I think it would be really nice if it was possible to subsample sequences per sample with replacement. That would be very helpful to put bootstrap values on community distance based dendrograms. Maybe you could add a sub.sample (replacement=T) option?
I was actually thinking of building 100 dendrograms based on let’s say unifrac distances that are based on resampled sequences with replacement=T. Then take these 100 trees and feed them into another software, maybe IQtree, and get bootstrap values for the nodes as well as a consensus tree from it. I think this is something you guys could also implement if you found it worthy. I am always missing bootstrap values on my distance trees and feel that adding these by subsampling sequences with replacement is a powerful way to get legit values.
Have you tried tree.shared (http://www.mothur.org/wiki/Tree.shared#subsample)? I think that if you use subsample and iters it will output a bootstrapped tree and possibly all of the replicate trees.
Yes I have seen it, but I thought it uses the subsample command which is without replacement. That means for the lower coverage samples I end up pretty much with the same sequences every time I subsample. This is no real bootstrap then and does not help that much assessing the uncertainty. I hope it is kind of clear what I mean and am happy to be corrected if wrong.
It would be perfect if tree.shared had a replacement=T option when the iters option is activated and if there were bootstrap values on the nodes of the output tree that come from bootstrapping sequences.