Developing phylogenetic methods

(15 April 2009) These are notes for an ongoing project. Please let me know if you have suggestions.
-- Sam Riesenfeld

I am working on developing a phylogenetic method for metagenomic data. The basic idea is this:

* Build a reference tree with complete sequences.
Iterate over the following two steps until all reads have been placed:
* Place one read on the tree, hopefully using pplacer (see note below).
* Re-estimate branch lengths and possibly topology using maximum likelihood 
(probably RAxML) for some portion of the tree containing the read just added.

Note: pplacer is a method developed by Erick Matsen while he was working with Robin Kodner. (See the summary of this method.) They haven't sent me a binary yet, though.

I'll add details as this develops. Right now it is really a rough idea.