PFT2 & EM3DR2 Tutorial

Finding Origins and Orientations and Making Reconstructions with PFT2 and EM3DR2

1. The process of determining origins and orientations is iterative. For each iteration the following pft2 parameters are defined (enter "pft2" to see more options):

Parameter	Definition	Input Value
-anglestep	The angle step size in degrees	e.g. 1.0 to 0.1
-bin	Image compression factor (optional)	integer (1, 2, etc.)
-mode	Different algorithms of pft2 used to determine origins and orientation	sphere_search global refine, etc.
input.star	Input micrograph and particle parameter files are given after all the program parameters and options	e.g. polyoma.star
-map	Starting map	image file
-output	This file will store the orientation and origin information	*.star
-phaseampl	For modes sphere_search, global, refine, and symmetry, the initial orientation is determined by Fourier amplitudes only. You may select the top n orientations via this initial (amplitude-only) method to be tested by Fourier amplitudes and phases (a more rigorous method) if you enter a value of n >1.	# of top orientations
-radii	Inner and outer radial limits pft2 will use when determining orientations and origins	use radii from step 8 (on preparation page), given in pixels
-resolution	High and low resolution limits in angstroms (in Fourier space)	usually the minimum resolution is 1/3-1/4 of the particle diameter, e.g. 90-25 angs.

2. Iteration #1: Find initial origins and Euler angles using a mode in pft2 called sphere search. Enter the command

pft2 -mode sphere_search -phaseampl 3 -output polyoma_search.star -anglestep 1.0 -radii 20,34,1 -resolution 120,25 -map polyoma_3d.pif polyoma.star

Note: add "> pft2_iter01.log" to the end of the command to put all screen output into a log file. Use "-bin 2" to speed up this and subsequent pft2 runs.

3. Select particles with acceptable correlation coefficients using the Bsoft program bpartsel. Enter the command

bpartsel -verbose 7 -output polyoma_01_sel.star -deviation -1 polyoma_search.star

The option "-deviation -1" selects particles that have correlation coefficients (a figure-of-merit) which are greater than or equal to one standard deviation below the mean. Enter "bpartsel" to find other selection criteria.

4. Make first reconstruction by running em3dr2

em3dr2 uses the information in a star file to generate a reconstruction. You must specify the input star file and the output file name of the map it will generate. The map generated in this step will become the starting map used in the next iteration. Enter "em3dr2" for a complete list of options. Enter the command

em3dr2 -output polyoma_map2.pif -symmetry I90 -resmax 24 polyoma_02_sel.star

In general to avoid Fourier truncation artifacts, you should increase the resolution by at least 0.5 to 2 angs. over that used in pft2. For example, if you use a high resolution of 20 in pft2, your resolution maximum should be 18 or better in em3dr2.

5. Iteration #2

This iteration refines the origins and Euler angles found in iteration #1 by running pft2 in global mode. The command to run iteration #2 is fairly identical to that of iteration #1 except for the following differences:

i. global mode (instead of sphere_search)

ii. polyoma_map1.pif (note that the starting map for each successive iteration will be the map generated by the previous iteration)

Run the command, pft2 -mode global -phaseampl 5 -output polyoma_02.star -anglestep 1.0 -radii 20,34,1 -resolution 120,25 -map polyoma_map1.pif polyoma_01_sel.star

After the pft2 run finishes, run bpartsel and em3dr2 for iteration #2, include the iteration #2 output. Enter the commands

bpartsel -verbose 7 -output polyoma_02_sel.star -deviation -1 polyoma_02.star
em3dr2 -output polyoma_map2.pif -symmetry I90 -resmax 24 polyoma_02_sel.star

6. Iterations 3 to n

Subsequent iterations will also require the correct starting map (the map generated by the previous iteration). You may also want to increase the resolution, decrease the angle step-size, and run a few iterations in refine mode in order to obtain accurate orientations and origins at relatively high resolution. Here are a few suggestions for iteration 03-09 settings:

Iteration	03	04	05	06	07	08	09
Angle step	1.0	1.0	1.0	1.0	0.5	0.25	0.1
Mode	global	global	global	global	refine	refine	refine
Max pft2 resolution	25	20	20	17	17	17	17
Max em3dr2 resolution	20	16	15	15	15	15	15
N for "phaseampl" option	10	25	50	75	49	49	49

Watch the resolution via bresolve (step 7). If resolution does not improve, then your analysis 1) is complete or 2) you should increase the high-resolution limit or other parameters and continue until no further improvement in resolution is observed. You may also want to observe the average figure-of-merit which is computed at the end of each pft2 run:

                     Average Std. Dev.
---------------------------------------
MAG factor           1.000     0.000
Figure-of-Merit      0.475     0.049

In general, the reconstruction quality will improve as the figure-of-merit improves. However, the figure-of-merit is sensitive to changes in radii and resolution, so comparisons are usually only valid when those parameters are identical.

7. Resolution assessment

bresolve is a Bsoft program that determines the resolution of your reconstructions (maps). Run em3dr2 to generate two maps that will be used by bresolve to determine resolution. Example for iteration 3: em3dr2 -verbose 0 -number 2 -output 03.pif -resmax 20 -symmetry I90 polyoma_03_sel.star

Then run bresolve. Example for maps from iteration 03: bresolve -verbose 7 -map 03_02.pif 03_01.pif > bresolve_03.txt

You may delete these temporary maps using the following command: rm 03_01.pif 03_02.pif

8. Script: iterations_01-09.tcsh

This script allows you to run pft2, bpartsel, em3dr2, and bresolve for nine iterations using a single command. Before running iterations_01-09.tcsh, be sure to enter all the appropriate parameters for your data set called for in the script. This script is based on the suggestions in step 6. These parameters have worked well for several icosahedral reconstructions. Note that focal pair images were used and combined, but the script could be adapted for other cases.

9. Viewing reconstructions.

The maps generated from the iterations can be viewed, for example, with the program DINO or Chimera (to view the overall three-dimensional structure) or with the Bsoft program Bshow (to view cross sections). If you generate a polyomavirus reconstruction using the data provided, you should end up with a map that looks like the following:

Surface rendering of polyomavirus reconstruction

Volume rendering of polyomavirus reconstruction, central section

Polyomavirus reconstruction. Surface rendering (left). Volume rendering of central section (right).

Return to main tutorial page

Return to main PFT3DR page

24 Mar 2008
Questions or problems? Contact David_Belnap@byu.edu