CG13749 looks like a paralog of IMPPP; other species have a longer C-terminus. In addition, in most species and also IMPPP, there are 4 exons with a different N-terminus, like this example in D. elegans: https://www.ncbi.nlm.nih.gov/gene/108149888. Alignments against the 6 D. melanogaster species currently available at NCBI show an insertion in 4/6 of those other assemblies vs Release 6, right near the end of the current CDS. That’s consistent with this gene being a segregating polymorphic pseudogene within Dmel populations.
I Spligned NM_136584.2 to CM_010529.1 to figure out the revised sequence. The only difference is the CM sequence has an extra TCGACGTCGG insertion. So the corrected transcript is something like this (also removing the extra intron I mentioned based on the cross-species evidence):
>NM_136584.2 Drosophila melanogaster uncharacterized protein (CG13749), mRNA
TGACAATGCATTTGCTTGGACTTCTCGTGCCGATCGCTTGCGCTGTTTTGTGCGTTG GTGAAAAATCCCT
TTACCATGGTCTCGCCCAAACGGAGACTAATCGACGATTCCATTGCTCCCTCAG CTGCCGATCTCCAGCG
GACATACGACGGACATGATGGGCCTCACGTTTTCGGAACACCCGGCAATCAAGTCTACATTCGGGGGCAG
AACGATGGCGTATACAAGGTGCCAGGAGTAGGTGGTCAGTTCCATCACGCATCGTCGCCAGCAGATCACG
TCTATACGGATGAGCAGGGGATCACGTATGTGCACAAAAAAGACGCTGGAGGACCAGGAACACATACGCT
TAGAGGACCAGGTCTAGCTGCTGTACAGCCCGCCTACTCCACTGTGCAGCCTGCGGGACTTCGCAATCCC
CAGTTTCATGTGGAGCGAAGTGGTCGCACTGTGGATGTTGGTTCTGGAGGATTTGTCGTCCAAAGGAGCA
GACGTAGTCCTCAGTTCCATGTGGAGCGTCCTGGTCGCACTGTGGATGTGGGTTCCGGAGGATTTTTTGT
CCAAAGGGGCAGGCGTAGTCCCCAGTTTCCTGTGGAGCGACCTTGGCAAAGGG tcgacgtcgg CTTTGAAGGACTGATTG
TTGAAAGGGACAGCCGCAGCATCAAAGATGGCCGCGTCCAGGGAGAAAACTTCGTGGCTCGTGACGATCA
AGCAGGAATTTGGGATGATAGTGTGTCGGTGTGGAAGCGTCCGGATGGACGAACTGTCTCGGTGGGCAAT
GACGGAAGTGTTATTGTTTCTGGACATGGGCACACGCTGCAATACTAATAAATAATGGGAAGCAGCCTTT
TAATTTACAGAGATTAGGTAGCTAAAATTCCAATGATATTAAATGAACAAGATTCAC
I ran that through ORFfinder to get the translation:
https://www.ncbi.nlm.nih.gov/orffinder/
>lcl|ORF5
MHLLGLLVPIACAVLCVAADLQRTYDGHDGPHVFGTPGNQVYIRGQNDGV
YKVPGVGGQFHHASSPADHVYTDEQGITYVHKKDAGGPGTHTLRGPGLAA
VQPAYSTVQPAGLRNPQFHVERSGRTVDVGSGGFVVQRSRRSPQFHVERP
GRTVDVGSGGFFVQRGRRSPQFPVERPWQRVDVGFEGLIVERDSRSIKDG
RVQGENFVARDDQAGIWDDSVSVWKRPDGRTVSVGNDGSVIVSGHGHTLQ
Y
(251-aa)
The revised N-terminus includes a predicted signal peptide mostly encoded in the revised first exon.
From FlyBase (LC): Nice detective work!  Our policy has been to annotate genes described as polymorphic pseudogenes as mutations in the strain (thus providing the correct CDS) and flagging with a comment that it may be a polymorphic pseudogene.  However, I see no evidence in the data presented in GBrowse and JBrowse on FlyBase for the additional intron at the 5' end.  I agree, sequence conservation supports it, but usually there would be additional support -- at least some vague evidence in the RNA-Seq.  I would be inclined to annotate 2 isoforms, one with the intron and one without.
Response from T. Murphy: The splice probably is rare. Here’s a view with the aggregate exon coverage in linear scale:
https://www.ncbi.nlm.nih.gov/genome/gdv/browser/genome/?cfg=NCID_1_48358646_130.14.22.10_9146_1587743476_1511586678
There’s a bit of a coverage dip in that region, but lots of signal. So two isoforms with and without the retained intron make sense.