Closed dustine32 closed 5 months ago
This works! At risk of being uninformed, what if the input-output step was put after the inference of all functions? The CTNNB1 is the output of the 'Beta-catenin is released from the destruction complex' and the input of the black box 'Beta-catenin translocates to the nucleus'.
Other cases where gene products or complexes can be output / input connectors would include signaling cascades and initiation of DNA replication.
@ukemi Yes, I think that's a very sensible suggestion though with how the code is currently set up, it'll take some more refactoring work to plug that in. We may be forced to do that at some point.
For #308.
Removes the conditional criteria blocking the sharing of
has_output
/has_input
small molecule instances for Reactome models.This intermediate sharing pattern was initially added for YeastPathways models and just I realized there was previously no filter to restrict this to only
SmallMolecule
BioPAX classes. I think this was ok for YeastPathways since their pathway intermediate input/outputs were always small molecule classes. However, this change broke a testtestInferProteinLocalizationProcess
for Reactome R-HSA-4641262 because a proteinCTNNB1
was an intermediate input/output in part to represent a protein transport activity and the new intermediate sharing pattern was breaking the inference logic. To fix, I just limited this intermediate sharing toSmallMolecule
.