VMD-L Mailing List

From: Andrew Dalke (dalke_at_ks.uiuc.edu)
Date: Tue Mar 25 1997 - 02:09:57 CST

Hello,

  This issue of VMDTech is several days late because we have been
working on the final details of the release of VMD version 1.2b1. It
is now available from ftp.ks.uiuc.edu in /pub/mdscope/vmd/ . We will
now start updating the VMD web pages to better reflect information
about this new version. If you have an older version of VMD we
strongly recommend you update it.

  If you download the code during the next week, please tell us if
there were any problems. While we tested the installation and
compilation several times on our systems, it is hard to ensure that
things will work as well elsewhere.

                  What's new with this version?

  The biggest change is that VMD now runs on machines other than SGIs.
In addition to the normal full source and SGI binary release, we
distribute precompiled binaries for HPs and Linux. The version names
and tested platforms are:

        IRIX5 -- works on IRIX 5.x and 6.x
        LINUX -- tested on RedHat with 1.2.13 and 2.0.27 kernels
        HPUX9 -- works on HP-UX 9 and HP-UX 10

The non-SGI versions use OpenGL, which is emulated using a graphics
package called Mesa that runs under X-windows. The SGI version still
uses GL, though we will soon have an IRIX6 version compiled for
OpenGL. (Some SGI hardware is GL based and the OpenGL version does
not as run well on these; similarly, some SGI hardware is OpenGL based
and the GL version doesn't run as well under these.)

  We've expanded many of the features already present in VMD. There
are a couple of new rendering types, like a fast solvent accessible
surface calculation and a C-alpha (and P) trace. The Tube method now
follows C-alphas (and P's) if no proteins (or nucleic) acids are
present. There are new rendering types, such as the oft requested
VRML and Postscript outputs. VMD now reads Amber files and reads and
writes Amber trajectories, so you can now convert between the two.

  Many of the changes are not visible in the GUI. We fixed a lot of
bugs in VMD, at present we know of no way to make it dump core.
Almost all of the commands will now print a short usage statement if
there was an error or if you enter the command by itself. For
instance, to see all the options on "color" just enter "color" . The
commands also have parameters for getting more information about each
options, so you can query about the current projection matrix or get a
list of available selection keywords. We also added more "traceable"
Tcl variables for performing actions based on different events. There
will be a VMDTech article on this in a few weeks. We improved the way
VMD interacts with a web browser and replaced how it does help. In
two weeks we'll have an interesting article by Jeff Ulrich on how to
use VMD to control a web browser by picking on different atoms.

                What's wrong with this version? or Why is it "b1"?

  As the title says, this is version "1.2b1", meaning it is the first
"beta" version. There are a few problems with this version and I am
hesitant to give it a full update. These problems are ONLY in the
OpenGL version of VMD and not in the GL version (IRIX5). They are due
mostly to our not quite understanding how the OpenGL library
interfaces with the XForms library upon which the GUI is based. They
are:

  o OpenGL and the XForms menus sometimes conflict
        - A form must haven been open at least once before using
          the pop-up menu from the graphics screen. Otherwise
          pressing the right mouse button in the graphics form
          WILL CAUSE VMD TO EXIT and report an X problem.
          The easy solution is to always open a menu, even if
          just to close it.
        - The popup is in a different colormap than the OpenGL
          display, but the colormap doesn't always switch when in use
  o Colormap flashing
  o The default window size information doesn't always work
  o Under Linux the standard error(3) causes a "Bus Error", as
        a temporary measure, we call _exit(2). This usually indicates
        a memory problem, but Purify (on the HPs) suggests nothing.

If there is anyone here who wants to help us fix these problems, we
will be quite grateful. While we're at it, we want to thank Steve
Searle for his help in finishing off the OpenGL port for this release.
Thanks Steve!

Also, while it isn't quite a bug, the DCD binary format is platform
specific, so VMD on an Intel box doesn't read a trajectories made on a
MIPS or PA-RISC machine.

                What will be in the final 1.2 release?

There's a lot of things we would like in the final release. In
addition to the problems mentioned above, which we are working on,
the three biggest features we will add are:

  o updated documentation -- what we've all been waiting for.

  o MSMS -- This is a fast molecular surface library from Michel
Sanner at Scripps,

  o more analysis commands -- We would like to be able to calculate
the surface area and volume, as well as get information about hydrogen
bonding (perhaps via HBPlus?) and nucleic acid base pairing. We also
want commands for computing standard structure values, like psi/ phi/
chi(n) /etc. for proteins and shear/ slide/ roll/ etc. for nucleic
acids.

  We are planning to have a Tk version of VMD instead of using XForms.
The 1.2b1 source distribution can be compiled for Tk and it works, but
all it does is open a window; there is no GUI for it yet. The three
advantages to Tk are that 1) it is programmable so you can add new
forms or modify ones that already exist, 2) it is available in source
code and 3) it works on non-X platforms and we are seriously
considering a MS Windows NT version.
  Tk will be added only if there is enough time.

                What is the time scale for future versions?

  This is somewhat hard to say as it depends on several factors.
Documentation and rewriting the GUI will take a lot of work (we'll
have to learn how to do Tk). We want to fix the problems with
OpenGL/XForms. And of course we are working on other projects as well
which limits development time.

  The biggest constraint is we would like to compile the final version
against Tcl 8. This is a byte-compiled version which offers a
several-fold speedup compared to the current version. Along with this
are updates to the associated Tk graphics packages, which we would
have to figure out. The posted schedule for the Tcl release is in
June, so don't expect the final version of VMD 1.2 out before the end
of summer.

  We plan to finish the documentation before then. When it is ready,
we'll notify subscribers to this mailing list.

                        What are some cool things?

  Hmm... Well, we redid the help interface so it calls a web browser.
Try "Help" from the popup menu. This calls the command "help
<keyword>", which matches the keyword to a url and passes to
vmd_open_url. That function uses the environment variable
VMDHTMLVIEWER to call a web browser. The default option, compiled
into VMD, is:

        netscape -remote openURL(%s,new-window)

which checks to see if a version of netscape is already running and,
if so, tells it to open the URL in a new window. I personally don't
like this option, so I set VMDHTMLVIEWER to

        netscape -remote openURL(%s)

which opens the URL in the currently active window. Actually, a
problem with this command is if a browser is not already open, nothing
happens, no window opens. Thus, if first try fails and the first 7
characters of the variable are "netscape" then VMD assumes the first
word (in this case "netscape" as well) is the browser name and calls
it directly.

  Or, you might try something like:

        xterm -e lynx %s

  Now, the "help" command can be used to call many things. Currently
you can do things like "help raster3d" to open up to the Raster3d home
page, or "help tcl" for the Tcl page. Since the look up is all done
in Tcl (see $VMDDIR/scripts/vmd/openURL.tcl) you can add more items,
if you wish.

  Speaking of cool things, several researchers here put together some
movies of their work based on the previous VMDTech article on making
movies. If you are interested in seeing them, take a look at
http://www.ks.uiuc.edu/Overview/movie_gallery/ . The biggest problem
with the movies is they are between 200 kB and 1.2 MB in size, which
does take a while to download even locally. This Friday's VMDTech
will explain how to make an animation that morphs between two
structure just like the "Kinesin" movie.

                                                Andrew Dalke
                                                vmd_at_ks.uiuc.edu