Unexpected behaviour when using the SplitMultiallelics Transformer with unbounded INFO fields

[simojoe]

Use case : I am parsing a VCF file with glow and am using the SplitMultiallelics transformer to obtain biallelic entries. One of my INFO fields is unbounded, meaning that it is defined with Number=. in the VCF header.

Unexpected behaviour : When splitting, the unbounded INFO is sometimes left intact, sometimes separated into the different alternate alleles.

Due to being unbounded, the said INFO field will have a variable number of elements for each VCF entries. When that number is equal to the number of alternate alleles, the SplitMultiallelics transformer will split it into individual parts, but it will keep it intact if it is not equal.

It seems to me that this is an edge case, because one would expect an INFO field to be either split or intact throughout a single file, not both.

Possible solutions :

• INFO fields can use Number=A to indicate that the number of elements is the same as the number of alternate alleles. The transformer could allow splitting only of Number=A fields.
• For more fine-grain approaches, an argument could be added to specify INFO fields to split (or those not to split)

[henrydavidge]

I agree that the current behavior is unintuitive / basically incorrect. Your first proposal seems reasonable to me. In general, would you expect an unbounded INFO field to remain unsplit?

[henrydavidge]

@simojoe I opened a PR to resolve this issue. If you have a moment, let me know if it meets your needs.