When to normalize data

Theory behind mothur
1

Dear All,

I am following Miseq SOP to treat my data, but I am curious when to normalize the data?

Should I normalize before cluster or after I get the OTU that I normalize the OTUs (sub.samples)? And, for the alpha diversity, do we also need to normalize reads among different samples, like beta diversity?

Thanks,
Rain

2

I subsample rather than normalize. summary.single (alpha diversity) allows for repeated subsampling just like summary.shared. I also make a single subsampled OTU table for analyses that only take species matricies

3

I would never recommend normalizing. As kmitchell states, just sub.sample to generate a subsampled shared file and then rarefy the original shared file in summary.single and dist.shared

4

Thanks Pat and Kmitchell.

Rain