XP:molecular function xp protein

Logical definitions of molecular functions in terms of proteins, as represented in PRO

Example:

[Term]
id: GO:0001532 ! interleukin-21 receptor activity
intersection_of: GO:0004872 ! receptor activity
intersection_of: OBO_REL:has_input PRO:000001382 ! interleukin-21

The structure of these are very similar to [[XP:molecular_function_xp_

Availability

Logical definitions:

• molecular_function_xp_protein.obo

Use in annotation

See slide 9 of http://www.slideshare.net/cmungall/go-pathwayinteractionintegration

unfortunately the arrows got shifted a little in the transition to slide share, but the basic inference is as follows:

If FIGF binds to human VEGFR2 AND
  human VEGFR2 is_a generic VEGFR2 AND
 any binding instance for which the target is an instance of generic VEGFR2 is necessarily an instance of GO:0043184 VEGFR2 binding
THEN
 FIGF has VEGFR2 binding activity

This uses the following logical definition from molecular_function_xp_protein:

[Term]
id: GO:0043184 ! vascular endothelial growth factor receptor 2 binding
intersection_of: GO:0005488 ! binding
intersection_of: OBO_REL:results_in_binding_of PRO:000002112 ! vascular endothelial growth factor receptor 2

and the following is_a hierarchy from PRO:

  is_a PRO:000018263 ! amino acid chain
   is_a PRO:000000001 ! protein
    is_a PRO:000001971 ! vascular endothelial growth factor receptor
     is_a PRO:000002112 ! vascular endothelial growth factor receptor 2 *** 
      is_a UniProtKB:P35918 ! VEGFR2 (Mus)
      is_a UniProtKB:P35968 ! VEGFR2 (Hsa)

This also works in cases where GO choose not to name highly specific binding classes. For example, if there we no VEGFR2 binding in GO, only VEGFR binding, then the inference would be to VEGFR binding (and if we later added VEGFR2 binding) we could automatically deepen the annotation.

  is_a PRO:000018263 ! amino acid chain
   is_a PRO:000000001 ! protein
    is_a PRO:000001971 ! vascular endothelial growth factor receptor
     is_a PRO:000002112 ! vascular endothelial growth factor receptor 2 *** 
      is_a UniProtKB:P35918 ! VEGFR2 (Mus)
      is_a UniProtKB:P35968 ! VEGFR2 (Hsa)

One limitation is that there are only mammalian mappings to PRO but this could easily be extended, e.g. using PANTHER.

This is only useful for a subset of the binding terms. We could use a similar strategy for terms like "X kinase binding", where the classification is based on the activity of the partner rather than the family