Open dosumis opened 4 months ago
We need to be able to see xrefs coming back in from external ontologies (particularly FBbt). These now live externally. @gouttegd - what is the easiest way to query for them? Can we use UberGraph?
Currently, only the FBbt mappings live externally (in a SSSOM mapping set). The mappings with the other species-specific ontologies are still present directly as cross-references in the -edit
file.
As for the FBbt mappings, they are converted to cross-references and injected into CL at build time, so they do appear in all released products as “normal” cross-references and they can be queried by any tool that knows how to access/display cross-references – which should include Ubergraph.
Here, no FBbt xrefs are visible for any of the terms listed because none of them are actually mapped to any FBbt term.
Many are grouping terms with logical defs, we can potentially use these logical defs to auto-roll textual definitions. @gouttegd - I believe you have an updated version of the software that FBbt uses for this. How easy would that be to deploy here?
Very easy. Note however that the software will only generate text definitions for terms that have a so-called “dot definition” (a definition consisting of a single dot, .
). It will not generate definitions for terms that have no definition annotation at all.
So we’ll need to either add a .
definition to all those terms, or I will need update the software to add the possibility to work on definition-lacking terms (feasible, but will take a bit more time).
Here, no FBbt xrefs are visible for any of the terms listed because none of them are actually mapped to any FBbt term.
All of the R photoreceptors in the table clearly should be. Can you add them to the mapping file?
Looks like there are some unmapped fish terms too.
c.curie | c.label | c.definition | c.database_cross_reference |
---|---|---|---|
CL:1000245 | posterior lateral line ganglion neuron | null | null |
CL:0008021 | anterior lateral line ganglion neuron | null | null |
CL:0000246 | Mauthner neuron | null | null |
All of the R photoreceptors in the table clearly should be.
In principle I am reluctant to map terms without a text definition to ensure the terms do refer to the same thing. In this particular case, I’ll admit the CL terms probably cannot refer to anything else, so OK.
Looks like there are some unmapped fish terms too.
For reasons unknown (at least to me), ZFA-CL mappings are represented as cross-references, but on the ZFA side (ZFA terms carry cross-references pointing to CL terms, not the other way round; some CL terms do have cross-references to ZFA but they are ignored while bridging the two ontologies together, the source of truth is ZFA).
This means that the ZFA-CL mappings cannot be seen from CL; they can be seen from ZFA only, or from the SSSOM mapping set that Uberon is deriving from the ZFA xrefs.
CL:0000246 is mapped to ZFA:0009149; the other two CL terms CL:1000245 and CL:0008021 are not mapped to anything in CL. That must be fixed on the ZFA side.
@gouttegd what does ZFIN need to do for CL:1000245 and CL:0008021 in ZFA?
@ybradford Find the equivalent terms in ZFA and add CL cross-references to them.
But that is assuming that ZFA does have equivalent terms for CL:1000245 and CL:0008021, which may not actually be the case.
CL:1000245 is posterior lateral line ganglion neuron
. From what I see, ZFA has a posterior lateral line ganglion
(ZFA:0001314) (which is already mapped to its Uberon equivalent UBERON:2001314), but it does not have a term for a neuron of that ganglion.
Likewise for CL:0008021 (anterior lateral line ganglion neuron
). ZFA has a anterior lateral line ganglion
) (ZFA:0001391) (already mapped to its Uberon equivalent UBERON:2001391), but no term for a neuron of that ganglion.
So it may not be something for ZFA to fix after all.
We have 71 terms for types of neuron that lack a text definition (see table below). We should define them
Strategies for finding more info:
@AvolaAmg - can you prioritise terms that are obviously applicable to mammalian brain?