In the clinicaltrialskg_tsv_to_kg_jsonl.py ingest module, we are not actually saving the NCT record identifier that underlies each edge. See this line of code, where we get the NCT identifier as variable nctid but then don't do anything with it:
So, the NCT identifier that is the basis for the edge is kind of lost, at that point. This should be fixed, with high priority. Need to figure out exactly how we are supposed to put this information in a KGX edge, however. Maybe in EPC information.