GO-CAM ectopic meeting July 19th 2017

Zoom URL

https://stanford.zoom.us/j/679970729

Agenda

Discussion of inference on Wnt signaling LEGO model

http://noctua.berkeleybop.org/editor/graph/gomodel:5966411600000538

Missing/New inferences:

contributes_to inference:

Added to RO: part of'(?x, ?y) ^ enables(?y, ?z) ^ molecular_function(?z) -> 'contributes to'(?x, ?z) e.g. GP(X) part_of complex(Y) enables MF(Z) -> X contributes_to Z

Paul suggests:

'contributes to'(?x, ?y) ^ part_of(?y, ?z) -> 'involved in'(?x, ?z)

This inference is considered safe because in GO-CAM models a whole complex will only be annotated if the complex is required to perform the molecular function in the context of the process. One could imagine cases with unsafe inference, for example a complex with a negative regulators subunit that regulates an activity that is part of a process. But Paul suggests that annotations rules should get around this: In this case there should be no annotation to the complex.

This may be one we have to revisit if odd annotations start showing up.

Annotation to compound molecular function => annotation to its parts

enables(?x, ?y) ^ 'has part'(?y, ?z) ^ molecular_function(?y) -> enables(?x, ?z)

e.g. if GP X enables ATPase coupled transporter activity' and 'ATPase coupled transporter activity' has_part 'ATPase activity' then GP(X) enables 'ATPase activity'

Inferences from molecular function regulators

https://github.com/geneontology/go-ontology/issues/13492)

e.g.:

X <— enabled by— protein binding —directly_positively_regulates --> protein kinase activity —enabled_by--> Y

=>

  X enables ‘protein binding activity' has_input Y *
  X enables 'protein kinase activator activity’ has_input Y *

\* Comes from a new rule: molecular_function(?x) ^ molecular_function(?y) ^ 'directly regulates'(?x, ?y) ^ enables(?z, ?y) -> 'has input'(?x, ?z)

e.g. if 'protein kinase activity'(X) directly_regulates 'protein binding activity (Y)and this is enabled by GP(Z) then X has_input Z

Wrong inferences that need to be trimmed:

regulates o part_of->regulates

https://github.com/geneontology/go-ontology/issues/13926#issuecomment-316054054

Strategies for trimming down superfluous inference

1. enables and other GP/Complex -> MF/BP relations never go in extensions. Inference of the ‘contributes to’ qualifier would make up for this.
2. some very high level classes should be blacklisted from extensions (molecular_function, regulation of molecular function. We could perhaps existing no annotation subset flags for these.)
3. If we have ‘X regulation’ then the GPAD has only GP involved_in 'regulation of X', not GP involved_in_regulation_of X.

This is hard to implement: simpler alternative would be to just drop this GP-GO term relation and rely on named regulation classes alone

Minutes

• On call: David H, David OS, Kimberly, Paul T.

Protein Complex Annotations

• New rule will allow for individual members of the complex to inherit the MF annotation with a contributes_to qualifier
• For BP annotations, if the MF is performed by a protein complex, and the MF is part_of the BP, then each member of the complex is involved_in the BP

enables o regulates

gene product x enables activity y; activity y regulates process z, gene product x involved_in 'regulation of z'