I’m currently working on a 16S soil microbiome dataset with ~140 samples using QIIME 2 (qiime2-amplicon-2025.7). During the DADA2 denoising step, about 25% of my samples end up with 0 reads after filtering.
I’ve already experimented with multiple trimming and truncation parameter combinations (trying to avoid overly aggressive filtering), but the issue persists regardless of the settings I use.
Do you have any ideas about what might be causing this? Could it be related to sequence quality, primer removal, or something upstream in the pipeline?
For context, I’m running the analysis on WSL within an institutional cluster (64 cores), so computational resources shouldn’t be a limiting factor.
Any guidance would be greatly appreciated—this is part of my master’s thesis, so I’m trying to troubleshoot this as thoroughly as possible.
Is it Illumina? I don't remember seeing something like that - it looks like each read (forward and reverse) was sequenced 2 times by 150 nt and then just joined...
What were the steps before DADA2? How were raw reads sequenced? Were any additional manipulations performed outside of Qiime2?
I think that those quality plots are the main reason you are struggling:
Some samples lose all the reads at the filtering step (based on the length or quality)
Other samples lose a lot of reads because they are marked as chimeric (not surprising based on the quality plots).
