Re: [Chimera-users] using chimera on 3D EM map
Hi Weimin, Thanks for explaining these problems. We intend to solve all of them (though it will take some time) and perhaps some already have solutions. 1) Volume eraser ball can't be made small enough. The ball can have any floating point radius value > 0. So you should be able to make it as small as you like. I've tried and had no trouble making it tiny. A radius value of 0 is not valid and the ball radius will be left at whatever it was prior to attempting to set it to zero. 2) There is no "undo" in volume eraser. Yes. This is a serious problem we plan on fixing. 3) No way to add density (as opposed to subtracting with volume eraser). I guess you mean you would like to use a 3d ball to mark the density you wish to keep instead of the density you want to erase. It might be easy to make a tool like this where you can paint the volume using a ball (like volume eraser). You could paint using one or more colors and then separate the different colored pieces into separate maps. The underlying code to do this is already in Chimera (color zone tool, and split map by color zone). It requires a way to place a marker that matches the painting ball -- basically just a "Color" button on the volume eraser dialog. Might be able to add this soon. I recently added plane by plane loop drawing for hand tracing surfaces that can be used with the new "mask" command to segment out the enclosed region (in current daily builds). But from what you say that is less useful in your high resolution case. 4) Statistical comparison of two maps. Chimera can report the correlation coefficient between two maps and it can optimize the position of one map within another. This tools was called "Fit Map in Map" in the current Chimera production release (version 1.2470) and is called "Fit in Map" (which also handles fitting models) in current Chimera daily builds. There is also a new option in "Fit in Map" that will update the correlation coefficient in real time as you move one map within the second map. Also there is a new "Fit in Map" option that will compute a simulated map from a PDB at a given resolution (like EMAN pdb2mrc) and will fit the simulated map and report its correlation in the experimental map. Are there other measures besides correlation coefficient that you want? From your example, maybe you want rotations in degrees between two different fits? Tom Weimin Wu wrote:
Dear Prof.:
I am one postdoctoral in Prof. Wenjiang's group. We have employed Chimera in our research and really appreciate it. However when our 3D map goes beyond 5A (see Feb. 28, 2008 Nature paper) which enabled us to see the boundaries among alpha helices and beta sheets, we meet some problems in the segmentation which we do think you could help us out.
One problem is that the viewer eraser ball is too big even the radius is set to zero. It prevented us from completely clearing the neighboring signals. There is no undo option when using the eraser function. Therefore we have to keep saving the intermediate maps.
Another problem is that there is no 3D volume editor which can let us add density in 3D EM map. Sometimes this is the most important for the segmentation, because it is hard to tell which density should be included if we go slice by slice, but it is much easier to tell directly in 3D map with the stereo function.
Another problem is that there is no statistical tool for analyzing the slight difference between two 3D EM maps. Therefore it makes hard to determine the fit quality of one map into the other, for example, when fitting one subunit into the other, there is some conformation difference between them, but we need quantitative values to evaluate the difference and make a story. Something like in two PDBs, we can fit one domain and check the difference of the other domain.
If I don't explain the problems clearly, please let me know. Many thanks,
weimin wu
Hi Weimin, One more comment on the problem of making the volume eraser ball small enough. The slider that controls the ball radius in the volume eraser dialog only goes in discrete steps -- just because the smallest detectable slider motion is one screen pixel. You can resize the volume eraser window to make it wider, so the slider is longer and that will give finer slider steps. Or you can just type in whatever radius value you want and not use the slider at all. Tom
Tom Goddard wrote:
Hi Weimin,
Thanks for explaining these problems. We intend to solve all of them (though it will take some time) and perhaps some already have solutions.
1) Volume eraser ball can't be made small enough. The ball can have any floating point radius value > 0. So you should be able to make it as small as you like. I've tried and had no trouble making it tiny. A radius value of 0 is not valid and the ball radius will be left at whatever it was prior to attempting to set it to zero.
2) There is no "undo" in volume eraser. Yes. This is a serious problem we plan on fixing.
3) No way to add density (as opposed to subtracting with volume eraser). I guess you mean you would like to use a 3d ball to mark the density you wish to keep instead of the density you want to erase. It might be easy to make a tool like this where you can paint the volume using a ball (like volume eraser). You could paint using one or more colors and then separate the different colored pieces into separate maps. The underlying code to do this is already in Chimera (color zone tool, and split map by color zone). It requires a way to place a marker that matches the painting ball -- basically just a "Color" button on the volume eraser dialog. Might be able to add this soon. I recently added plane by plane loop drawing for hand tracing surfaces that can be used with the new "mask" command to segment out the enclosed region (in current daily builds). But from what you say that is less useful in your high resolution case.
4) Statistical comparison of two maps. Chimera can report the correlation coefficient between two maps and it can optimize the position of one map within another. This tools was called "Fit Map in Map" in the current Chimera production release (version 1.2470) and is called "Fit in Map" (which also handles fitting models) in current Chimera daily builds. There is also a new option in "Fit in Map" that will update the correlation coefficient in real time as you move one map within the second map. Also there is a new "Fit in Map" option that will compute a simulated map from a PDB at a given resolution (like EMAN pdb2mrc) and will fit the simulated map and report its correlation in the experimental map. Are there other measures besides correlation coefficient that you want? From your example, maybe you want rotations in degrees between two different fits?
Tom
Thanks for your reply. I will try the eraser ball again. To add the density, I really need something like 3D brush. I have tried a series of Markers and split by colors, however it is hard to choose marker position and choose color zone radius, because we need separate the signals within 4A distance, especially when two subunits cross over and one beta sheet is very close to the beta sheet on the other subunit. For statistical comparison of two maps, I need not only correlation, but also rotation degrees, translation, especially for the hinge region, something like rmsd in pdb comparison. Thanks, weimin
On Mar 6, 2008, at 11:21 AM, Weimin Wu wrote:
Thanks for your reply. I will try the eraser ball again. To add the density, I really need something like 3D brush. I have tried a series of Markers and split by colors, however it is hard to choose marker position and choose color zone radius, because we need separate the signals within 4A distance, especially when two subunits cross over and one beta sheet is very close to the beta sheet on the other subunit. For statistical comparison of two maps, I need not only correlation, but also rotation degrees, translation, especially for the hinge region, something like rmsd in pdb comparison.
Hi Weimin, Another possibility is to make a surface, and then use the surface to divide the density map. This can be done with recent versions of Chimera. I don't know if this is a good way to work with your data, but if you want to try, here are more details: (a) in Volume Viewer, show a single density plane (Features... Planes in Volume Viewer menu) (b) use Volume Tracer to draw a line in that plane, actually a string of markers (Place markers on data planes and Place markers while dragging) http://www.cgl.ucsf.edu/chimera/docs/ContributedSoftware/ volumepathtracer/volumepathtracer.html#dragplace (c) repeat (a) and (b) for a stack of planes (d) use Volume Tracer to join the stack of lines into a surface (Features... Surfaces in Volume Tracer menu) http://www.cgl.ucsf.edu/chimera/docs/ContributedSoftware/ volumepathtracer/volumepathtracer.html#surfaces (e) use the command "mask" to use the surface to divide the data http://www.cgl.ucsf.edu/chimera/docs/UsersGuide/midas/mask.html Best, Elaine ----- Elaine C. Meng, Ph.D. meng@cgl.ucsf.edu UCSF Computer Graphics Lab and Babbitt Lab Department of Pharmaceutical Chemistry University of California, San Francisco http://www.cgl.ucsf.edu/home/meng/index.html
Hi Weimin, I wonder if a 3d brush will be useful. The problem is you cannot see the depth in a 3d view of the data clearly. Amira provides a 2d brush which is much less trouble. The painting tool I suggested would have you push a button to paint within a sphere (maybe some other shape) and it would be a simple matter to allow you to drag it like a brush without the button press to test if that is useful. I'm not sure I understand the rotation and translation parameters you want. Is it that you have one map fit within another in 2 orientations and you want the rotation and translation relating those 2 orientations? Tom Weimin Wu wrote:
Thanks for your reply. I will try the eraser ball again. To add the density, I really need something like 3D brush. I have tried a series of Markers and split by colors, however it is hard to choose marker position and choose color zone radius, because we need separate the signals within 4A distance, especially when two subunits cross over and one beta sheet is very close to the beta sheet on the other subunit. For statistical comparison of two maps, I need not only correlation, but also rotation degrees, translation, especially for the hinge region, something like rmsd in pdb comparison.
Thanks, weimin
My problem is that I have few subunits in the asymmetry unit. I fit one into the other subunit, and I try to figure out how good is the fit. Visually I can see they fit well, but there is some difference which I think it is due to conformation difference, however I need quantitative data to report the difference, or a tool as simple as rmsd in the PDB comparison. I once tried to convert the density map to pseudo PDB (vol2pdb in situs), but it is hard to open that PDB in chimera. weimin Thomas Goddard wrote:
Hi Weimin,
I wonder if a 3d brush will be useful. The problem is you cannot see the depth in a 3d view of the data clearly. Amira provides a 2d brush which is much less trouble. The painting tool I suggested would have you push a button to paint within a sphere (maybe some other shape) and it would be a simple matter to allow you to drag it like a brush without the button press to test if that is useful.
I'm not sure I understand the rotation and translation parameters you want. Is it that you have one map fit within another in 2 orientations and you want the rotation and translation relating those 2 orientations?
Tom
Tom Goddard wrote:
Hi Weimin,
One more comment on the problem of making the volume eraser ball small enough. The slider that controls the ball radius in the volume eraser dialog only goes in discrete steps -- just because the smallest detectable slider motion is one screen pixel. You can resize the volume eraser window to make it wider, so the slider is longer and that will give finer slider steps. Or you can just type in whatever radius value you want and not use the slider at all.
Tom
Another problem when using volume eraser is that the keyboard accelerator "es" doesn't work. My system is Linux x86_64. weimin
Hi Weimin, To use the "es" keyboard shortcut with volume eraser you have to turn on keyboard shortcuts using menu entry Tools / General Controls / Accelerators On. They are not enabled by default. I should add an "enable shortcuts" button to the volume eraser dialog or automatically enable them when you show that dialog. One reason for not enabling shortcuts by default is that accidentally typing in the Chimera window can do many unexpected things as there are more than a hundred shortcuts. http://www.cgl.ucsf.edu/chimera/docs/ContributedSoftware/accelerators/alist.... Tom
participants (4)
-
Elaine Meng -
Thomas Goddard -
Tom Goddard
-
Weimin Wu