Make protein sets OWL Union entities in GO-CAMs about pathways2go HOT 6 CLOSED

(note from Geneva meeting:

See pyruvate kinase tetramer - it is a set of complexes. Make a union of the complexes as the controller entity. In reaction phosphoenolpyruvate + ADP => pyruvate + ATP from glycolysis. Complexes need to maintain has_part relations to proteins.
Tabular annotations from this one should look like:
Gene_product enables MF )
See also 'part_of o enables -> contributes_to' geneontology/minerva#110

from pathways2go.

See 'p-S189,T193-MAP2K3, p-S207,T211-MAP2K6 ' in the reaction: 'Phosphorylated MKK3/MKK6 migrates to nucleus' in the pathway 'activated TAK1 mediates p38 MAPK activation' for another example of a protein set (that is not a complex).

from pathways2go.

This is working according to issue description (hence closed it), but no one has confirmed they like the output. Ping @ukemi @thomaspd @cmungall See e.g. http://noctua-dev.berkeleybop.org/editor/graph/gomodel:284362402 and look for 'union'

from pathways2go.

@bgood
From 2019-03-01 discussions at NYU: given that the sets identify the gene products or complexes that are individually capable of performing the reaction, we can't think of a better way to represent this.
Still to do: make sure that the resulting annotations (in GPAD) are correct.
Note that the above model is currently failing reasoning due to logical inconsistency.

from pathways2go.

Sets are supposed to be ordered lists. Here is an edge case. Reaction R-HSA-73548 "(d)CMP or UMP + ATP <=> (d)CDP or UDP + ADP (CMPK1)" has as input the set [dCMP, CMP, UMP], and as output the set [dCDP, CDP, UDP]. The intended meaning is that dCMP can be converted to dCDP, or CMP can be converted to CDP, or UMP can be converted to UDP. The first item on the input list becomes the first item on the output list, and so forth.

Further, this reaction is catalyzed by a set of two enzymes, which is intended to mean that each of the two enzymes is capable of catalyzing each of the three reactions listed above.

from pathways2go.

@vanaukenk apologies for reasoning problems in noctua-dev. I have a request in to fix them. (Not the fault of the models.) If there is a specific question I can answer, I can run tests in my local environment. I can tell you right now though that the gpad output is not going to contain genes from within the union or intersection constructs as it stands. The export code will need to be adapted to look in there. How would you want the output to look like - for sets/complexes ?

@deustp01 any ordering of members within sets such as these is lost in the BioPAX translation. Though more verbose, I think it would be more useful computationally to break shorthand reaction definitions apart into the different reactions that represent. Of course that is a Reactome decision to make.

from pathways2go.