replace/enhance isX patterns with SWRL rules, e.g. for phosphorylates

[cmungall]

Example: http://purl.obolibrary.org/obo/RO_0002447

has axiom:

(capable of o is kinase activity o has direct input) subPropertyOf phosphorylates

this infers entity-entity precise relationships given a GO-CAM pattern

this pattern is familiar to owl ultra-geeks but is probably bamboozling to others. It doesn't help that ontobee doesn't show self-reflexivity axiom for http://purl.obolibrary.org/obo/RO_0002481. Another issue is that this is not in EL.

It would be cleaner to write this as a SWRL rule. We can have an optional transform to the self-reflexive pattern.

thoughts @balhoff ?

[cmungall]

?a1 enabled_by ?gp1
?gp1 a GO:'kinase activity'
?a1 directly-regulates ?a2
?a2 enabled-by ?gp2
->
?gp1 phosphorylates ?gp2

similar pattern for ubiquitination etc

[balhoff]

I agree the SWRL rule is less mysterious and would actually be supported in some GO reasoning situations (Arachne or Whelk). However it wouldn't contribute to Tbox classification; maybe this doesn't matter since the current Self axiom isn't either. ObjectHasSelf is supposed to be in EL, but as you say ELK doesn't support it (I checked 0.5 also).

[cmungall]

Good point. Still useful for a lot of ABox use cases.

[cthoyt]

I noticed that the only two post-translational modifications available in OBO-RO are phosphorylates and ubiquitinates. Is it non-trivial to add more of these relationships if they're being defined this way?

I'm interested in this because we capture several more PTMs in BEL (amination, nitration, nitrosylation, glycation, etc.) that live under the GO term cellular protein modification process (GO:0006464). I'm not (yet) an ultra-geek, but it seems like defining more of these using this SWRL rule might not be so difficult.

Should I make a new issue to continue this discussion?

[cmungall]

Is it non-trivial to add more of these relationships if they're being defined this way?

Trivial but a bit tedious. I think we should try and devote a couple of hours to getting our templating framework set up so this becomes a matter of filling in terms in a CSV. @balhoff would dosdp-tools/spec require any work to include SWRL?

Yes, go ahead and make a new ticket, thanks!

[dosumis]

?a1 enabled_by ?gp1 ?gp1 a GO:'kinase activity’ ?a1 directly-regulates ?a2 ?a2 enabled-by ?gp2 -> ?gp1 phosphorylates ?gp2 On 28 Sep 2018, at 01:34, Chris Mungall notifications@github.com wrote:

Is it non-trivial to add more of these relationships if they're being defined this way?

Trivial but a bit tedious. I think we should try and devote a couple of hours to getting our templating framework set up so this becomes a matter of filling in terms in a CSV. @balhoff https://github.com/balhoff would dosdp-tools/spec require any work to include SWRL?

It should be reasonably straightforward to extend the DOSDP spec to support adding rules. Needs support for OP vars (deliberately omitted to now - but I should probably add these anyway). Do you anticipate adding very large numbers of these? Will they be self-contained or part of templates specifying a broader pattern?

[nlharris]

Is this still needed (and if so, who could work on it)?

[cthoyt]

This is related to #522, which will get some discussion at the RO meeting next week