filtering when there's a bimodal distribution of pairwise distances

[nhoffman]

We noted that many Fusobacterium nucleatum sequences are being lost in 'deenurp filter-outliers' - one possibility is that there are two distinct populations and we lose the smaller one. First thing to follow up is:

• get all F. nucleatum sequences from rdp named (there are ~150)
• align with muscle
• make a tree - are there two distinct clades?

If this turns out to be the issue, will need to consider whether to treat each subpopulation separately (ie, identify centroid and draw radius for each).