Select genomes to sequence by NSTI

[zaneveld]

One enhancement we've discussed for the next PICRUSt release is using NSTI values to inform strain selection in sequencing. This would be a new script inside PICRUSt that greedily selects genomes for sequencing in order to lower NSTI scores for a given environment, represented by an OTU table.

Required Inputs:
-- Tree (assume Greengenes by default?)
-- Count Table (needed only to know which strains have information).

Optional Inputs:
-- Target OTU table. If provided, look to minimize NSTI vs. the organisms actually present in the OTU table.
-- File providing a list of available strain ids [e.g. if someone is looking at a culture collection]. If not provided, assume all tips can be cultured (will be much slower).

Outputs:
-- Greedily selected set of strains for sequencing with estimated cumulative reduction in NSTI per strain.