VCF output is not compatible with current version of PySAM

[davidlougheed]

Hello,

Thanks for your work on the Straglr tool.

I am currently attempting to benchmark STR genotyping software, but due to Straglr's VCF 4.5 output (which is not yet fully supported by htslib, although partial support landed in 1.22), PySAM (which wraps htslib 1.21) seems to struggle with these VCFs and I get errors that seem to relate to END tags (see my discussion on the Truvari tool issues page here: ACEnglish/truvari#290).

Right now, it seems PySAM/htslib 1.21 need these END tags to calculate SV length in some cases. As a result, I cannot benchmark Straglr output.

Do you think it would be possible to add back in these deprecated END tags to the variant records+header until full VCF 4.5 support lands in these libraries?

[readmanchiu]

Sorry for the late reply. No problem, I will add back in the END tags. But just a reminder that SV lengths of TR expansions cannot be calculated by simply subtracting the start from end position, as they are just coordinates of the loci in the reference genome. The SV length can be looked up from the RB field in the vcf file.

[davidlougheed]

I believe END is more to get the size of the annotated region in the reference genome, rather than the length of the variant allele. Thanks for your response!

[readmanchiu]

INFO END is deprecated in v4.5, but I assume it can be included for backward compatibility
I think END is supposed to be inferred from FORMAT LEN now in v4.5, which I overlooked and haven't included in the vcf output. I'll probably add FORMAT LEN in too for the next release.