Hello,
For NTA, the only database available is the GO: Biological Process database. If you would like, you could download the results and look at the seedsInNetwork.txt file to get the interesting gene list and independently perform your own ORA analysis.
For the ORA enrichment, you are correct in how the reference set is selected. All of the genes that are in the network make up the reference set, and the identified nodes from the analysis are the interesting list.
If you have any questions, please let me know!
Best,
John