I’ve been having some problems with classify.seqs() and large training sets in mothur 1.15. These same training sets used to result in mothur 1.12 using about 2GB memory, and everything functioned great. Now, when it gets to the “Calculate template probabilities” step, instead of climbing to 2GB memory, it climbs to 11GB memory plus another 23GB in virtual memory. It crashed the computer a few times, and caused a restart. (Mac OS X 10.5) I’m not sure what has changed in handling memory from 1.12 to 1.15, but it must be a bug of some kind I imagine.