Weirdness in Ubergraph around RO and BFO labels (two labels for one CURIE)

I ran a small script that grabbed some labels from Ubergraph for some CURIEs. And got some errors (the following uses Oaklib)

(ORION) [dkorn@ht1 obo-labo]$ python check_for_mapping_conflicts.py
WARNING:root:Multiple labels for BFO:0000050 = part of != part_of
WARNING:root:Multiple labels for BFO:0000051 = has part != has_part
WARNING:root:Multiple labels for RO:0002202 = develops_from != develops from
WARNING:root:Multiple labels for RO:0002202 = develops_from != develops from
WARNING:root:Multiple labels for RO:0002211 = regulates != regulates (processual)
WARNING:root:Multiple labels for RO:0002297 = results in formation of != results in formation of anatomical entity
WARNING:root:Multiple labels for BFO:0000062 = preceded by != preceded_by
WARNING:root:Multiple labels for RO:0002087 = immediately preceded by != immediately_preceded_by
WARNING:root:Multiple labels for RO:0002496 = existence starts during or after != start stage
WARNING:root:Multiple labels for RO:0002497 = existence ends during or before != end stage
WARNING:root:Multiple labels for RO:0004022 = disease arises from feature != disease has basis in feature

I dug into RO:0004022 as a test case. When I look at the sparql I see two options for the label:" "disease arises from feature" and "disease has basis in feature"

But checking the OWL for RO it looks like the correct label is "disease has basis in feature". Not sure where the other label is coming from.

Start with the Ubergraph SPARQL results for RO:0004022 and compare them with the RO OWL shown in the report. Trace where each of the two labels enters the integrated store; done means the source of the duplicate is identified and the expected single-label behavior is documented or corrected.