It strikes me that chimera.uchime() is not utilizing multiple processors? It would be nice if a uchime job was forked per sample limited by the number of processors specified.
It should be. Can you post the command you ran?
I am following the MiSeq SOP step-by-step using 20 processors on a dataset with 78 samples. Inspection with top in Linux shows a single mothur/uchime process. The command was:
chimera.uchime(fasta=jt.trim.contigs.good.unique.good.filter.unique.precluster.fasta, count=jt.trim.contigs.good.unique.good.filter.unique.precluster.count_table, dereplicate=t)
mothur v.1.32.0
I can confirm it forks as expected with mothur v.1.32.1 if you add the processors=20 argument.