Entrez/VMS (was: Don't be too high-tech!!!)

David Mathog mathog at seqvax.caltech.edu
Mon May 10 13:30:00 EST 1993


Sigh, here we go again with the X windows vs. terminals religious war.  
Being a service provider, you'll find me squarely in the "least common 
denominator" camp.  Sure, if you have a cutting edge computational problem
your solution may not be portable, but that hardly applies to a simple
query system like Entrez!  Also, there has been a thread in here that
slapping Xwindow software onto PCs and Macs makes them adequate Xwindows
servers.  Perhaps for some things, but I've found MacX, for example, to be
pretty unsatisfactory:  performance is often slow and jerky; arcane tricks
are often needed (like having to set root windows to be tiny to avoid
running out of memory before the desired rootless windows open); the weird
key combinations needed to simulate a 3 button mouse feel odd; and cutting
and pasting between the X application and native documents is generally not
possible.  On points 1,3, and 4 terminal emulators are clearly superior,
while for point 2 I will concede that considerable arcane knowledge is also
required for the terminals/emulators (baud rates, parity, and such). 

Let's not get ridiculous about GUIs.  NetEntrez is 100% text:  the queries
are text, as are the qualifiers and the results. If anything the GUI
interface makes it *more* difficult to put a query together because one has
to do several (to me) nonintuitive operations to formulate boolean logic. 
The GUI also keeps any terminal users from utilizing the software, which is
a serious fault at some sites.  In the present case, it isn't too bad for 
us because anybody who wants to can run NetEntrez on their Mac.  The
downside to this is if they decide to retrieve something using NetEntrez it
ends up on their Mac instead of on our VAX where the sequence analysis
software lives.  There is also more support work required helping people
set things up on their Macs than on one central machine, and it is 
somewhat difficult recovering these costs (for some strange reason they
won't let us charge them for connect time on their own machines :-).) 

However, the thing that really amazes me, especially considering the 
evident Unix bias of the program's authors, is that by enforcing a GUI
interface they have reduced the usefulness of NetEntrez as a research tool.
Imagine that Entrez had a "command line" interface (ie, like a standard 
Unix command). If that existed one could keep a set of scripts around, run
them at some reasonable interval (say, once a month), diff the output, and
so pick out new and interesting entries automatically.  Under Unix, VMS, or
any other command line operating system this would be trivial for a *user*
to set up. There is not a strictly equivalent GUI operation, but just
running the query from the GUI at monthly intervals would be much more
work, especially if the query returned a large number of hits or was 
complicated and so difficult to formulate.

NetEntrez is a pretty nice program.  Entrez's neighboring analysis, in
particular, is a unique and valuable service.  On the other hand, for
routine search and retrieval Entrez is less convenient than Gopher
(personal opinion, obviously).  I'm waiting for the NCBI to release usage
numbers for the Net server and for CD-ROM purchases.  A while back I
suggested that a Net version of Entrez would be preferable to the CD-ROM
version. Now that most of us have a choice, it will be interesting to see
if the prediction pans out. 

That's my two cents,

David Mathog
mathog at seqvax.bio.caltech.edu
Manager, sequence analysis facility, biology division, Caltech




More information about the Bio-soft mailing list