cluster.split and unifraq

My computers are currently busy so I’m planning future analyses. I’d like to try unifraq of the sequences (al la “mothur > unifrac.unweighted(tree=final.phylip.tre, name=final.names, group=final.groups, distance=lt, processors=2, random=F, subsample=4419)”) but the only way that I’ve been able to build my sequence distance matrix is using cluster.split (after following the SOP I still have 300k preclustered “uniques”). Is it possible to use clearcut to generate a nj sequence tree on a distance matrix that doesn’t have overlap between phyla?

Ok so I selected representative sequences for each 95% otu and am trying to build a tree with clearcut, (65k seqs). it’s been running 4 days but hasn’t crashed