Classifying particles into two groups based on occupancy

Particles should be aligned again (see b12.ali) against the refined volume, as opposed to the reference volume.

All the files below are contained in occupancy.tar.gz


b01.dcf: B01.dcf uses the select files generated by b16.cla. It creates view files for each of the 83 projections.

b02.cpy: copies them into different directories.

b03.fit: (optional) shows sensitivity of the procedure. Particles must all have the same noise file.


Use SPIDER to create a spherical mask such that the desired ligand is contained within the sphere. Verify in explorer.


b05.pmk: projects the mask in the 83 directions.

b06.mth: Thresholds the 2D masks so they are binary again (they are not binary after being projected).

b09.clk: The main classifier batch file.


CA SI creates SEQ files

CA S (optional, if you would rather to classify on the pixel level) N = 5 factors (PIX#, IMC, EIG)

CL KM works directly on images. It needs the number of pixels under the mask, applies factor to every pixel.
2 specifies IMC; 1 to use SEQ made by CA SI
x40 = 2 = no. of classes
Calculate 2 averages according to class, the calculate resolution between averages.

AS DC creates avg*** and var*** files for both classes. You now have 83 x 2 averages, but which one has the ligand?


b10.sub: difference between averages & significance map

b11.fsm: always use the mask picstat002.ext

b11.stc: ?????? (unimportant)

b12.dcf: Finds which group (1 or 2) has the higher average. Then creates seltotal_h and seltotal_l, which contain the numbers of the aligned particles (in ali/sar*****), with high or low occupancy, respectively.

b23.bph: make the volume

The above routines do not use defocus. You can try to sort back into defocus groups.


Created: 9/14/01    Bill Baxter