Hi @Alban_OTT,
It is definitely true that you will sometimes see organelle rRNA sequences in your data, and they will often be classified during taxonomy classification. For example, in Greengenes, mitochondrial rRNA are often classified as:
k__Bacteria; p__Proteobacteria; c__Alphaproteobacteria; o__Rickettsiales; f__mitochondria; ...
and chloroplast rRNA are often classified as:
k__Bacteria; p__Cyanobacteria; c__Chloroplast; ....
(Chloroplasts descend from Cyanobacteria, and mitochondria from Rickettsiales - hence their placement in these lineages in this taxonomy.)
In general, these don’t seem to cause much of a problem, and I don’t consistently remove these sequences from my data sets. In some cases (for example, in this study) since plant surfaces were being sampled we had an overwhelming amount of chloroplast sequences (somewhere around 50% of the sequences were chloroplast sequences, if I remember correctly), and we filtered these from the dataset based on their taxonomic assignments.
In the next release of QIIME 2 we will be improving support for removing these types of sequences based on their taxonomic assignments. We can follow up with you on this post to let you know when that is ready.