I am looking to clarifying how merge-seqs and feature-table merge work for my situation. I have sequence data from different runs, but overlapping sample IDs. Each of these datasets have been demux-ed and deblur-ed separately.
I found that the sequence quality of some samples were better than the same samples in other runs. When merging sequences and tables, will samples with better sequence quality overwrite those that are worse? Otherwise, what is the best approach for this?
Not at all — the quality score data are not retained after denoising so there is no way to do this.
When merging tables with overlapping IDs, the matching sample IDs will sum together. That is the point of this command: to merge overlapping samples/features, not to somehow choose which is the better of the two.
If you do have some preference for samples from one table or the other, you should filter out the low-quality samples manually before merging.
When merging sequences, quality does not matter — the ASV feature ID is unique to a given sequence so you do not need to filter the sequences before merging; instead merge all sequences, then filter out sequences that do not appear in the merged table (i.e., if you do the pre-merging sample filtering that I described above).