Cross reference guidelines

[pgaudet]

Hello,

Working on the definition and general cross reference guidelines, I wonder if the following sourcs should be allowed as def/general cross references:

• Reactome (maintained outside GO): for eg def xref of MPT synthase complex >> Reactome:R-HSA-947581
• ChEBI : not a source of definitions fo GO terms? for e.g. regulation of vitamin D biosynthetic process defxref: CHEBI:27300
• • is it OK as a source of synonym xreferences? see 2,5-dihydroxypyridine catabolic process => CHEBI:16364

[deustp01]

In the "definition" guideline, you note that Wikipedia and other web site links can be unstable. A patch for Wikipedia, discussed on a recent ontology developer call, is to make the definition point to a specific version of the Wiki page. The same can work for Reactome. Each event has a stable identifier, R-HSA-947581 in the MPT example. Each such identifier has a numerical suffix, which is increased each time the event annotation is changed in a substantial way. The current MPT annotation, for example, is R-HSA-947581.2.

For maintenance, it would be prudent to make the version suffix part of the GO definition AND to trigger a review of the GO term when its Reactome partner is updated, in case that update reflects a change in our understanding of the relevant biology that should propagate to the GO term. The number of these changes is probably small, and most will have no effect on GO term definitions, so the workload should be tolerable. Sanity check, please, @adamjohnwright

[pgaudet]

Thanks @deustp01

Can you tell us how you make that permanent link? At the last ontology call we could not figure it out.

[adamjohnwright]

I have been talking about this wiht @pgaudet here is the template of the url that is stable and can be used with the stable ids: https://reactome.org/content/detail/R-HSA-94758. This link will always work as long as the ID has not been made obsolete. We are in the process of making a system that will let the user know that the ID has been made obsolete with further information. Using this URL structure should be used if I am understanding your usecase correctly.

[deustp01]

At the last ontology call we could not figure it out.

I think I lumped two issues together. We definitely want to support links to Reactome reaction instances R-HSA-number in GO definitions. The confusing bit that I dragged in was that, for a number of years after we first wanted to do this, our links were not completely reliable. That problem has now been fixed, to yield the reliable scheme Adam describes just above.