Reference allele value for Number=R VCF INFO attributes about bdg-formats HOT 6 CLOSED

@fnothaft @jpdna I believe this one is a blocker for the next point release (0.10.1 or 0.11.0, however we decide) as I can't finish the implementation in bigdatagenomics/adam#1288 without a decision on this.

from bdg-formats.

Add new reference value fields

I like the approach in your 6. above of adding new explicitly named ref prefixed fields.
If the decision was made to do that, then what further problems still remain?

from bdg-formats.

Option 6 only helps for the specific case of Number=R VCF INFO attribute values that map to fields. We'd still need something for generic Number=R VCF INFO values in VariantAnnotation.attributes (1-4).

Then after many years dealing with corporate DBAs and other such naming standard b.s., I no longer abbreviate in field, method, or class names, so it'd be reference*. :)

from bdg-formats.

How about we have a rule that for all Number=R attribute map "fields" that the the String that is the the value must itself by a comma delimited list with exactly two fields (post splitting), ref and alt

from bdg-formats.

I'd go for that if we also went with option 5 for AD,ADF,ADR. Added to description above.

from bdg-formats.

Yeah I see why you'd want option 5 over 6 then for consistency, that's be fine with me.

We could also make attributes a map of array where often the array would be length one (rather than have comma delimited strings in some cases), but I am not sure if I like the potential overhead of that nesting. Anyway, that attributes map representation decision would be easy to change in future though if we changed mind.

from bdg-formats.