Dave wrote: > Thought you might be interested in some feedback I was giving the Jmol > Developers in regards to my efforts to find OR develop a replacement for > Rasmol/chime to be used to teach protein structure and functional > chemistry in our medical science / biotechnology degrees at our > university. > > I built a vis tool using java 3D extension but the performance (frame > rate) above 4000 coordinates was very poor... even after removing > antialiasing and other taxing processes. We also looked at pymol, > molView, FPV, cn3d (C++), mage/kinemage (Java - from Richardson lab at > Duke Uni.) and of course rasmol and chime.
I have never used Java3D, but I am surprised/disappointed that the performance dropped off so quickly. > The major deficiencies of all of these included: > > 1) excellent quality images in 3D space with really nice rendering, but > when it came to interacting with the molecule, performance (frame rate) > was very poor for large molecules (such as over 50,000 coordinates, I > tested with a 59,000 coordinate structure file using spacefill/vdw > representation, apparently there are some files already that have > greater than 99,999 coordinates, so the need to handle large files was > important) OR when you were at maximum zoom and rotating the molecule > with a mouse. So rotation/translation was very clunky on those apps > with nice graphics. Maximum zoom is also a performance problem in Jmol. Once the size of the individual spheres gets above 100 pixels (?) then the performance drops dramatically. If someone has a real need to fix this then we can talk about it. > 2) Those programs that could handle smooth rotation/translation of the > very large structures had very, very poor quality graphics. So if you > used one to visualize an interaction/characteristic of a molecule and > wanted to use it in a presentation - it would look very second rate... > and the aim of the software is visualization... so poor quality > rendering/buffering of the image was also a key requirement ! OK > 3) Functionality and the ability to take command line inputs was > essential - especially for teaching. It just provides versatility in > working with molecules and allows the running of scripts. OK > 4) portability because we have people with Mac's and PC's and Linux > machines ! Yes, I think that many people in academic environments have this issue. > SwissView is also a program we use and considered, but it requires a lot > more time to teach with because of the large and advanced function > range it possesses, so is more suitable for teaching people already > skilled with rasmol/chime/jmol. And it�s a predominantly point � click > format also. Actually, I've never even seen SwissView. But I have heard that it has very good functionality. I need to download it some day. > I actually tested the old V8 of Jmol and was a bit disappointed by it's > performance in No. 1 above, I spent a *huge* amount of time trying to make Jmol v8 run better. But the problem is the Java graphics routines and the overhead of making all the graphics calls. Many people think that the Jmol v8 graphics routines look better (because of antialiasing) but there is a significant price for it. > But when i checked out the V10 beta code and > compiled it, blitzed No. 1 requirement above !!! Well, that is good to hear :-) > Jmol is a long way > ahead of any other open-source vis application I found, in all 4 area's > above ! The new graphics engine is excellent. And the specular and > perspective used for visualizing the structures contributes alot to the > realism of the images. I am glad that you mentioned the perspective view. Not many people mention it, but I think it is a nice touch. > There are other features that would be nice, > but I�m sure their on there way (solid ribbons, solid ribbon cartoons, > iso-surfaces etc). We still have a lot of work to do, but we are making good progress. > So I was really excited by V10 Jmol !.. We also have SGI machines in the > school we use for rational drug design but these aren't any good for > teaching biochemistry secondary/tertiary structural protein chemistry > as Rasmol/chime were. Have you tried Jmol on your SGI machines? I had access to an SGI Octane a while ago and the Java implementation was terrible. I could never get Jmol v8 to run on it. > I�ve modified Jmol application V10 code to include a summary of the PDB > file, which collates all the information for each molecule represented > (expression system, chain identifiers etc) which can be saved as an > .rtf; a window for viewing the entire raw PDB file (without having to > open it in third party app), and a window for reading collated PDB file > remarks. > > I docked the script window with the graphics panel and put the > measurements panel behind it on a tabbed panel. I also added a browser > pane to it where we will have tutorials. I'm just updating the browser > window so it's http 1.1 and javascript compatible and can authenticate, > and we�re still to write the actual tutorials for the browser window. > So it�s heavily geared for educational purposes and providing as much > information to the user about the structure as possible, as well as > easy access to essential functions � measurements and command line. > > There are some screen shots of what we�ve come up with so far located > here > > http://jmol.sourceforge.net/dchandler/screen_1.jpg > http://jmol.sourceforge.net/dchandler/screen_2.jpg I think that these tabbed windows are excellent! The GUI of the Jmol application has not received very much attention because of the work on the core 'viewer' (which is shared by Jmol and the JmolApplet). The idea of incorporating a browser window into the application is an interesting way to tackle the problem of presenting instructional material while still offering the benefits of a dedicated application. I am *very* glad that you shared these images with us. > ... still 2-3 months away from finishing it up (as I�m doing a PhD in > Molecular Biology � crystallography studies of a protein, and doing the > programming just on the side in my spare time... PhD + sparetime = not > enough hours in the day !). > > Hope this gives you some ideas and info on where we think Jmol stands as > a tool for protein / molecular visualization, and it�s excellent > applications as a tool to replace rasmol/chime ! Thanks very much for sharing this with us. Keep us informed as you make progress! Miguel ------------------------------------------------------- SF.Net is sponsored by: Speed Start Your Linux Apps Now. Build and deploy apps & Web services for Linux with a free DVD software kit from IBM. Click Now! http://ads.osdn.com/?ad_id56&alloc_id438&op=click _______________________________________________ Jmol-users mailing list [EMAIL PROTECTED] https://lists.sourceforge.net/lists/listinfo/jmol-users
