RDP Reference Database in QIIME2 format

[Nicholas_Bokulich]

Hi @KMaki,

@KMaki

The feature ids seems to match up, but is there a way to check if they are labeled as numeric?

They do not look numeric, so assuming all follow that format that is not the issue.

@KMaki

The file shows that it was updated, but do I need to add any flags/options?

No, sounds like it worked! You have a related (but new) error, so that means that the windows-style line breaks were the issue before.

@KMaki

Does the classification automatically terminate if something in the taxonomic search results is not present in reference taxonomy. Is this error saying that identifiers S002156889, S000615995, and S000269333 are missing from one or both of the files?

That's right. This error is specifically saying that those IDs are in the sequences but not the taxonomy.

@KMaki

Interestingly, if I extract the variable region from the RDP database based on primers and run the Vsearch workflow, I am able to classify taxonomy.

probably because those IDs are no longer included in the top matches; they may be excluded by extract-reads (e.g., if they are too short/long) or trimming off the rest of the sequence alters the kmer profile (which vsearch uses to queue up the first N hits for alignment).

This command might help you find all IDs that are unique in one file or another (no guarantees this will work, I'm sort of ad-libbing based on the file snippets you shared above):

grep '^>' rep_set_99_rdp.fa | tr -d '>' | cat - rdp_qiime_taxonomy.txt | grep -v ';' | sort | uniq -u

Then you can confirm that those IDs are really missing from one file but not the other like this (you only need to run this once or twice, just to make sure the command above worked):

id='paste-the-id-here'
for f in rep_set_99_rdp.fa rdp_qiime_taxonomy.txt
do
  grep $id $f | wc -l
done

Want to give that a spin and see how many IDs are missing?