Brownie is a program for analyzing rates of continuous character evolution and looking for substantial rate differences in different parts of a tree using likelihood ratio tests and Akaike Information Criterion (AIC) statistics. A manuscript describing the method appears in the May 2006 issue of Evolution. The manual is available here; see below for info on downloading the program.
There are two versions of the Brownie software. Brownie 1.0 is a set of MATLAB modules. It requires the Stats package for MATLAB and is no longer updated. Brownie 2.0 is a completely rewritten C++ program available as a compiled application for Mac OS X and also available as source code. Brownie 2.1 can read Nexus tree and data files and can perform analyses across a set of weighted input trees (such as trees weighted by posterior probabilities or bootstrap values) in order to deal with tree uncertainty. Note that Brownie 2.1 uses code (including the useful Nexus Class Library) from Paul Lewis, Rod Page, and the GNU Scientific Library.
The latest version was released on 10 March 2009, (3 months, 3 weeks, 2 days ago)
Note that the compiled version uses a faster Nelder-Meade simplex for optimization than the most recent source code does. (*simplex2 rather than *simplex). The GSL 1.12 includes this faster version, but at least one user has found that it aborts on huge datasets (doesn't give a wrong answer, just makes the program quit). I've reverted the code to use the older version, and the next release will include the slower code.
Subscribe to brownie-announce (group for new program releases and bug alerts, extremely low traffic).
Strongly recommended for all downloaders.
Subscribe to brownie-users (group for help discussions with the program author and other users).