After creating a rooted tree from my original feature-table I decided to filter samples out of the feature-table. I went ahead and used qiime feature-table filter-samples . I ran qiime feature-table summarize on my new feature-table and those samples are not there-so far so good. I then ran qiime diversity core-metrics on my newly filtered feature-table and made emperor plots. However, when I looked at the emperor plots those samples that had been filtered out of my original table are still present. How do I get rid of those samples?
Hi @pjtorres! Are the Emperor plots identical when you compare filtered to unfiltered? It sounds like the filtered table may not have been used for some reason, or perhaps an existing output directory was used that had unfiltered Emperor plots in it.
Can you please verify again that the filtered .qza file is filtered as expected (e.g. using qiime feature-table summarize) and then rerun core-metrics on the filtered table? Please supply a new output directory (one that doesn’t exist) to core-metrics. Do you still see the same results in the Emperor plots?
Can you please verify again that the filtered .qza file is filtered as expected (e.g. using qiime feature-table summarize) and then rerun core-metrics on the filtered table?
Yep, verified again and yes it was filtered properly. I even went ahead and took my filtered feature-table and exported my data as you recommended [here] (Is there any way to summarize taxa plot by category?) and then ran beta_diversity_through_plots.py on qiime 1 and my filtered samples are not there- everything looked good.
Please supply a new output directory (one that doesn't exist) to core-metrics. Do you still see the same results in the Emperor plots?```
They look about the same, but are not exact. The samples I wanted filtered include sample 29,112,46,55 and 56. When I open my emperor plot, go to the Visibility tab and click on scatter: Description I still see those samples in the emperor plot with the original NoFIlterbray_curtis_pcoa_results.qza-emperor.qzv (743.3 KB) and filtered feature-table FilteredMFbray_curtis_pcoa_results.qza-emperor.qzv (749.6 KB), but not on the emperor plot generated on qiime1.
Thanks for providing those files @pjtorres! It looks like your filtered metadata file has 6 fewer samples, and the Emperor plot also has those 6 samples filtered out. For example, the unfiltered metadata file has sample ID 112, and that sample is missing from the filtered metadata file and Emperor plot.
I noticed that the Description column has sample ID 112 labelled as 17 (in the unfiltered file), and sample ID 174 is labelled 112 in both filtered and unfiltered files. Is it possible your sample IDs and Descriptions got mixed up at some point? Inspecting the sample ID column in the Emperor plot will show you what samples are present.