Daniel Cameron

The other interpretation of the specs is that percent-encoding works on any of the 8 reserved characters anywhere in a VCF file. I'm less keen on this interpretation as it's...

Rereading the specs, s1.2 could do with a bit more clarification. Namely: - Percent encoding is ONLY used for reserved characters - Explicitly state which fields can/cannot be percent encoded....

Also, should we have a category for lenient parsing of things that are technically non-compliant? E.g. `DP=2.0` fails the integer regex but there's no loss of information if you parse...

This has not been standardised and is not currently a priority. If there has been movement in this regard (e.g. tools in the wild that use non-standardised fields allele-specific filtering)...

My reading of the specs is that the first option isn't actually valid: Section 1.4.2: `The Number entry is an Integer that describes the number of values that can be...

>Is there appetite for modifying the spec to be unambiguous? Yes, but not in a backwards incompatible manner.

> However I'll also add it then goes on to say: >If a call cannot be made for a sample at a given locus, . must be specified for each...

> The VCF specification does not dictate one cannot have the combination of GT=0/0 + GQ=0. I think it's fair to say it is not currently forbidden, despite the fact...

>I think we have enough concerns to explicitly ban the use of 0/0 for missing genotype. We already do: `If a call cannot be made for a sample at a...

Closing as this scenario is already covered in the VCF specifications. The relevant GATK output is not compliant with the VCF specifications and there are no plans to change the...