Improve ID management when using DOSDP

[matentzn]

One problem I continuously stumble over now with the new data pipelines for some of the organisms is the problem of novel ids. I think we need to be able to leave the defined_class column in the DOSDP TSV file empty, and have a process that, whenever a field was left blank, populates it with a fresh id from a given id range before compiling the pattern to OWL. The reason why I wanted to stick the issue on the ODK repo rather than DOSDP is because I am not sure we want to overload DOSDP having to be forcibly aware of the already occupied id space - that could be scattered across multiple modules and/or TSV files.

My suggestion is to have a simple python script shipping with the odk that loops through all the TSV files, and, when happening over an empty field, drawing a new ID. What I am not sure yet is where and how the various id ranges should be stored (say one for patterns, one for manual).

[matentzn]

@cmungall @balhoff @dosumis

[balhoff]

Doesn't EBI have a server API that addresses this?

[dosumis]

https://github.com/EBISPOT/urigen

It keeps register of IDs used - so doesn't need to look through TSV s & OWL files to check.

Don't think there's a live service anymore. We looked into running this for VFB, but IIRC, we couldn't get it working out of the box. Think it needs some updating.

[gouttegd]

@matentzn No updated in 5 years, is that still desired?

I could consider doing something for that in my dicer tool.

[matentzn]

Very low priority and should if at all be done by implementing @balhoff idea of the robot mint command (I think there is a pr for this that is 6 years old). I am also happy to just close this this has very little ROI.

[gouttegd]

should if at all be done by implementing @balhoff idea of the robot mint command

FYI, as I mentioned here the mint command as envisioned has been implemented as part of my KGCL plugin.

But that command can only mint ID within an ontology, not within TSV files.

[matentzn]

Ok, as I said, for me, this is very low priority. It is super easy to spend 2 minutes adding a few new Ids in a google sheet, and the many ways in which automatic generation of ids can go wrong (thinking of runs across branches, etc), makes it IMO not worth the hassle.

[gouttegd]

OK, keeping this open in the “wishlist” – stuff to do in the parallel universe where we have a lot of time on our hands. ;p

[matentzn]

I would close it to clean the tracker and have whoever wants this reopen this with a strong case

[gouttegd]

Since you’re the one who opened the ticket in the first place, if you don’t think yourself it is worth doing, then OK for closing.