Taxonomy classification results are totally different from two analysis method

[yuanyuan12543 (Junli Zhang)]

Hi, I used classify-sklearn and classify-consensus-blast to do the taxonomy classification. But the output q2-feature tables are significantly different.
I am not sure if it is normal. Which one should I choose for further analysis?

Thanks!

Junli

[TurboQiimer]

Hi,
Please share more info:
Commands
Files/ Barplot screenshots
Qiime2 Version

Thanks
Qiimer

[yuanyuan12543 (Junli Zhang)]

The script for classify-sklearn is like this
#qiime feature-classifier fit-classifier-naive-bayes \

--i-reference-reads NCBI_blast_ref/sequences.qza \

--i-reference-taxonomy NCBI_blast_ref/taxonomy.qza \

--o-classifier NCBI_blast_COI_classifier.qza

qiime feature-classifier classify-sklearn
--i-classifier NCBI_COI_classifier.qza
--i-reads rep-seqs-dn-97.qza
--o-classification rep-seqs-97-COI_taxonomy.qza
The script for blast taxonomay is like this
qiime feature-classifier classify-consensus-blast
--i-query rep-seqs.qza
--i-reference-reads NCBI_blast_ref/sequences.qza
--i-reference-taxonomy NCBI_blast_ref/taxonomy.qza
--p-perc-identity 0.97
--o-classification rep-seqs-blast-COI-ref-BLAST-TAXONOMY.qza
--verbose
For both of them, I used qiime taxa collapse to merged the otu table and taxonomy. The feature tables are in the attachment. feature-frequency-detail.csv (698 Bytes) feature-frequency-detail.csv (991 Bytes)
Thanks1

Junli

[ChrisKeefe]

Hi @yuanyuan12543!
There’s a great discussion in this topic comparing QIIME 2’s various classifiers, with links to primary sources and a bunch of anecdotal information from developers/advanced users.

Take some time to read those resources, and if you have more specific questions, feel free to post them here.

Best,
Chris