Hi, my last post on the issue of analysis of proteomics data ended on a
pissy note about problem of a novice analyzing data.  I apologize-I'd just
gotten done with a rather long meeting on proteomics profiling with someone
who wanted our group to contribute thousands of dollars to one of these
projects, as a "collaboration", thinking his lab could do little except make
the samples, once.  

In honesty, I must admit we have been looking at other software for protein
profiling from shotgun proteomics data that is available, and we also find
most of it lacking when comparing large datasets.  We have some software
(IsoformResolver, first described in Resing et al, (2004) Anal chem), that
was designed from the beginning for this type of experiment, and has been
tested with 1-2 million DTA files.  It does a good job of reporting the
isoform ambiguity and uses spectra counts to provide a convenient comparison
matrix that can handle a very large number of samples (we've tested it on
40).

We also have several good ways of validating peptide assignments, if you do
not have your favorite method.  for comparisons, I usually run at 3-5% FDR,
but a larger FDR can be tolerated in comparisons, because the problems go
into the set of protein hits that have higher CVs anyway.

I can help people out with analyzing that data sitting on their desk, for a
small facility fee to cover the cost of the person who will be sheparding
the data through the system, and can provide a consulting fee of $50/hr for
academic people, to help them understand how to analyze the data.  If you
like the software, it is freeware, but will take someone with some
background in running prototype type applications, to set it up in your lab.

If interested, I can send a couple of PDFs of use of our software in
analysis of datasets.

Katheryn resing