Hi @Jasmine84,
we have discussed this topic a lot on the forum, e.g., here — it is a bit risky to decide that anything detected in a negative control is contamination (as opposed to cross-contamination from real samples). I would advise you to proceed with caution unless if these are obvious contaminants.
You can filter specific taxa from a feature table or from sequence artifacts as described in this tutorial.
You will need to use feature-table filter-features
to filter the feature table before constructing a taxonomy barplot.
We do have an open issue to support interactive filtering from within a barplot, but I am not sure there are any plans to implement this in the near future...
I hope that helps!