Dealing with singletons and rare OTUs

Hi Patrick
The debate over singletons in 454 data (true diversity or sequencing errors?) is not yet settled. Recently I read one paper (Gottel et al 2011), where they removed rare OTUs by retaining only those that occured in more than 5 samples. I think this is a meaningful way, in the sense that all singletons were not treated in the sameway. Can we apply similar filter using MOTHUR?

I should add that I’m not a big believer in removing singletons. If you follow our SOP, your error rate will be very small. Although it of course is still above 0 and we aren’t able to remove every chimera, removal of singletons or rare OTUs seems like cutting your hand off to deal with a paper cut. The singletons contain information about how well you have sampled a community. More importantly, this approach gives the false impression that there is some abundance-based threshold between real and erroneous sequences. Remember - sequencing errors and chimeras aren’t always singletons…