Comparison of particle picking methods using the PCOMP batch file

The batch files pcomp.bat and pcomp_p.bat produce lists of particles picked in different ways, hereafter called "method A" and "method B".

An example

  • The input micrograph was a 2000 x 2000 window from a 2.92A resolution micrograph.
  • Method A: ran particle-picking (pick.bat) on the raw micrograph.
  • Method B: ran particle-picking (pick.bat) on the CTF-corrected micrograph (defocus = 32800).
  • Inputs to pcomp.bat are the Particles/coords/sndc00* files produced by pick.bat. Also, a threshold is needed because particles picked by both often don't have the exact same coordinates. Threshold = allowed pixel difference.

    Outputs:

  • doc file BOTH.DAT lists particles picked by both A and B. 
                A           B        A xcoord   A ycoord    B xcoord    B ycoord

    1 6  1.0000      3.0000      843.00      43.000      843.00      43.000    
    2 6  2.0000      4.0000      170.00      81.000      170.00      81.000    
    3 6  3.0000      5.0000      47.000      91.000      47.000      91.000    
    4 6  4.0000      6.0000      523.00      103.00      523.00      103.00    
    :     :           :            :           :           :           :   
  100 6  105.00      119.00      1591.0      1803.0      1595.0      1803.0    
  101 6  106.00      120.00      935.00      1815.0      935.00      1819.0    
  102 6  107.00      121.00      123.00      1835.0      123.00      1835.0    
  103 6  108.00      122.00      39.000      1843.0      39.000      1847.0
  • A_not_B: doc file listing particles picked by A but missed by B. 
    ;bat/dat   05-NOV-2001 AT 13:26:15   a_not_b.dat
    1 1  12.000    
    2 1  20.000    
    3 1  67.000    
    4 1  72.000    
    5 1  91.000


    These particles were found in the raw micrograph, but not in the CTF-corrected micrograph.

  • B_not_A: doc file listing particles picked by B but missed by A. 
     ;bat/dat   05-NOV-2001 AT 13:26:16   b_not_a.dat
    1 1  1.0000    
    2 1  2.0000    
    3 1  7.0000    
    4 1  8.0000    
    5 1  21.000    
    6 1  27.000    
    7 1  57.000    
    8 1  65.000    
    9 1  72.000    
   10 1  73.000    
   11 1  75.000    
   12 1  76.000    
   13 1  85.000    
   14 1  88.000    
   15 1  90.000    
   16 1  93.000    
   17 1  100.00    
   18 1  108.00    
   19 1  113.00

    These particles were found in the CTF-corrected micrograph, but not in the raw micrograph. Note ser00076 here may be the same as ser00067 above, but they are displaced by a number of pixels greater than the threshold, and therefore are not listed as being the same particle.

  • Totals.dat a doc file listing numbers of particles found by each method. 
             picked      picked      picked      In A,       In B,
          by A        by B       by both    not in B    not in A

    1 5  108.00      122.00      103.00      5.0000      19.000

    Conclusions: both methods seemed to find particles that the other missed, but method B (CTF-correction) found more particles.

    It's not enough just to see how many "particles" were found by each method. You have to look at what they picked (perhaps they picked more junk).

    Source file: batch/pcomp/index.html     Source updated: 11/5/01    Bill Baxter