Dear Chimera Team: I think your work on Chimera has been of very high quality. In fact, I've used it to produce print quality images for an upcoming book of mine. However, I've had some problems with displaying some large molecules, for example, 1AYM and 1P5Y. I use the PDB file for the molecule, but specifically the assumed biological model file, not the the asymmetric unit file. I've been able to open the file in programs like Deepview and Raswin, but I can't open the file in Chimera. I'm running Windows 98 with about 256MB of RAM, and what happens when I attempt to open the assumed biological model file Chimera runs for over an hour trying to open the file and then crashes with a page fault message. I certainly would like to use the image capabilities provided by Chimera, so can you help me with this? Or is this simply a limitation of Chimera?
Hi, The problem opening the biological unit PDB file for virus capsids 1aym and 1p5y is that you are running out of memory. The asymmetric unit is small 1aym (human rhinovirus 16) 7000 atoms 1p5y (canine parvovirus) 4400 atoms but with the 60-fold icosahedral symmetry, these become 420,000 atoms and 264,000 atoms. Chimera requires about 2-3 kbytes per atom, so you will need a computer with at least 1 Gb and preferably 2 Gb of physical memory to open these. With less memory the machine will become extremely slow as it uses your hard-disk to cache data that won't fit in the physical memory. I opened the 1aym biological unit on a Mac desktop machine which has 1.5 Gbytes of physical memory in a few minutes. Chimera was using about 1 Gbyte after it opened. To open the model with less memory you could make a new PDB file containing only the CA backbone atoms, or just CA, C, N, and O backbone atoms. That would require writing a script to filter out the other lines of the file. We could give you the backbone only files if you want. Another approach is to just open the asymmetric unit and use the Chimera multiscale tool (Tools / Multiscale / Multiscale Models) to make the 60 copies. This does not copy all the atoms so it does not need alot of memory. You open the asymmetric unit and press the "Make models" button at the bottom of the Multiscale dialog to display the full capsid. Here are images of 1aym and 1p5y I just made this way: http://www.cgl.ucsf.edu/home/goddard/temp/1aym.png http://www.cgl.ucsf.edu/home/goddard/temp/1p5y.png You can also show some of the capsomeres as ribbons or all atom displays. But if you use Multiscale and try to show the full capsid at atomic or ribbon level, then it will try to copy the asymmetric unit 60 times and you will again need alot of memory. More images of virus capsids using the Multiscale tool are on our web site http://www.cgl.ucsf.edu/Research/virus/ and on the VIPER web site http://viperdb.scripps.edu/ Tom
Thomas: Thanks for your very helpful information. It provided enough detail for me to make some important desisions about how I'll produce images for my book. You also confirmed my assumption that I was running out of RAM. I decided to use the Multiscale Utility for create the image, and it works fine for my needs. However, it seems that I have no control over which colors are used. Is that true?
Hi, You can change colors for the virus capsid models made with the Multiscale tool. Select one protein surface with the mouse using ctrl-left mouse click on it. Then press the "Copies" button near the top of the Multiscale dialog to extend the selection to all copies of that protein. Then press the color button (the square with the raised ridge) near the middle of the Multiscale dialog and adjust the color of the selected chains. The Actions / Color menu entries do not work for changing the Multiscale surface colors. Unfortunately, none of the Actions menu entries act on surfaces created with the Multiscale tool. All the controls for manipulating multiscale models are in that one Multiscale dialog. It's a confusing user interface resulting from the fact that the Multiscale tool is built on top of the core Chimera capabilities and has no way of changing the behaviour of core user interfaces like the Actions menu. Tom
participants (2)
-
Private -
Thomas Goddard