We have at least 2, but likely more, datasets that are conglomeration of runs that have different read lengths. I have to process them separately with dada2 to maximize each dataset when running dada2-denoise-paired. What would be your recommendation for merging? Ideally, I would like to merge them immediately after the denoising so that I don’t have to mess with processing both datasets separately, but wasn’t sure the best way to concatenate that won’t introduce some variable that leads to issues downstream.
That should work, assuming that each run has the same forward and reverse primers (and are all paired-end). So you should be able to just directly merge your feature tables with feature-table merge.
If you don't have the same primers for every paired-end run, then you need to analyze things separately because your sequences would not be directly comparable.