, especially regarding vcf files.

Questions

Dear Open cravat team

1. Does your web version of this platform support HGVS annotation of the VCF file ?
   If yes, what is the name of the application module on the web version of this platform ?

2. I have also noted with your web version, that your annotator does not include any "PASS" variants to annotate from a VCF file. Furthermore, for many variant coordinates, the coordinates in the original VCF file do not match your output annotated file.
   Have you been observing this issue ?

3. What is the maximum VCF file size that can be uploaded to be annotated using your web version of the platform ?
   I noted that large files such as over 50 Mb cannot be uploaded.

Our response

Hi User,

You should get HGVS c. and p. notation for all relevant variants. These will be in the "Base Annotation" column group's "cDNA Change" and "Protein Change" columns. The corresponding transcript is in the Transcript column. This will usually be the MANE Select transcript for that gene. You can find c. and p. for all mapped transcripts in the "All Mappings" column.

Regarding the un-annotated variants you're seeing: the current behaviour of our system is to not annotate variants where no sample has the alternate allele. OpenCRAVAT looks at the GT field of the VCF for all present samples, and annotates variants where at least one sample is heterozygous or homozygous for the alternate allele. We don't consider read depth or PASS, just the GT field for the sample(s). This has caused issues for some users, and we're working to change behavior to annotate every variant. In your vcf, the genotype call for the missing variants is "./.". According to vcf spec, that should mean that no call was made for that sample at that allele. Given that most users expect annotation regardless of GT call, we're going to shift to providing it.

You can check the read depth of your variants on the variants tab, within the VCF Info group. The Alternate reads count is also a column in that group.

The reason the input VCF coordinates and output coordinates may not match is an artifact of how we're numbering indels. We are working to fix this and thanks for bringing it to your attention. In the meantime, you can refer to the Extra VCR Annotation column group to identify the original position and allele notation from your input VCF. Look for column names "VCF Position", "VCF Ref Allele", and "VCF Alt Allele".

We limit the size of input files uploaded to the web server to balance server resources among users. It can probably be raised above 50MB though. What would be a good size for you? We could consider raising above 50MB.
With regards to RefSeq transcripts, ideally we would like to add them but we are small team with many ongoing projects so this is not a priority right now.

Best,
Kyle Moad