Splicing

[d-laub]

Related to #24 and #38, how should splicing be implemented? Spooner et al. choose to "not apply any VCF changes in introns or in splice sites" in Haplosaurus and claim this was more correct than to include them. We could do this too (likely won't be terribly hard) and it seems reasonable for an initial implementation. However, later on, would it be better to generate personalized pre-mRNA transcripts and then compute (and cache) splice site predictions (e.g. from SpliceAI) to incorporate intronic/splice site variants?

@bschilder

[bschilder]

Yes I think replicating the Haplosaurus approach makes the most sense for spliced transcripts.

And I agree, adding the ability to create pre-mRNA sequences, and then let splicing-specific models figure out which exons will be included/excluded, is the way to go down the road.