Hopefully multiple users will speak up on this topic so you can learn from as many perspectives as possible, because this statement is very true:
Sure enough. I will try to describe a few ways:
Taxonomy classificaiton is pretty much the same as for 16S (it’s the upstream steps that differ)… q2-feature-classifier has multiple different classification methods and they all work quite well for fungal ITS… see this paper for details (adjusting defaults can improve precision for fungal classification):
I wrote this tutorial a couple years ago and I am not sure much has changed, so I still recommend this basic workflow, but let’s see if others propose some upgrades:
I recommend the latest version… whether to do fungi only or all euks depends on how many non-fungal reads you expect. In general, at least a few plant (or other non-fungal) ITS reference sequences is enough to provide an outgroup for identification of non-target hits.
We actually benchmarked a few different versions of UNITE recently, take a look here for details:
YES, definitely! This is a big issue with ITS… the hypervariable length means that read-through can be a common issue (depending on your read length), so you need to trim the forward primers AND sometimes the reverse primers and adapters that appear in the reads (for more details, this has been discussed in other topics on this forum, maybe also the tutorial linked above).
There are two main ways to do this in QIIME 2:
q2-cutadapt can be used to trim at primer sites (forward and reverse)