I have run a same set of data with 2 different qiime2 versions: 2018.8 and 2018.11 and i got quite different results for the taxonomy classification. I have tried the classifier produced by both 2018.8 and 2018.11. Therefore, i think the error comes from dada2 files. For the results from 2018.11, there are some remarkable increases of some taxa in some samples which are not the case in the results from 2018.8. I am wondering how it can make such a difference.
I am really confused. Could anyone help to explain this problem?
I think the defaults in DADA2 may have changed between those versions (although I thought that happened earlier last year than August…)
Provenance would be really helpful here. Could you provide either the taxonomy files or some visualizations using those taxonomy files? Then we can see if there was any significant differences in parameters passed (either by default or by argument).
@Toan,
Thanks for sending those. You classified your sequences using entirely different reference databases (or rather, reference sequences clustered at different % similarity) — that would have a massive impact on classification results! Check out the data provenance in your QZVs for more details.