Hi @KMaki,
First, I must say: congratulations on making it this far! Once you get the RDP database working seamlessly with QIIME 2 I'd encourage you to share the QZA files somewhere (e.g., zenodo) and link to it on the forum (in a "community contributions" topic) if you are interested. Others in the forum community have asked about how to format the RDP database files for use with QIIME 2, so if you are happy to share these you'd be helping all boats float higher!
Now to the error:
It sounds like at least one sequence ID is in the sequences but not the taxonomy. Two possibilities:
- are the feature IDs numeric? If yes, this is likely the issue (they are being interpreted as numbers in one file but characters in the other), and changing to non-numeric would be an easy fix.
- special characters, especially line breaks, have been reported to cause this issue in the past, see here for a diagnosis and fix:
Feature classifiier consensus vsearch - key error - #7 by Nicholas_Bokulich
I strongly suspect #2 is the issue — let me know what you find!