Complex shape mismatch analysis

[ukemi]

The entries in this ticket will consist of an examination of the Complex shape mismatch violations on this document:

https://docs.google.com/document/d/1BWgjgdDLyAUdgRs4hwPanh1WYr8tZQOjY77kVV8ATus/edit?tab=t.0

[ukemi]

Model R-HSA-1592389 Not a metabolic model

Explanation: The 'enabled by' relationship from 'serine-type endopeptidase activity (GO:0004252)' points to a protein-containing complex (GO:0032991), but it matched the CellularComponent shape instead of ProteinContainingComplex. Expected one of: Protein-containing complex (GO:0032991), Information biomacromolecule — proteins, nucleic acids (CHEBI:33695).

• I'm not really sure what's going on here, but this model also has the Multi-type entity violation. If I delete all the offending protein-containing complexes, all the violations on this model go away.
• If I delete all of the single enablers from the complexes, all of the violations go away.

Conclusion- These violations seem to be stemming from the issues that are reported in #395 and #396. If we fix that problem, we fix an entire set of violations.

[ukemi]

Model R-HSA-189451 has four empty shapes, one Reacto Entity mismatch, and one Complex shape mismatch. Metabolic model

Explanation:The 'enabled by' relationship from 'protoporphyrin ferrochelatase activity (GO:0004325)' points to a protein-containing complex (GO:0032991), but it matched the CellularComponent shape instead of ProteinContainingComplex. Expected one of: Protein-containing complex (GO:0032991), Information biomacromolecule — proteins, nucleic acids (CHEBI:33695).

• Reaction R-HSA-189465 is enabled by a complex that consists of a dimer of the FETCH gene product (UniProt:P22830) and an iron-sulfur cluster (CHEBI:49601). For some reason the CHEBI entity is not being filtered out of the enabler slot. If I remove the irons sulfur cluster from the 'complex', the error on the individual is resolved. PD comment: yes, in Reactome the complex has a protein part with a UniProt ID and a chemical part (the iron-sulfur complex) with a ChEBI ID.

Conclusion- This is throwing an error because the CHEBI entity is not being properly filtered from the enabler. I'm not sure why it says it is matching the cellular component shape. In the ShEx, protein-containing complex can have parts ( InformationBiomacromolecule OR ProteinContainingComplex). As far as I can tell, this excludes small molecules, so this would explain why it doesn't match the shape. The CellularComponent class has no restrictions on has_part.

[ukemi]

Model R-HSA-71403 two complex shape mismatches and three REACTO entity mismatch errors metabolic model

Explanation: The 'enabled by' relationship from 'succinate dehydrogenase (quinone) activity (GO:0008177)' points to a protein-containing complex (GO:0032991), but it matched the CellularComponent shape instead of ProteinContainingComplex. Expected one of: Protein-containing complex (GO:0032991), Information biomacromolecule — proteins, nucleic acids (CHEBI:33695).

• This is another error that is created by the inclusion of the iron-sulfur cluster in the enabling protein complex for reaction R-HSA-70994. The complex is still broken up into sub-complexes. I believe the GO-CAM modeling group has decided that these should be simplified into a flat list. Nonetheless, if I manually remove the iron-sulfur complexes, the error is resolved.

Conclusion- This is another case of an iron-sulfur cluster being incorrectly included in an enabling protein-containing complex.

[ukemi]

Model R-HSA-9766229 two missing labels, two complex shape mismatch and two Multi-type entity errors. Not a metabolic model

Explanation: The 'enabled by' relationship from 'peptidase activity (GO:0008233)' points to a protein-containing complex (GO:0032991), but it matched the CellularComponent shape instead of ProteinContainingComplex. Expected one of: Protein-containing complex (GO:0032991), Information biomacromolecule — proteins, nucleic acids (CHEBI:33695).

• Reaction R-HSA-9935547 is enabled by a set of three individuals; a monomer of CTSS which is represented correctly in the model, a dimer of two fragments of CTSB which is represented as a protein-containing complex of two fragments from the CTSB gene, a dimer of two fragments of CTSL which is represented as a protein-containing complex of two fragments from the CTSL gene. Manually removing the CTSL and CTSB and replacing them with the UniProt entries for the gene products corrects the error. In fact, it corrects all the errors.

Conclusion- This is another case where an enabler is made up of two or more fragments of the same protein and it is being imported as a protein complex that has_part something invalid. These need to be simplified to be just the genes