Following trimming, my forward and reverse reads seem to be very high quality except for at two positions (25 and 143) in the middle of the forward read:
So for the DADA denoising step I ran the following command without truncation. Will the max error parameter most likely resolve the two base pairs (25 and 143) that are lower quality? Or should I probably use --p-trunc-q 20 or 30 to get rid of reads with lower quality data?
Additionally, I noticed the min length of my representative sequences is 95 when the amplicon size is normally ~390 bp. Is this normal or should I try to filter out the shorter sequences? It seems my reads are good lengths so I am not sure why they would be merging to form reads that are so short?
These stats are very solid, you're retaining ~70-90% of your reads which I would say is above average. I wouldn't bother explicitly trimming low quality bases, the dada2 algorithm is good at taking quality scores into account. If you want to length filter your sequences you can do so with the --p-trunc-len-* parameters, but this isn't necessary and will make you lose bases that you don't need to. That being said I am surprised you have a merged sequence that's only 95 bp... you can always blast sequences you're suspicious about and use qiime feature-table filter-features to remove them if desired.
