With regard to a request from Eric Martz (emartz@microbio.umass.edu),
        recently, asking if anyone had a library of flow data files from a
        variety of cytometers, I sent the attached reply. He suggested I
	send it also to "cytometry@flowcyt.cyto.purdue.edu" as well, where it
	might generate additional responses and, perhaps, more interest in the
	subject. 
----------

- From robb Mon Jul 26 15:12:22 1993
To: emartz@microbio.umass.edu
Subject: Re: Illustrative data by ftp
Cc: jet@evangeline
Content-Length: 2043
X-Lines: 38        

	Eric

	Having a repository of typical files from various systems is a good
	idea! I am also interested in testing my program IDLYK, (written in
	IDL), with all of the more common flow data file formats. It's very
	clear that the FCS 'standard' has been interpreted and implemented
	in a number of different ways. Perhaps one outgrowth of having a
	common source for files of various persuasions will be better con-
	vergence on a single interpretation of the standard. I will add a
	few files to the library once one is set up. 

	The same general idea might be used to set up a library of routines
	which perform specific functions in the various computer languages.
	For example, there are surely different ways of calculating means
	and CVs, or performing the K-S test, but some are more robust or
	efficient than others. This might eventually serve as an appendix
	or addition to the Data Analysis & Computers section of Robinson's
	"The Handbook of Flow Cytometry Methods".

	Parenthetically, this would also constitute a first step in the
	self-validation of flow cytometry software by the user community
	itself. Feeding back on the proposal to set up a library or
	collection of 'good' flow data files, I would suggest the following:
	The information deposited along with each file should include entrys
	regarding the number of cells positive in the various parameters
	along with means, CVs, standard deviations, etc.. For files with
	cell cycle data in them there should be a summary of results of the
	fitting analysis (G0/G1, S, etc.) and the program used to obtain those
	results.

	I would be interested in helping to set something like this up. There
	is a lot more software available now than ever before, but there isn't
	any assurance that it actually does what it's supposed to do. A group
	of data files, with quantitative information on their contents, will
	at least provide the means to compare the performance of one analysis
	program with the results produced by another. 

	Robb Habbersett - robb@big-geek.lanl.gov