Closed dhimmel closed 3 years ago
Thanks @dhimmel, this has been a gradual clean up in EFO but I will definitely look into this a little more and see if there are any EFO-originating invalid cross-references that need cleaning up.
These have now been looked at and fixed during our other clean ups, therefore I'll move this to done.
We're using
oboInOwl:hasDbXref
relationships to translate information from other terminologies into EFO terms.There are a few easy-to-detect xref values that are likely invalid. I imagine xrefs are often imported from elsewhere? So some of these are likely upstream problems. But others might originate with EFO.
whitespace
There are some xrefs with leading / trailing whitespace (not all whitespace renders below, but I looked for when the trimmed version of xref did not match xref):
mesh identifiers
Using the following regex for valid MeSH IDs:
^[CD][0-9]{6}([0-9]{3}|)$
.I didn't include xrefs to MeSH tree locations, which were almost entirely from UBERON and reported upstream in https://github.com/obophenotype/uberon/issues/698.