I have run Indicator analysis in R and would like to create a heat map using feature-table heatmap with a feature table that contains only the indicator sequence variants. I tried running feature-table filter-features (keeping the indicator sequence variants) but the output table has does not contain all my samples. Is there a way to keep all samples in the output table? If not, is there a way to specify which sequence variants to include in the heatmap via feature-table heatmap?
filter-features should not drop samples, only features. Are you sure the input table contained all samples? Please run qiime feature-table summarize on the input and output tables to check — if you are still having trouble, please post those QZVs, the filtering command that you used, and your indicator species list here so we can help troubleshoot.
Thank you Nicholas for your quick response. I've attached my input and output table summaries (created via 'qiime feature-table summarize' and my indicator species list.
The filtering command I used:
$ qiime feature-table filter-features
--i-table rarefied_table_e1185_Acropora_millepora_remove_unhealthy.qza
--o-filtered-table rarefied_table_e1185_Acropora_millepora_remove_unhealthy-INDICATOR-TAXA-ONLY
--m-metadata-file Indicispecies_Acropora_list.txt
The input table has 66 samples and the output table has 23 samples. qiime feature-table filter-features says "Any samples
with a frequency of zero after feature filtering will also be removed." Some of these indicators likely have a frequency of zero in some samples (making them good "indicator taxa" of my treatment groups).
I apologize for multiple correspondences but I discovered that some of the features have an extra character "X" at the beginning of the feature id. I corrected this and ran the following filtering command:
I apologize — I was mistaken and you are correct. I have raised an issue to address this (e.g., make dropping empty samples optional) and I agree this could be bad behavior in some circumstances.
For now, I'm not sure there is a great workaround that is 100% in ... the best approach is probably to export your feature table and generate something in R, where you can have more control to customize this plot. Sorry I don't have a better solution at the moment!