My lab and I have performed 2 separate MiSeq runs and have been using DADA2 for our 16s analysis. I am curious if analyzing all of the samples (one run contained 16 samples as a test pilot and the other contained 96 samples) from the two runs together at the same time with DADA2 could cause any problems. For example, does the error model generated by DADA2 change substantially between runs on the same MiSeq instrument? Could the samples from each run generate cross-talk that could impact downstream analysis or accurate identification of ASVs? Could it be impacting results in some other way that I haven’t even thought of yet? Thanks so much for your help!
Since batch variability is a real possibility even on the same machine I would advise denoising them separately and combining them after. I would think its possible to run them together if the runs were very similar to each other, but it’s probably just safer to do them separately. There’s a big list of technical issues that MAY occur between the runs, including contamination of reagents, barcode jumping, residual amplicons, contamination etc.
In order to merge them you would have to use the exact same denoising parameters between the two runs though.
If you you are looking to be combining a lot of projects in the future, you could also use deblur to combine them and run them together since deblur uses a static error model. Its a bit more convenient in that regards.