problem with classifier during deblur denoise-other

Hello,
I am using a classifier based on a dataset downloaded from the SILVA browser, which only includes fungi sequences.
In detail, I download the sequences in Fasta format, separate the sequences from the taxonomy and have formatted both files as * .fna and taxonomy_all_levels.txt, respectively; no errors returned during classifier training. The error returns when I use the classifier during “qiime deblur denoise-other”, this command returns:
Invalid value for “–i-reference-seqs”:
‘fungi_ref-seqs.qza’ is not a QIIME 2 artifact
(.qza)

Anyone have any suggestions?

Thanks in advance
Federica

Ciao Federica,
welcome in the forum.

First of all, never used Silva for Fungi but glad to learn it contains that too.
Can I ask you if you can provide the full command you used for that?

In general, any qiime2 plug ins works with qiime2 artefacts as input.
I suspect you trying to import the database sequence references and the respective taxonomy in qiime 2 artefacts, for this please have a look at the tutorial on how to prepare a reference https://docs.qiime2.org/2020.2/tutorials/feature-classifier/.

On preparing the reference, please be sure to that the taxonomy you using is all at 7 levels and with the format compatible with qiime2 (in general Silva taxonomy is usually compatible with qiime2 but it may contains many more taxonomical levels).

My final point would be on the use of the deblur plug in, its purpose is to denoise the sequencing data (probably pre-processed for merging into a single sequences and quality filtering), and is not for taxonomy assignment.

Among the deblur output you will find an artefact including the representative sequences, and it is this one that should be use for the taxonomical assignment step wit. Again, I think
it would be useful to work on the “Moving Picture tutorial” even if is not strictly on fungi (https://docs.qiime2.org/2020.2/tutorials/moving-pictures/).

Good luck
Luca

Welcome to the forum @Federica_Chiappori! And thank you @llenzi for your detailed advice!

I have just a couple things to add:

deblur does not use a trained taxonomic classifier; you will want to use the reference sequences that are used as input to feature-classifier fit-classifier-naive-bayes. From the filename you’ve listed I think you are already aware of that, but I just want to clarify here for other readers that the discussion of classifiers is not actually related to the issue at hand here.

This cryptic error message is also not really pointing in the right direction. It sounds like you used this file to train a taxonomic classifier, so it must be a valid QIIME 2 artifact. This error is also sometimes raised when a file does not exist, so I expect that is what is occurring here.

Good luck!

Ciao Luca
thanks for your reply!

I use exactly the commands in the https://docs.qiime2.org/2020.2/tutorials/feature-classifier/ tutorial.

The taxonomy file includes 13 levels, similar to the file “taxonomy_all_levels.txt” included in the Silva directory. I remove all taxonomy not reach 7 levels.

The deblur plug-in is included in the early step of my pipeline, exactly with the purpose you describe. The command denoise-other of the deblur plug-in require a reference-sequence, and the examples that I found use the reference sequence employed in the classifier building.

Thank you for the suggestion of the “Moving Picture tutorial”, I try also this pipeline.

best
Federica

Thank you Nicholas for your reply!
Using the input file of the feature-classifier, the deblur denoise-other doesn’t return errors.

Best
Federica

1 Like