We have installed two qiime2 with different versions which are 2017.12 and 2019.10 by conda.And we used the same classifier file supplied by yours.Interesting, we used the same 16S sequence files but we get total different taxonomy tables. The taxonomy table from qiime2-2017.12 showed about 20 genus but which from qiime2-2019.10 only showed 3 genus.WHY?
Welcome to the forum, @hyystvar!
As in you processed the same RAW data in each version? or you processed in one version and then classified taxonomy with both versions? I suspect you mean the former, in which case this is most likely an issue with, e.g., changes to default parameters for denoising methods, not an issue with the taxonomy classifiers themselves.
Much has changed since the end of 2017… and changing defaults in, say, dada2 will impact how reads are denoised and merged, impacting downstream parameters. So using the same exact processing parameters that worked with one version may not necessarily work for another… you must always pay close attention to the outputs at various control points, e.g., sequence yields after denoising/OTU clustering.
None of this is to say that there are bugs in one version or the other, only that much has changed and a pipeline that worked in 2017.12 may need to be adjusted for 2019.10 even for the same data… in theory, with careful attention to parameters and outputs at each processing step, you should be able to more or less replicate the 2017.12 results with 2019.10.
I am guessing you are re-processing data because you originally analyzed with 2017.12 and now want to reprocess since you are getting close to publishing or use these data in another way? 2017.12 results should remain valid, and forward compatible with current releases (e.g., if you want to apply methods that have been released more recently) so re-analyzing with 2019.10 should not be necessary.
I hope that helps!
This topic was automatically closed 31 days after the last reply. New replies are no longer allowed.