defocus refinement in mrefine_sge.com

Submitted by mjb on Thu, 2011-09-01 07:12

Forums

Frealign

mrefine_sge.com splits up particles into batches without taking their micrograph number into account.

Together with FDEF=T, FASTIG=T, FPART=F (and in this case a split-up "increment" of 1000) this leads to the following result:


1999 62.33 207.08 162.42 -2.50 -7.88 71950. 74 28786.7 29588.1 45.00 64.67 64.67
2000 176.60 252.38 222.86 1.42 -2.67 71950. 74 28786.7 29588.1 45.00 62.02 62.02
2001 292.27 195.89 294.59 -1.01 0.14 71950. 74 28291.8 29275.2 1190.92 63.59 63.59
2002 0.61 250.89 214.22 -3.87 -1.83 71950. 74 28291.8 29275.2 1190.92 67.20 67.20

This refinement differences are especially strong when only a few particles from a micrograph "slip" over to the next particle batch.
Is there a solution for this (except increasing the increment)?

Best regards
Matthias

mjb Thu, 2011-09-01 12:34

And along the same lines: Is a line like the following in a par file generated by frealign refinement normal or is the very big astigm. angle (no spaces!) indicating that something is going very wrong?


48658 102.57 254.10 351.23 1.38 2.80 71950. 1577 29900.0 29900.018379.64 67.09 -0.17

version: frealign_v8.09_110505

params:

thresh_reconst 90.0
thresh_refine 25.0
pixel_size 2.08
dstep 15.0
WGH 0.07
CS 2.0
kV1 300.0
radius 185
PBC 100.0
BOFF 35.0
DANG 200.0
ITMAX 50
MODE 1
XSTD 0.0
RBFACT 0.0
FMAG F
FDEF T
FASTIG T
FPART F
dfsig 150.0
IEWALD 0
res_reconstruction 7.89
res_low_refinement 250.0
res_refinement 13.0
start_process 2
end_process 10
first_particle 1
last_particle 70392
increment 1000
symmetry C1

box size 240x240, data has previously been reconstructed to a resolution of ~12A in xmipp...

niko Thu, 2011-09-01 13:19

You will need to change the script to split the stack into batches of particles collected from the same micrograph. I attach a script that I wrote to do this. You will need to adapt it to your own needs.