Re: repository for flow data files

Dave Coder (dave@nucleus.immunol.washington.edu)
Wed, 28 Jul 93 10:22:13 -0700

I have a fairly good synoptic collection of data files composed of
FCS formats (at last count about 30-odd variants), proprietary
formats, etc. of list mode and histogram data. Many of the previous
problems in reading data will disappear when everyone writing data
files adheres to the standard described in FCS 2.0. Despite this
panglossian musing, all the legacy data from previous years will be
around to cause problems and it should be useful to have some
reference material for verification.

As noted by previous posters, data files are one issue. Verified
analysis of the data is another.This means both statistical analysis
(do you get the same numbers as I do?), and more fundamentally, does
your picture look like the "real" one? Gross problems such as byte
order are quickly identified, but things like axis reversals on
2-parameter histogram files may not be apparent. Reference pictures
(device-independent PostScript or HP PCL files should have broad
usability) would solve this problem.

So the data collection begins to sound like the standard zoological
museum or herbarium: Type specimens are stored with complete
taxonomic descriptions, illustrations of important features, and
verification data by the collector/depositor. For future taxonomic
work, such reference collections provide specimen data traceable to
THE standard.[Just where in France is the meter bar these days?]

Of course the collection should be available over the Internet by
anonymous ftp. This will require some (not much) disk space, a quick,
multiuser computer with a reliable Internet connection, and somebody
to maintain the collection. Once the archive is assembled, the time
commitment of the Curator should not be large. For those not yet
with access to the Internet, the cost of making hardcopy (disk and/or
prints) may be prohibitive in the absence of a person whose time is
partially dedicated to the task.

The collection of data, provision of illustration files,
verification, and the writing of documentation and user notes, will
take time and effort. I'd be willing make a contribution.

Overall, it's probably an essential archive if a degree of
commonality of data exchange is to be attained.

Dave Coder
University of Washington
Internet: dcoder@u.washington.edu
tel. 206-685-3014
fax. 206-543-3480

And remember, the nice thing about standards is that there are so
many to choose from.