Project

General

Profile

Actions

Bug #3547

open

ctffind4 hard to control defocus range

Added by William Rice over 9 years ago. Updated over 9 years ago.

Status:
In Code Review
Priority:
Normal
Assignee:
Category:
-
Target version:
-
Start date:
09/10/2015
Due date:
% Done:

0%

Estimated time:
Affected Version:
Appion/Leginon 3.2
Show in known bugs:
Workaround:

Description

Ctffind4 seems unable to find the right defocus if it is far from the "nominal" defocus. I tried adding a large range (100 steps of 0.1 um) but this had is search a very small range.

From the log:
Starting image 1 ( skip:0, remain:50 ) id:294181, file: 15sep01d_02en-a
... Pixel size: 1.47
[CTF run] method: unknown | runname ctffind4run2
[CTF param] def1: 1.93e-06 | def2: 1.95e-06 | angle: 48.5 | ampcontr
0.07 | defratio 1.011
[CTF stats] conf_30-10: -0.07 | conf_5peak:
0.007 | res_0.8: 100.0A | res_0.5 100.0A
Defocus search range: 19276 A to 19296 A (1.93 to 1.93 um)
... running ctf estimation at Thu Sep 10 07:48:42 2015
input = 15sep01d_02en-a.mrc
output = 15sep01d_02en-a-pow.mrc
apix = 1.47
kv = 200.0
cs = 2.0
ampcontrast = 0.07
fieldsize = 1024
resmin = 50.0
resmax = 7.0
defmin = 19276.6
defmax = 19296.6
defstep = 1000.0
expect_astig = 500.0
phase = no
newline =


Files

Screenshot - 09132015 - 08_18_44 PM.png (599 KB) Screenshot - 09132015 - 08_18_44 PM.png ctffind 2d graph William Rice, 09/13/2015 08:21 PM

Related issues 1 (1 open0 closed)

Related to Appion - Feature #3950: Create CTF test dataset within DockerAssignedNeil Voss02/15/2016

Actions
Actions #1

Updated by William Rice over 9 years ago

sample sessions over carbon to check:
15sep01d - yACOB gOMEZ HOLDER TEST
15sep09k - test drift after OL flow adjustment and zeolite cycle

Actions #2

Updated by Anchi Cheng over 9 years ago

  • Project changed from 138 to Appion
  • Status changed from New to In Code Review
  • Assignee set to Neil Voss
  • Affected Version set to Appion/Leginon 3.2

Defocus input to ctffind4 is in angstroms. Fixed in r19140

Actions #3

Updated by William Rice over 9 years ago

Program seems to want defocus step in um. Entering 1000 as the step results in following error:

... Committing data to database
Traceback (most recent call last):
File "/opt/myamisnap/bin/ctffind4.py", line 325, in
imgLoop = ctfEstimateLoop()
File "/opt/myamisnap/lib/appionlib/appionLoop2.py", line 27, in init
appionScript.AppionScript.__init__(self)
File "/opt/myamisnap/lib/appionlib/appionScript.py", line 83, in init
self.checkConflicts()
File "/opt/myamisnap/bin/ctffind4.py", line 63, in checkConflicts
apDisplay.printError("Please keep the defstep between 0.0001 & 2 microns")
File "/opt/myamisnap/lib/appionlib/apDisplay.py", line 65, in printError
raise Exception, colorString("\n * FATAL ERROR *\n"+text+"\n\a","red")
Exception:
  • FATAL ERROR ***
    Please keep the defstep between 0.0001 & 2 microns
Actions #4

Updated by Anchi Cheng over 9 years ago

Bill,

Sorry, I meant that in the code the gui value was supposed to convert to Angstrom before it calculates the range to input to ctffind4 program but the old code did not do that, which is why it gave tiny range when you put in the proper value in micron in gui.

The change will become live tomorrow.

Actions #5

Updated by William Rice over 9 years ago

Ctffind4 still seems to fail when actual defocus is far from nominal. It seems to convert properly from um to the expected A (see log below) but defocus it found is obviously wrong (see attached screenshot). Looking at the logs it seems to often have strange values for resmax (here 37A; should be 6 as entered) and expected astig (here 83.7; should be 300 as entered)

Defocus search range: 1000 A to 59408 A (0.10 to 5.94 um)
... running ctf estimation at Sun Sep 13 20:10:05 2015
input = 15sep09k_013sq_01hl_04en-a.mrc
output = 15sep09k_013sq_01hl_04en-a-pow.mrc
apix = 1.47
kv = 200.0
cs = 2.0
ampcontrast = 0.07
fieldsize = 1024
resmin = 50.0
resmax = 37.03
defmin = 1000.0
defmax = 59408.9
defstep = 1000.0
expect_astig = 83.7
phase = no
newline =

/usr/local/bin/ctffind4
15sep09k_013sq_01hl_04en-a.mrc
15sep09k_013sq_01hl_04en-a-pow.mrc
1.47
200.0
2.0
0.07
1024
50.0
37.03
1000.0
59408.9
1000.0
83.7
no

Actions #6

Updated by Neil Voss over 9 years ago

  • Assignee changed from Neil Voss to Anchi Cheng

ResMax and astig are auto-assigned if using '*Use best values from Database*' is checked to help get a better fit. If you want only you value uncheck the box. For some reason the pop up help is not working the CTFFIND4 page, but it does specify as such on the CTFFIND3 page.

I forgot to include user resmax, so should be better now, so we should have a smarter resmax setting in r19149.

      if ctfvalue is not None and self.params['bestdb'] is True:
         ### set res max from resolution_80_percent
         gmean = (ctfvalue['resolution_80_percent']*ctfvalue['resolution_50_percent']*self.params['resmin'])**(1/3.)
         if gmean < self.params['resmin']:
            # replace only if valid Issue #3291
            self.params['resmax'] = round(gmean,2)
            apDisplay.printColor("Setting resmax to the geometric mean of resolution values", "purple")
Actions #7

Updated by Anchi Cheng over 9 years ago

  • Assignee changed from Anchi Cheng to Neil Voss

The code change in ctfestimate.py probably has wrong indentation. gmean is not defined if self.params['bestdb'] is False. Therefore if gmean < self.params['resmin'] will crash the program.

Actions #8

Updated by Anchi Cheng over 9 years ago

Also, resmin is at very low resolution, so why replace resmax with a value that depends on it ?

Actions #9

Updated by Neil Voss over 9 years ago

  • Assignee changed from Neil Voss to Anchi Cheng

Thanks for checking Anchi, should be better now.

Actions #10

Updated by Neil Voss almost 9 years ago

  • Related to Feature #3950: Create CTF test dataset within Docker added
Actions

Also available in: Atom PDF