building really huge trees

I’d like to look at faith’s d for a huge soil dataset (275k otus following the 454 SOP). I’ve managed to build one giant distance matrix of the OTU rep sequences that is 338gb on our lab server which has 128gb ram. I’m trying to clearcut (I’m not sure if clearcut has to read the entire dist file or not) it but don’t have high hopes that this will work. I’ve considered building phyla trees (I used cluster.split to generate OTUs to begin with) but the dist for the proteos alone is ~200gb. Are there other options out there for looking at phylogenetic diversity on these big datasets?

You can try to shove it into clearcut, but I wouldn’t have high expectations.