I have just made a non-coding isoform transcript (Y73F8A.32b) which will appear in WS227.
It has extensive evidence from the RNASeq alignments.
Y73F8A.32b is on the reverse sense to the gene Y73F8A.33.
There is no RNASeq evidence for Y73F8A.33 being transcribed. It has a 3’ RST and modENCODE evidence of a poly-A site, but no other hard evidence for its existence.
It was tempting to remove Y73F8A.33, but it has ab initio gene predictions from twinscan, Jigsaw, mGene and Genefinder and has numerous protein homologies (from proteins with zinc-finger domains - there also appears to be a transposase domain about a third of the way along it). Some of the structure of Y73F8A.33 uses weak splice sites and it is possible that it is either a pseudogene, or its structure should be altered to avoid overlapping the exons of Y73F8A.32b.
On balance I think it is it is probable that Y73F8A.33 is a pseudogene and that Y73F8A.32b is a non-functional transcript which is being spuriously transcribed, but further evidence is required to confirm this.