My question is about running the denoising step on individual runs. If I were to do this step on, for example 10 individual runs representing a longitudinal period of a year, would this yield better outputs for downstream analysis, rather than if I were to carry out the denoising step on all 10 runs pooled together?
What would be the negative implications of pooling it vs. individual runs via DADA2?
There has been lots of discussion on which approach to use, and there really isn't a great solution to pooling or not, you may have to change other parameters during pooling, and/or sequencing depth as discussed here.
I personally like to use --p-pooling-method pseudo --p-chimera-method pooled along with --p-min-fold-parent-over-abundance 8, or 16, but not higher, as mentioned here.
Thanks for the post links, there were very informative. I'll check out these parameters and compare their effects on the outputs with my current outputs (without these parameters). Thanks for your help!