feature counts by sample? and how to do correct rarefy

Hi @moonlight,

Unless you are using the taxa collapse action, when you assign taxonomy by the typical methods in feature-classifier you are simply creating a new taxonomy artifact, not changing your feature-table. This is an important concept to be aware of in QIIME 2, in that you are keeping these objects separate.

Generally speaking, removing taxa from your feature-table without a very specific reason is not a good idea. I typically do all my analyses at the ASV level and call their taxonomy when needed and avoid collapsing my ASVs down to taxonomy since you lose some resolution there.
Which taxa do you not want? Is there a specific reason why you need to remove some? I ask because, this can bias your results and is not a typical approach, unless you are removing something like chloroplast or mitochondria.

Yes, but you don't get to control that overlap region, rather that is dictated by the primer sets you used. If the primers flank a region that are 550 in length and you have a 2x300 bp run then yes that would mean they theoretically will merge with 50 bp overlap.

The option of justConcatenate is not available in Qiime2 as there is in native DADA2 in R, and I would advise against doing something like this in general. See this recent thread on the topic. ITS analysis has some specific considerations that you need to be aware of, see this ITS tutorial for more details. If you are failing to merge these reads, you may consider just using your forward reads only.

Please see my initial answer from above:

In other words, you already have this table in this format within feature-table.qza.

Again, answered previously:

So, this is a summary of your feature-table, and in fact this holds some of the information you were asking for. For example: L4S137 has a total frequency count of 9820 across however many features. You can think of this as the total number of reads of that sample, not total number of unique features.

No, the Feature Count value in that table corresponds to sum of all counts across all features in that sample. Yes, you can think of features, ASV, and OTUs as the same units, though they are produced differently.

Sure, like I said this feature-table in QIIME2 and OTU table in QIIME 1 are the same.

Yes this is in fact the only definition of rarefying and the plugin does exactly that. It randomly throws away reads (regardless of their feature designation) from each sample to reach the set threshold.

Why? I'm not following this logic.

By using the feature-table summarize action as you have before. This is exactly the info you need.

I think this is again stemming from your misconception of what Feature Count is referring to in that table. See above answers.

A couple of additional comments, a lot of the answers I've provided here are easily found in those tutorials that you have seen, again I would advise you to read through those carefully. Also, it is always a good idea to search through this forum first with regards to questions you have, in that it is very likely that the same or related question has been answered already. This also brings me to my last point, in that, in the future please limit your questions to one per thread, or at least keep all questions related that topic. If you have additional, unrelated questions, please start a new topic for those. This helps us keep the forum organized and easy to search through for questions.