The files fdr.R and fdr.m are R (or Splus) and Matlab (or Octave) code
for implementing the False Discovery Rate (FDR) methodology for
multiple testing.  In particular, the code allows one to increase the
power of the FDR method by estimating the proportion of alternative
hypotheses.  This code accompanies the article:

Ventura, V., C.J. Paciorek, and J.S. Risbey.  2004.  Controlling the
proportion of falsely-rejected hypotheses when conducting multiple
tests with climatological data.  Journal of Climate, in press.  Also
Carnegie Mellon University, Department of Statistics technical report
775 (www.stat.cmu.edu/tr/tr775/tr775.html).

The code includes the main fdr function and four helper functions
(fdr_basic,fdr_master,storey,prop_alt) which can be used on their own,
but are mainly designed as internal functions to be called by fdr().

To use the code in R or Splus, download the file and source it in
either language: 
> source('fdr.R') 
Then call the function fdr() with a
vector of p-values as the first argument.  Other arguments are
optional and are described as comments within the code for the fdr()
function.  The function returns the indices of the significant tests.

To use the code in Matlab, download the file to the directory from
which you call Matlab, or place it in a directory in Matlab's path.
Call the function with a vector of p-values and the FDR threshold
(which we call the qlevel) you choose.  Other arguments are optional
and are described through the help utilities and as comments within
the code for the function.  The function returns the number of
significant tests and the indices of the significant tests.

The code is also usable in Octave, but you need to put each of the
four helper functions in their own .m files.

The code was written by Chris Paciorek and is distributed under the
GNU General Public License, GPL version 2 or later.  Please contact me
at paciorek AT alumni.cmu.edu with bug reports or other comments.