--normalize_type creates invalid VCF for translocations

[wdecoster]

Hi,

I think --normalize_type on a BND representing a translocation creates an invalid VCF, as it fills in the END INFO field with a value that is on a different chromosome, which may end up being END < POS. While that seems, according to the VCF4.2 specifications, not explicitly forbidden, it does lead to errors with bcftools and has been clarified in the VCF4.3 specifications "End position on CHROM ".

See for example also https://github.com/samtools/hts-specs/pull/436

Cheers, Wouter

[mkirsche]

Hi Wouter,

Thanks for bringing this to my attention! I will plan on updating this in the next Jasmine release to conform with these standards.

Thank you!! Melanie

[wdecoster]

Hi Melanie,

That's great. Thank you for the very useful tool. I think this also comes from jasmine, but I think you also added the 'STRANDS' and 'AF' info fields, but these are not yet defined in the VCF header?

Cheers, Wouter