I come to you with a question regarding the feature-classifier plugin. I would like to know if there is a way to obtain, for each ASV, let's say the 5 most probable taxonomic assignments rather than the first one (together with confidence values). In concrete terms, I am looking for results formatted as follow:
I thought the --p-maxhits option of consensus-vsearch could achieve it but I must be wrong since the table extracted from taxonomy.qza still displays only 1 taxon per ASV.
I am interested in a solution applicable for any of the 3 available classifiers (sklearn, consensus-blast or consensus-vsearch).
Hi Ben,
As far as I know there is not a easy way to do this. It seems like you are wanting to get rid of the consensus part of the commands. The easiest, but most labor intensive way I can think of doing this is by doing the taxonomy by hand using something like web blast.
The issue is that none of the other qiime commands are going to recognize taxonomy in that format. So any further use of that taxonomy is going to need to be programmed/done by hand by you as well.
If you're up for some hacking you could probably piggyback one of the functions like _compute_consensus_annotations used by vsearch and blast. And have it spit out the lists of top hits.
