This site at http://www.bios.unc.edu/~fwright/
is intended to replace the data supplement site for the Lemon et al. Bioinformatics
paper, formerly at http://thinker.med.ohio-state.edu,
which is temporarily down.  The data link on this page is a big zipped file
(150Mb) with essentially all the relevant files within it.  I am sorry that it 
is not very well annotated, but if you sort by file type you can see e.g. 
the Word files that contain supplementary appendices and the
longer manuscript LEMONWJBioInfFinal as a pdf file , etc.  
The CEL files are in a zipped file inside the big file,
and our LWF expression estimates are in a text file lwfBioscdata, where
"lwf" stands for Li-Wong full and "Biosc" refers to the so-called bioscaling
we used.  The Affy MAS 4.0 results are text files with names like
fbss.MAS4AD, where "fbss" stands for "fetal bovine serum stimulation" (our
abbreviation for the study), and "AD" stands for average difference.

Perl scripts have extension .pl, and Splus scripts extension .s.
William Lemon wrote a C program that can give the LWF estimates much faster than
the glacial pace of the others, but I don't have the source code.  I have an
executable for it that runs on Linux.  I have all the .DAT image
files as well, but they are too big to put on the site.  Let me know if
you want them (probably not).

Tab-delimited text files for the original CEL data are somewhat more 
convenient than the original CEL files, if you wish to deal with the probe-level
data.  They are in the zipped files as Stim.1, 5050.1, etc, and should have
self-explanatory column headings (PM1, MM1, etc. for perfect match and mismatch
probes).  I clipped it off at 20 probe pairs, as almost all probe sets had 20
probe pairs, although a few have fewer and a very few have more.