Merge regulates_o_has_participant & has_regulatory_target ?

[dosumis]

regulates_o_has_participant has been used by MGI to link regulatory process to cells, e.g.

GO:0022409 positive regulation of cell-cell adhesion regulates_o_has_participant(CL:0000031)
GO:0045089 positive regulation of innate immune response regulates_o_has_participant(CL:0000040)
GO:2000035 regulation of stem cell division regulates_o_has_participant(CL:0000047)
GO:0050769 positive regulation of neurogenesis regulates_o_has_participant(CL:0002322)
GO:0050728 negative regulation of inflammatory response regulates_o_has_participant(CL:0002476)

It is logically a parent to a whole family of regulates_o relations and to some newer relations that have this pattern.

has_regulation_target has been used much more extensively, but only for participants of much lower granularity - mostly genes and gene products. Here's a breakdown of the IDs:

has_regulation_target(DDB
has_regulation_target(ENSEMBL
has_regulation_target(GO
has_regulation_target(MGI
has_regulation_target(NCBI_Gene
has_regulation_target(PR
has_regulation_target(PomBase
has_regulation_target(RNAcentral
has_regulation_target(SGD
has_regulation_target(UniProtKB
has_regulation_target(WB

The current definition of has_regulation_target is extremely vague, probably accounting for how controversial a relation this has been. We now have a number of new sub relations that are much more clearly defined. Many annotations should be pushed down to these:

• regulates_transport_or_localization_of
• regulates_expression_of
  • regulates_transcription_of
  • regulates_translation_of
• regulates_activity_of

In the meantime we need a general enough definition for has_regulation_target. One option is merge* it with regulates_o_has_participant.

=> usage: Use this relation to link a regulatory process or function with a participant in the regulated process or function.

comment: Please investigate using more specific relations before choosing this one.

The disadvantage of this is that it ignores the granularity difference in usage, but I'm not at all sure how to fold that into a definition, and the distinction is not, I think, particularly useful for inference.

@ukemi, @RLovering, please provide feedback on whether this option is satisfactory.