Mothur splitting an OTU in a strange manner

Hi everybody,

I have something which I can’t really understand happening with my data…

I have a dataset that I’ve been analyzing with both QIIME and Mothur. What really puzzles me is that I realized that two samples which in the QIIME analysis were completely dominated by the same OTU (and when I went back to the sequences just after the denoising and I aligned them the unique sequences forming that OTU were 100% identical between the two samples), have now been split by mothur in two different OTUs so that sample A has only OTU 1 and sample B has only OTU 2 ! The total number of reads for that OTU found in the two samples are very similar between the two analyses, so I assume that more or less the two denoising had the same sequences passing that step… also all the rest of the OTUs and their patterns of distribution and abundance across the different samples are very comparable between the two analyses.

What could have happened and what could I do to further check on this?


Unfortunately, there’s a bazillion variables that you could be using and so it’s pretty hard to comment. One suspicion is that QIIME is merging things in the denoising step that mothur isn’t merging. Also, depending on when you look at the sequences, things might be merged at different stages. If you can generate a simple case that generates this result, I can take a look.