DEXSeq & _PAR_Y genes - paralogues are grouped with counterparts for differential usage analysis

[SamBryce-Smith]

This comes back to the _PAR_Y issue I noticed previously morrislab/qapa#49

SLC25A6's paralogue has a single 3'UTR/PAS. But it is included in differential usage output, presumably because it is grouped with the non-paralogue which has multiple 3'UTRs.

I think it would be best to keep these independent. This originates from QAPA stripping version IDs from gene IDs, so will maybe need some additional parsing. Will need to dig into this more deeply to come up with most straightforward workaround