Visualization of X-ray Tomograms using Chimera

Tom Goddard
February 6, 2008

Topics

What does Chimera do?

Interactive visualization and simple analysis of 3-dimensional data.




Atomic models and sequences

Morphing between conformations.

Molecular assemblies

Crystallography maps

Single particle reconstructions

EM tomography

Demonstration

X-ray tomogram of dividing S. pombe cell provided by Dula Parkinson.

Dividing S. pombe (yeast), tilt angle 0. Segmentation. Vesicles colored by density: high to low red, yellow, green, blue, black. Nucleus is orange.

Which Software to Use?

ChimeraAmiraIMOD
Good

Easy to use. Free. Extendable.

Developers nearby. Collaboration.

Focused on map analysis.

Flow-graph programming is flexible.

Good 2D plane by plane data analysis and segmentation.

Free.
Bad

Focused on molecular scale.

Map capabilities developed for electron microscopy at molecular resolution.

High cost per machine license.

Commercial vendor may focus on large customers' needs (medical imaging).

Limited extendability (C++).

Limited 3D capabilities?

I have not used the program enough to evaluate it.

Chimera Setup and Support

Demonstration Details

Show original tilt series data, pombe21.tif, 90 image multipage tiff file.
Enable plane display.
Put one threshold at left edge of histogram at middle height, one in middle
  of histogram at top, rightmost one not quite at right edge to eliminate border.
Chimera does not calculate tomograms from tilt series.
Can make movie from tilt series.
Note the bright illumination in the upper right, and overlap of another
  cell at bottom for many view angles.
Close tilt series.
Open tomogram pombe21.mrc, 4103.
Adjust surface threshold lower (~ 0.005).
Adjust step from 8 to 2.
Surfaces common for EM single particle maps. Volumetric style more used
  for tomograms which are noisier and have less even brightness.
Switch to solid mode, step 4, adjust middle threshold ~.007, low to 0.003.
Rotate. Switch to step 2 for nicer detail.
Switch to plane display.  One plane.  Note step is now 1.  Drag slider
 through range of planes (much faster than Play since planes are skipped).
Move lower threshold to zero and height 1/4 to see low signal features.
Switch to y axis plane and drag slider.
Plane by plane display is standard to see maximum detail and low signal detail
  in EM tomography.
Remove pipette by masking out (setting to zero) volume outside cylinder.
Switch to surface, step 4.
Create cylinder with command "shape cyl" using one command shortcut oc.
Change color of map (light yellow) to distinguish cylinder.
Resize cylinder by selecting and using sz shortcut and button 3 drag.
Move cylinder to be coaxial with map, shortcut ao activate only selected
  and aa activate all, or at active toggle, shortcut or orthographic projection.
Mask to cylinder, one command oc, "mask #0 sel", hide surface hs.
Made map smaller.  Show outline box, original map outline box, then unshow.
Smaller can allow volumetric rendering at full resolution.  Size is still not
  below 256, so does not help in this case.
Align volume grid with axis of two cells.  Helps with plane by plane display.
Use subregion selection panel, drag box, toggle on/off rotate box and align
  with higher density cell including membrane separating cells.
Drag box sides in close to cell membrane.
Press Resample button, hide original map, click of subregion selection mode,
  hide subregion selection panel, raise threshold to see inside, step 2, step 1.
To make vesicle boundaries clearer, white background with silhoutte edges,
  width 2.  Back to black, no edges.
Show plane display along z, now aligned with long cell axis.
Show all planes, adjust brightness, use step 1, map size now below 256,
  rock by hand, hide planes panel.
Remove cell membrane to get a clear view of vesicles at low threshold.
Use same surface masking done for pipette removal, only shape is harder.
Use Gaussian smoothed data to get the cell membrane shape.
Use Gaussian width 20 voxels. Low threshold 0.004, blue mesh, step 2.
View end on, clip half way through using side view, white background
Toggle selectability (ts), select blue, scale with mouse (sz) to fit inside.
Mask to remove cell membrane "mask #4 sel", hide blue mesh.
Enable silhouette edges, width 2.
Can now use low surface threshold and see hollow vesicles without blocke view.
Use step 2, point out a hollow large vesicle near center.
Next want to mark individual vesicles.
Can color and measure volume of connected blobs.
Color several different colors.  Choose disconnected ones.  Threshold 0.00645.
Note that volume grid spacing is not set so volume is in voxels.  Setting
  spacing to 20 nm, then volume would be reported in nm^3.
Connectivity only sufficient for coloring well isolated blobs.
Color some connected blobs.
Another serious problem is that Chimera does not have a way to remember this
  coloring.  Change threshold and show color vanishes.
Use a more flexible approach of placing spherical markers on vesicles.
Use volume tracer, drop marker on spot, adjust map transparency 0.7, marker
  color dark blue for contrast.
Adjust mouse mode to move markers, turn off linking.
Move marker to center of vesicle and resize (alt + mouse drag).
Raise threshold, quickly drop 20 markers without resize or move, lower
  threshold place a few more.
Explain that move and resize would be next phase.  Show moving some.  Save
  much time by moving many in one view.
Open a marker file with centered and sized markers on all vesicles and nucleus.
Coloring in movie was according to average density within vesicles.
Chimera doesn't have a builtin tool for measuring average density in spheres.
  But a Python a one page script can do it.  Might show it, explain blocks.
Select markers and open script avedensity.py, explain render-by-attribute
  histogram and color markers by average density after unselecting nucleus.
Show mesh to increase transparency of map.
Might note alarming fact that hand is flipped from movie, by observing 4
  densest red markers and nucleus.  Probably a map handedness problem.
Modified script hollowness.py computes ratio of ave density in shell +/- 20%
  of marker radius over ave density inside 80%.  Indicates hollowness.
Open script, color with render-by-attribute.
Show z plane with 2 red markers, select markers in plane, invert selection,
  hide others, invert selection, hide and show red and blue balls to show
  that hollowness is indicated.

Average Density in Sphere Python Script