[Jmol-developers] Java-less Jmol == "JSmol"

[Robert H. <ha...@st...>]

Jmol Developers --

Some of you may know that since the end of August, after coming back from
the ACS meeting in Philadelphia, I became convinced that we needed to
explore ways of taking Jmol to the next level -- the JavaScript/HTML5/WebGL
level. I began experimenting with various ways of doing this -- first with
Google's web toolkit (a disaster), and then  with compiling Jmol into
JavaScript using the amazing package Java2Script, created by Zhou Renjian.
If you have followed my code check-ins at all, you know that this has
involved a substantial amount of refactoring of the Java code, and that was
a challenge, but it was worth it.

Today I can point you to a not-exactly-fully-functional version of Jmol (we
are calling it JSmol or maybe J2Smol for now) that uses WebGL for its
rendering (based on GLmol, by Takanori Nakane). With the help of a St. Olaf
student, Duncan Blue, we are busy testing this and, of course, finding
problems daily. But the enormous progress just these last few days prompts
me to let the cat out of the bag.

http://chemapps.stolaf.edu/jmol/jsmol/test2.htm

Consider it alpha. I do believe we have solved some of the LONG standing
problems with JavaScript-only molecular visualization. For example:

- All ASCII file readers should work exactly as in Jmol itself (tested so
far: XYZ, MOL, PDB, CIF)
- VDW surface and solvent-accessible surface creation is no problem
- partial charge calculation using MMFF94 is implemented
- MEP mapping of isosurfaces works beautifully
- JVXL surface files are read just fine
- SMILES/SMARTS matching works
- all database access (RCSB, PubChem, NCI-Cactus) is exactly as in Jmol/Java
- protein and nucleic acid cartoons work just fine
- LCAOCartoons work just fine
- "load append" and multiple file loading/frames are no problem
- Jmol scripting is fully implemented
- Jmol math is fully implemented

The list goes on and on. (Though obviously there are bound to be some
limitations.)

Some of you will recognize that many of these capabilities (all of these
capabilities?) are completely unheard-of in JavaScript. This goes way, way
beyond what any currently available JavaScript-only option provides. There
are no server-side tricks being played here. It is just exactly as
client-side as the Jmol Java applet itself. Surfaces are being calculated
and mapped on the fly, and they look and handle almost as well as in Jmol
itself. I know others have struggled for a long time to do this, and all I
can say is that Java2Script is amazing. It took a bit of optimizing and
adapting, some careful coding, but in the end that library just flies. We
are fully autobonding a protein in 125 ms, creating VDW surfaces for small
molecules and even proteins in about 1-3 seconds. It's really something to
behold.

What's absolutely amazing about this is that I have not created a new
program at all. "All" I've done is used Zhou's incredible
Java-to-JavaScript compiler to allow the Jmol trunk Java code to be turned
into a JavaScript alternative at the click of a button within Eclipse. The
Java classes and Jar files are all still there. It's one development
stream. It's really very bizarre. But, wow, does it work. And the code
management is a snap -- There are not two versions of Jmol here -- there is
just the trunk Java version code that can be compiled either into class
files or their JavaScript alternatives. The debugging it totally bizarre --
stepping through the compiled Java/JavaScript code using Firebug. Who'd
have thought it possible??

What we have is a perfect combinations of efforts -- Zhou's compiler,
Takanori's GLmol interface, and Jmol. I cannot overemphasize the importance
of the contributions of these two guys. Perhaps some day we can meet, but
already I can tell you that none of this would be possible without their
obviously huge efforts. I have to tell you that this has been perhaps the
most exciting, rewarding, and at the same time totally weird two months of
my life. Seriously, two months ago I thought there was absolutely no chance
we were ever going to see Jmol transcend Java, and, in fact, I argued that
it couldn't be done, but here we are - surfaces and all! I'm still catching
my breath.

Go ahead and exercise it some if you want on that test page. Explore the
code. Don't expect perfection -- this is definitely a work in progress.

I'm off to Washington, D.C., to spread the word...


Bob

ps: TODO includes draw, labels, echos, animation, spinning, minimization,
MOs, binary file loading, and a number of other things.


-- 
Robert M. Hanson
Larson-Anderson Professor of Chemistry
Chair, Chemistry Department
St. Olaf College
Northfield, MN
http://www.stolaf.edu/people/hansonr


If nature does not answer first what we want,
it is better to take what answer we get.

-- Josiah Willard Gibbs, Lecture XXX, Monday, February 5, 1900