CTF correction

Submitted by hsosa on Fri, 2010-07-30 14:32

Forums

ctffind & ctftilt

Is there a program, other than frealign, to CTF correct (amplitudes or just phase flipping) a single image using the values obtained from ctffind3 (i.e. correction taking into account some astigmatism). I have a few programs that correct CTF on single images but they all assume no astigmatism.

Thanks

Alexis Fri, 2010-07-30 14:50

Reasonably recent versions of imagic do 2D CTF flipping including astigmatism (commands ctf2d-flip or ctf2d-operations). You would need to convert the CTF parameters given by CTFFIND to the convention used within imagic to describe astigmatism (of course, you'd also need to convert your images to imagic file format).
If that sounds like something you'd like to do, I can probably look up the conversion for you.

Axel Mon, 2011-03-07 11:41

Hi Joshua,

I recently had to solve the same problem, so I thought I would pipe in here.

The angular conventions follow the following formula. Imagic defocus angle = 90 - ctffind3 defocus angle

I have found that imagic headers are very good for doing the bookkeeping. Thus, I usually write out particles for each micrograph into an Imagic stack (using e2boxer).

Here is a short example of how to get defocus parameters out of a ctffind3 output file, and correct the defocus angle.

set defocus1 = `grep Final $FILM.ctf | cut -d" " -f5`
set defocus2 = `grep Final $FILM.ctf | cut -d" " -f9`
set ctffind_defangle = `grep Final $FILM.ctf | awk '{print $3}'`
#correct ctffind to imagic defangle convention
set defangle = `calc 90-$ctffind_defangle | awk '{print $3}'`

I then call a script that modifies the headers using the following script

./headers.b $F_INPUT $defocus1 $defocus2 $defangle

where defangle is the defocus angle using the imagic convention.

headers.b contains the following:

echo "! IMAGIC program: headers - Writing to $1 Def1 $2 Def2 $3 Defangle $4"

/gware/imagic/linux64/stand/headers.e WRITE
LABEL
LABEL
DEFOCUS1
INT
$2
$1
EOF

/gware/imagic/linux64/stand/headers.e WRITE
LABEL
LABEL
DEFOCUS2
INT
$3
$1
EOF

/gware/imagic/linux64/stand/headers.e WRITE
LABEL
LABEL
DEFANGLE
INT
$4
$1
EOF

I hope this helped.

Axel

PS. Although I have tried to make sure that everything in these scripts is correct, you may want to troubleshoot them yourself. Let me know if you find any inconsistencies.

Joshua Wed, 2011-08-24 12:14

Hello Axel,

Can IMAGIC handle the parameter from ctftilt? I think it does not consider the tiltangle.
What's your suggestion to do ctf correction from ctftilt.(Spider? XMIPP? MRC image processing suite?) And how to convert the parameter?

Axel Wed, 2011-08-24 14:43

Hi Joshua,

As far as I can tell (from asking Alexis) Imagic does not handle the tilt parameter. He suggests you could try it, however.

Whenever I have performed RCT reconstruction, I have just used Spider, and have relatively little experience using ctftilt.

Sorry I couldn't be of more help.

Axel

Alexis Wed, 2011-08-24 14:53

Joshua,

As Axel said, I don't think there is built-in functionality in Imagic to use results from CTFTILT.
However, you could use an approach similar to Axel's descrption above, where you would compute the defocus parameter per particle (as described by equation printed out near the end of the CTFTILT output), and then pipe the per-image defocus values into the headers for your stack of images.
Then you could use ctf2d-flip (for example) to correct the CTF within imagic.

I think the latest versions of Imagic feature tighter integration with CTFFIND, but I haven't had a chance to play with this yet.

Joshua Thu, 2011-08-25 00:52

Thanks, Alexis ＆ Axel
I wonder if there is an efficient way to compute the defocus and update the headers.
Or should I use Excel to key in every location of the particles and calculate the defocus then update the headers one by one.

Alexis Thu, 2011-08-25 09:24

Hi Joshua,
You must have a list of coordinates of the particles you picked, right? You can use this list as an input to a script to handle the calculations for the defocus values, and the same script can be used to plug these values into the image headers, as described above by Axel.
HTH
Alexis

niko Fri, 2010-07-30 16:28

You can also use the program ctffinda from the MRC image processing suite. ctffinda was the 'prototype' of ctffind3. It takes the output defocus parameters of ctffind3 directly (no conversion necessary). ctffinda has an IAPPLY flag which has to be set to T (for true) to apply the CTF.

hsosa Mon, 2010-08-02 13:07

Thanks all. I will probably use the MRC or the spider routines (thanks CS for the angle conversion code) as I dont have imagic available right now.

DS57 Tue, 2011-01-18 19:47

Dear All,

I'd like to correct for the CTF using the method that cvsindelar (Chuck) suggested i.e. using TF CT with inputs from ctffind3; but want to double-check that several of the input parameters that have not been specified in this thread are correct:

Firstly, my real space micrograph image is 9680 x 9680 pixels; does this correspond to the number of spatial frequency points in the TF CT operation?

Secondly, the micrograph is discretized with a pixel size of 1.33A/pixel; is the maximum spatial frequency in the TF CT operation 1/(2*1.33) ~ 0.38 A-1?

Having computed the binary phase contrast transfer function (with a sign of -1), I've multiplied this with the FT of my micrograph, and computed the inverse Fourier transform to produce the CTF corrected image. Is this correct?

I'd be extremely grateful for any responses -- I obviously need to make sure that this step is done properly.

Thanks!

DS57 Wed, 2011-01-19 08:27

Thank you very much for your quick response -- especially clarifying point 1. I thought the documentation was misleading!

I was planning to check that there was correspondence between the calculated and simulated PS, so I will certainly follow up that suggestion.

Thanks again!