frodo and o on sun sparc stations and sgi indys

[Gerard Kleijwegt]

In article <1994Jan21.231508.11607 at csi.uottawa.ca>, misrael at csi.uottawa.ca (Mark Israel) writes:
|> In article <2hn1db$rpm at ursula.ee.pdx.edu>, Martin Zwick <zwick at ursula.ee.pdx.edu> writes:
|> 
|> > (1) does frodo and/or o run on sun sparc-stations?
|> > (2) (if so,) who is using them on these platforms?
|> 
|>    All the crystallographers *I* know use either SGI or Evans &
|> Sutherland hardware.  I know some NMR people who have SparcStations,
|> but even they use IRISes for interactive graphics.
|> 
|>    What with PHIGS and OpenGL, machines are becoming increasingly
|> interoperable.  You may soon be able to *run* whatever you want on
|> a Sparc (or even on an IBM PC), but whether you'll get acceptable
|> performance is another matter.

O actually runs on E&S Freedom machines now; this is a Sun workstation
with an Evans & Sutherland display.  The graphics is superb in terms
of speed.  The lines are a bit thicker than on the ESVs, but still
thinner than on SGIs.

|> 
|> > (3) what are the relative merits of sun and indys as graphics machines?
|> 
|>    Sparcs are excellent for computation, static displays, and complex
|> GUIs, but interactive graphics is not really their forte.
|> 
|>    A collaborator has tried out Frodo/TOM on an Indy, and reports that
|> performance is acceptable.  I would stay away from the 8-bit Indigo 
|> and similar machines -- Frodo/TOM runs on them, but the picture will be 
|> neither depthcued nor antialiased.
|> 
|> > (more specifically, for $10k, which would be the better machine?)
|> 
|>    Don't forget to budget for peripherals.  Are you going to buy dials
|> (highly recommended), or are you going to try to control everything 
|> with the mouse?  Are you going to buy CrystalEyes hardware stereo, or
|> will you settle for side-by-side split stereo?

The 24-bit Indy is quite okay, until you start using raster images
(sec. structure cartoons).  A Freedom would be an alternative, but
keep in mind that a lot of software won't run on it.  In other words,
if you cannot afford a graphics and a compute engine, stick to SGI.
If you can, buy an Indy or a Freedom or an HP plus, for example,
a DEC Alpha (OSF/1). (This is my advice, not an endorsement.)

|> 
|> > (4) what are advantages/disadvantages of frodo vs. o?
|> 
|> Advantages of O:
|> 
|> --  It's by far the most powerful program.  You can customize it to
|>     do almost anything, not just the things that crystallographers
|>     normally want to do.
|> 
|> --  It is (as far as I know) the only program that has Real Space
|>     Fit, for identifying bad parts of your structure.  (Real Space
|>     Fit is not to be confused with Real Space Refinement --  that
|>     we have.)
|> 
|> --  It may be "the way of the future" -- the program that's conquering
|>     the world.  I don't think the rest of us can compete with the
|>     rate of software development on O.

Allow me to add:

-- there's a lot of software that works with and around O, but not
   FRODO (e.g., OOPS [rebuilding accelerator], VOIDOO [for cavities
   and surfaces], DEJAVU [recognising protein folds], RAVE [real-
   space electron-density averaging] and OPLOT [what-you-see-is-
   what-you-get PostScript pictures of O displays])

-- it's the original; if Alwyn thinks of something new, you can
   try it in your lab a few weeks/months later and you don't have to
   wait until your generic model-building program includes the
   same functionality

-- user-support has improved a lot recently; there's an info server,
   an extensive FAQ list, a WWW-server, and an ftp-server where
   the latest versions of the software, plus any new programs, are
   made available to the user community within hours of their
   implementation in Uppsala

-- O features lots of tools which make "quality control" of
   protein structures easier (real-space fit, peptide-flip,
   rotamer side-chain); OOPS will judge your structure on a
   number of criteria and automatically generate macros which
   reduce rebuilding time by a factor of 2 to 10

-- O is more than a xtallographic model-building program; you
   can also use it for simple modelling and for creating
   publication-quality pictures (with the "Sketch" commands
   plus the semi-transparent surfaces in the SGI-version)

|> 
|> Disadvantages of O:
|> 
|> --  The learning curve is quite steep.
|> 
|> --  You may have to spend quite a bit of time writing macros to get
|>     it to do what you want.

True, but ... We've recently written a tutorial which is currently
being used by students in our Protein Engineering course.  Sometime
this or next week, I shall put it on the ftp-server.  My guess is
that a protein crystallographer with only a very basic knowledge
of Unix will be able to become a fairly accomplished O-user within
two days.  The tutorial, called "O for Morons", covers about two-
thirds of the O commands, but not the purely crystallographic ones
(since I wrote it for this Protein Engineering course).  Nevertheless,
a number of commands (pertaining to quality control, structure
analysis, generating an initial model from CA-coordinates, limited
rebuilding and mutant design) which are also used in xtallography
are covered.  Later this year, we hope to write a second tutorial
("O for Frodons" ?) which explains the model generation, building
and rebuilding stages in detail.

|> 
|> --  It's a bit of a hog.  Even if the user is doing nothing (unless
|>     he's iconized the graphics window), O will consume resources.
|>     Your background jobs won't get the CPU time they deserve.  If
|>     you move the mouse outside the O window to do something else,
|>     O will still read your mouse buttons, and unexpectedly rotate
|>     the molecule.

If this is supposed to be the major disadvantage of O, let me give
you a one-line macro to solve it:
- create a file called "pause"
- put the following line in it:
  print #Hit RETURN when you want to continue ... #
- put the pause command on your menu:
  menu @pause on
- whenever you want to take a break, hit the "@pause" item
  on your menu
- when you come back, hit the Return key

|> 
|> misrael at csi.uottawa.ca			Mark Israel

--Gerard

******************************************************************
               Gerard J. Kleywegt              ___  
  Department of Molecular Biology              | |  /\
                Biomedical Centre             /\ -- ||
            University of Uppsala             || || ||
                          Uppsala             || || ||
                           SWEDEN             || \/ --
                                              --  |__|
  E-mail: gerard at xray.bmc.uu.se
******************************************************************
           "He's probably pining for the fiords ..."
******************************************************************
  The opinions in this mail/post are fictional.  Any similarity
   to actual opinions, living or dead, is purely coincidental.
          (Nevertheless, they're mine, mine, MINE !)
******************************************************************