This may be unfeasible, but I am noticing that when I use cluster.split with splitmethod=classify, some of the temporary files have the cutoff threshold changed to 0.01 or 0.02 from 0.15. It would be great if I could set a parameter so that, for most data, I can say that I won’t accept a threshold below 0.03 and have clusters for most of the temporary files made at 0.03. For the taxa where the threshold is changed to less than 0.03, perhaps these taxa can be pulled out and handled differently so that the clustering threshold that is actually desired can be made (for example, give the option to make distance matrices with threshold 0.03 and cluster this way? Or take the most abundant sequences as OTU seeds and map other sequences to them? I think that’s how USEARCH works?).
I like using mothur, but I’m a bit annoyed with my computer’s inability to handle cluster.split without a cutoff indicated, and when I indicate a cutoff it always is made to a distance that is too narrow. I don’t mind using OTUs at a 0.02 distance but my collaborator is upset by it. I’m trying everything I can think of to try and get clusters at 0.03 without crashing the computer, but so far I haven’t had luck.