Feature-classifier error during training

Hey all,

I’m currently using the “Moving Pictures” tutorial to help me analyze my own data, and have gotten to the stage where I try to train my feature classifier on the 341F/805R primers for my current data set.

qiime feature-classifier extract-reads
–i-sequences 99_otus.qza
–p-trunc-len 212
–o-reads ref-seqs.qza

What ends up happening though is that I get an error message stating

Plugin error from feature-classifier:

No matches found

I’ve tried using the primers provided in the files from the company that did the sequencing, and I’ve tried using the primers found in the tutorial and I keep ending up with the same error!

Here’s hoping I can get some insight :smile:


Hi @leahtee,
Thanks for posting! Interesting error. This statement seems particularly diagnostic:

That seems to indicate that the reference sequences that you are using may be somehow faulty. A couple questions:

  1. Where is the 99_otus.qza file from? If you could also point me to the link you used for download that would be great (just in case we need to try rebuilding this on our end)
  2. Could you also please share the commands that you used to import this? And any other steps that you used between downloading and extract-reads.

Could you also please share the full error traceback — either run this command with the --verbose flag added to the end of the command, or read the contents of the log file that is listed in the error message. This will help us pinpoint precisely where the problem is occurring if the reference file does not lead to an obvious fault.


Thanks for your reply @Nicholas_Bokulich!

  1. The 99_otus.qza file came from gg_13_8_otus.tar.gz downloaded from https://docs.qiime2.org/2017.12/data-resources. I’m assuming it’s the same as the one from http://qiime.org/home_static/dataFiles.html because I’ve tried both and neither seems to work.

  2. Imported with:

qiime tools import
–type ‘FeatureData[Sequence]’
–input-path 99_otus.fasta
–output-path 99_otus.qza

qiime tools import
–type ‘FeatureData[Taxonomy]’
–source-format HeaderlessTSVTaxonomyFormat
–input-path 99_otu_taxonomy.txt
–output-path ref-taxonomy.qza

After downloading, I moved the greengenes file into the directory I’ve been working within, but unzipped the files (i.e. the .txt and the .fasta) and moved them out of the original file folder (gg_13_8_otus.tar.gz) and into the directory.

Lastly, here is the error message with the debug info

Plugin error from feature-classifier:

No matches found

Debug info has been saved to /tmp/qiime2-q2cli-err-ksh878ys.log

Thank you so much for your help!

Very strange. I have replicated your commands locally and have no issue — the output contains 406010 sequences.

Here’s a different thought — the primer sequences that you are inputting might contain special characters that are that are throwing off the extraction. Writing commands in some word processors and text programs will sometimes automatically create these special characters as part of auto-correct, and copying/pasting from websites is another frequent cause. Do you maybe want to try typing out the commands entirely in your terminal?


P.S., this does not look like a bug in the code, so I don’t think I need to see the error traceback after all, but the full error log for the command you used would be contained in /tmp/qiime2-q2cli-err-ksh878ys.log according to the output that you provided. just letting you know in case you have future support questions in the forum that will require this, or if you ever run into an error and want to take a look under the hood. thanks!

Hi @Nicholas_Bokulich! I tried it again, and got the same error as before, even with typing in the commands and sequences, rather than copying and pasting them.

Do you think it has to do with the primer sequences themselves (i.e. that I’m using a sequence that is too long)? Or maybe it’s my truncation, should I be using the 120 instead of the 212 I’ve chosen?

Sorry if these questions seem kind of naive, I’m new to QIIME!

Thank you so much!


No, it cannot be either of those. I have replicated every command you ran, and successfully extract several hundred thousand sequences.

Could you please share the output of the following command?
qiime info

Could you also upload the 99_otus.qza file?

Sure thing!

System versions
Python version: 3.5.4
QIIME 2 release: 2017.12
QIIME 2 version: 2017.12.1
q2cli version: 2017.12.0

Installed plugins
alignment 2017.12.0
composition 2017.12.0
cutadapt 2017.12.0
dada2 2017.12.1
deblur 2017.12.0
demux 2017.12.0
diversity 2017.12.0
emperor 2017.12.0
feature-classifier 2017.12.0
feature-table 2017.12.0
gneiss 2017.12.0
longitudinal 2017.12.0
metadata 2017.12.0
phylogeny 2017.12.0
quality-control 2017.12.0
quality-filter 2017.12.0
sample-classifier 2017.12.0
taxa 2017.12.0
types 2017.12.0
vsearch 2017.12.0

Application config directory

Link to the 99_otus file:

Thanks again for all the help :):smiley:

Thanks for sharing the file! Problem solved — you imported aligned reference sequences. Instead, you should use the unaligned reference sequences (in the “rep_set” directory of the greengenes files).

Let me know if you are still running into trouble with the unaligned seqs!


Ohhhhhhhhh! That worked! Thank you so much!!


This topic was automatically closed 31 days after the last reply. New replies are no longer allowed.