I’m reviewing some details on our new D. sechellia annotation, and noticed that D. melanogaster CG42658 looks like it’s missing a second, more-prevalent two-exon transcript that retains the second intron of CG42658-RA.
Addendum: I think the N-terminus may be too long. Looking at alignments to Drosophila orthologs, we see two things:
	1. A set of more closely related species that do have protein models starting at the same Met, but they’re all “LOW QUALITY” with frameshifts near the N-terminus. These are induced by alignment of the D. melanogaster protein, where its longer N-terminus will align but with frameshifting indels.
	2. A set of more distantly related species with proteins starting at aa-42, starting with MELL (or similar in other species)
From FlyBase (LC): That looks like an improvement to me -- the FB annotation is based on the MIP07152 cDNA, which I suspect is just invalid.  The 5' end extends beyond the extent supported by RNA-Seq. Plus, the third exon supported by this cDNA is in a different ORF -- which is not conserved at all. Looks like we should delete -RA and make a new transcript conforming to your suggestions.