MSO has no link connecting transcripts and their parts to genes

[cmungall]

SO allows us to trace back from exons, introns etc to transcripts (via part_of) and from transcripts to genes (via member_of).

Admittedly member_of may be a confusing relationship here, but at least there is some kind of chain of relationships reflecting the central dogma, this seems to have been lost in MSO

[msinclair2]

Aren't the parents of these terms defined as being part_of some transcript?

E.g. exon has superclass 'transcript region', defined as

'sequence molecular entity region'
 and ('part of' some transcript)

[mikebada]

It was a while ago, but the member_of assertions for exon, etc. seemed janky, and so there was a discussion of alternate representations. Exons and introns can refer to either DNA or RNA, and it was decided to make the base exon/intron classes refer to the RNA. (Making separate DNA and RNA exon and intron classes was also considered.) I'm certainly open to discussion about this.

Note that a number of assertions using 'transcribed from'/'transcribed to' have been made.

[keilbeck]

Mike and Michael,
The relationships between parts and wholes, genes and transcripts is a key part of SO with respect to genomic features. We had a lot of conversations where this was stressed.

[msinclair2]

Karen,

I know. According to what I can see, these were retained. Please see my comment above.

I would like Mike to comment further on this since the way these relations were to be captured in MSO was apparently decided a long time ago when the ontology was first built.

[mikebada]

Michael and I agree that the mereotopological assertions are very important, so much that we actually added an extra step into the pipeline to make sure that these are kept even in the SO. I think this particular example is tricky, as since exons and introns can refer to either the DNA or RNA entities, they could be asserted to be parts of either genes or transcripts. I'd lean toward creating DNA and RNA exon and intron classes and correspondingly linking them, but I'm of course open to other views.