I'll just take a second to add my $0.02 to the discussion...
In my experience (Running a sequence analysis "facility" in Industry)... It is
better to try and support a limited number of packages (GCG and IG in our
case) on a centralized system (a vax cluster in our case) than to try and keep
up with what users want and need on a micro based system. Working with a
centralized setup allows the system manager to keep data as up to date as
is possible.. the key to a sequence analysis facility if you ask me. Also,
you can keep the software as up to date as possible as well, without having to
deal with a large number of pseudo-managed micro's.
I think that the responsibility of the Sequence Analysis facility manager is
to 1) provide the most up to date data, and 2) the tools to work with these
data. Other than that, the user can use his Mac or PC based system to draw
pretty pictures and the like, but he(/she) can get to the data on the
centralized system with standard tools. The system manager can also, if they
choose, provide databases in formats for micro based packages (boy, i wish
they would provide the tools for reformatting these databases into their own
proprietary formats...) and distribute them on the centralized system.
If I were to set up a facility today... I think that I would start with the
GCG package, running on a high powered unix workstation, perhaps a multiuser
IRIS or the like. I would then concentrate on delivering the data to the
system in as timely a manner as possible, and then (in my spare time) work on
a more user friendly way to help users with the package, and work in some nice
additions to the package (Blast3, clustalV, and others). I would try NOT to
work with any micro based packages. You are just asking for management
troubles that way. After all that, you can still put up such niceties as GDE,
PIMA, and other tools that users would probably love to have access to.
Unfortunately, a vax based system won't have as many nice tools to
incorporate, but then you won't have to spend your time working them in
either :). Also, with the unix based package, you can have (a limited number
of) people access the system using Xwindows, by running an xserver on their
Macs or PC's. A nice advantage for a small facility.
As far as the system manager goes, I agree, most of us are Molecular
Biologists that have "coverted". I don't think that you will find many a
system manager that can "pick up" enough mol bio to really help out the
scientific users of such a system. I also think that you will probably have
to look pretty hard to find a competent person to do this from the start, or
be prepared for a couple of years of on the job training (which can be
painful). I also think that it is important for the system manager to work as
a sort of "power user/consultant". I would venture that most of us have, from
time to time, provided research guidance, or advanced sequence analysis for
programs, on an "as needed" basis. Putting together a well oiled system, and
working with its tools provides you with a knowledge of the tools that most
"power users" envy. This is an added incentive to find that converted
molecular biologist that is computer savvy.
Well, that was more like $0.10, but what the heck.
elliston at msdrl.com
Keith O. Elliston elliston at msdrl.com uunet!av8tr!elliston
AA5A/U N9734U elliston at mbcl.rutgers.eduelliston at biovax.bitnet
"Beware of pseudo-experts with a mission and a grudge, especially if they are
lawyers pretending to be scientists." -- H.W. Lewis in 'Technological Risk'