Thought it might be time to contribute a Nickle's worth here: $0.02 and
then some for a couple of different threads.
Qualifications/Training for a Molec.Bio.Comput.Resource Manager:
I went through a typical science training/employment path: BS,Chem;
PhD,Biochem; Postdoc; 11+ years on the research faculty at a med school. None
of my work ever involved wet bench molecular biology, but I became something of
an electronic molecular biologist by developing an interest in using and
supporting (as something of a local 'guru') the GCG package when it was made
available in the environment. Throughout most of my science 'career' I had
opportunities to use one type of computer or another and developed some
programming skills which were useful in data processing. At one point I took
some formal programming courses and then other courses from a Computer Science
Dept (which convinced me I was just a competent computer user, NOT a computer
My experience to date suggests that a MbCR manager must be able to
'speak the language' of the research scientist. In my previous life, I once
was discussing protein sequences and sequence analysis software with our VAX
System Manager. At one point, he stopped me and asked, "What's a 'residue'?"
It never dawned on me that he wasn't following my comments because of a
'language gap'. More often than not, discussions with research scientists here
at UGA have occurred in a similar fashion: they don't think to define all the
terms/acronyms/etc used (nor should they have to); they simply discuss things
in a way that presumes I'm 'with it' enough to understand what they're talking
about. Fortunately, in contrast to a system manager w/ no bio background, I
usually manage to hang in there.
I don't really know whether there's an advantage to the way a MbCR
manager is trained (ie, comes to know what needs to be done and how to do it).
At this point in time I'd guess there are a number of examples of managers who
are 'fugitives from the lab' (From today's posts it's nice to see that there
are managers with Computer Sci backgrounds; good to know this game CAN be
played more than one way). However, I do think the molec bio computer support
game has enough significance to warrant the establishment of a true training
program and a career path for interested individuals. (I wont claim this as a
unique idea as I think it has appeared on the net in the last year or so.) I'm
not sure that training necessarily be deeply rooted in CS (being a 'deep
thinker' re algorithms may not be essential for the management tasks of a
MbCR), but given the relevant range of tasks (user management/training/support,
system support/development, software/database maintenance/updates/etc), a well
trained manager should come out of a program being reasonably bilingual:
speaking 'science' and 'computers'.
I don't think I'd be as competent at system management today if I had
not been tutored by a system support person who largely spoke only 'computers'.
This was an integral part of 'training' that only began after I took my
current job. It also provided me with colleagues on the computer side of
things and futher bolstered bilingual capabilities. At this stage of our
facility, we continue to benefit from access to University system support
expertise while having a science/computer person 'at the helm'. Thus, our
scientists don't have to worry about the details of having the facility up,
running, and current - because that's my job [you didn't think I recommend
firing myself, did you? :^) ]. They hopefully also feel that they have someone
with whom they can effectively communicate in order to obtain help, to ask for
new capabilities, to suggest directions for facility development, etc...
One other observation in this topic: 'system managers' can have their
jobs evolve to where the wearing of many hats is almost required. So, when
we're currently discussing the who and what of a system manager, at some point
maybe it would be worth thinking whether the job can really be done by only one
How to do a MbCR: PCs/Workstations/Minis/...
We run a local area VAXcluster, have the GCG and IG molec bio
packages, and a variety of other software as well (BMDP stats, Phylip, Biosys,
Paup, other popgen/molec evolution programs). This setup, which was chosen
to some extent because GCG was only available for VAX/VMS when we began,
provides for a centrally managed, distributed computing facility in which the
availability of CPU cycles (for DB searches, multiple alignments, etc) and disk
storage can be easily modulated. (Yes, I know there's the ability to 'cluster'
Unix systems, but my impression from colleagues working on UGAs RISC/6000
setup is that a VAXcluster is more easily managed and more stable.) So far we
haven't had to place any limits on who can do DB searches, how many they can
have in the queue at any time, or when they can do them. With about 4.5 GB
disk space, we're holding our own as far as having storage for user, software,
and database files. Acquisition of additional disk space continues as a
priority as does the upgrade of hardware to support what we do. Our facility
pay its operating expenses by charging subscribing laboratories $1000/yr, which
in these times of hard-to-come-by grant money is being viewed with increasing
frequency as being 'too much'. The philosophy of the facility's Faculty
Steering Committee is to apply for grants or to obtain other funds that will
support major hardware upgrades. Reducing subscription fees is a goal, but the
method of doing this remains to be determined. Some choices: obtaining
additional subsidy, reducing range of available software, scaling back to
hardware which might be less capable or not as available to as many
The above should not be taken as a sales pitch for the way to do
things; it's just notes on what we have, how we work it, and some of the
considerations involved with how it operates. In a recent thread of this
newsgroup, there have been discussions of pros/cons re lab/dept/central
facilities. What should be clear is that there isn't only one way of getting
compute power for the molecular biologist. If you have the need, you
can find a solution given the time, money, and personnel you have available.
Is there an _optimal_ solution? Well, there's still 'the grass is always
greener...' problem: I know of groups who tired of the PC-based PD/commercial
software route and moved to a supported central facility, as well as labs who
experinced access to a central facility and, deeming it too expensive, looked
to PCs and the internet for the source of their compute power. I don't know
if it's reasonable to expect that there be an optimal way. After all,
molecular biology continues to evolve, computer and software power continue to
increase, and databases and what we want to do with them continue to change.
Doesn't that guarantee that a _best_ solution to providing/obtaining computer
capabilities will always be a moving target?
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _
/ Michael J. Weise, Ph.D. \ Univ.of Ga. BioScience Computing Facility \
( weise at bscf.uga.edu \ Dept.of Genetics UGa, Athens GA 30602 )
\ _ _ _'Tis_only_me_speak'n._ _\_ _ _ _ _ _ _ (706) 542-1409_ _ _ _ _ _ _ /