Is pre.cluster still necessary for MiSeq data?

Considering the lower error rate of MiSeq sequencing compared to 454, why is pre.cluster still a recommended step in the SOP? Why still keep it at 1 difference per 100 bases?

Has anyone actually tested the effect of pre.cluster on a mock community sequenced using MiSeq?


The MiSeq error rate is about the same as we were getting by 454 with shhh.flows. So we do use pre.cluster. See the Kozich paper for data.