CTF estimation from Power spectrum

I am looking for a way to estimate CTF from average power spectrum.

In reply to by niko

Dear Niko,

I am looking for a way to get the defocus and astigmatism information from the power spectrum (not from the image).

I have made class averages from power spectra and try to get the defocus and astigmatism value to correct CTF for original particle images. Original images were made from the sample over the holey film in cryo condition. The CTF from the each particle was defined poorly due to the absence of carbon film in the imaged hole area and small particle sizes. After classification and averaging of the power spectra, more defined power spectra were obtained. CTFFIND3 seems to take only image as an input but not the power spectrum.

In reply to by bghan

Yes, you are correct that CTFFIND3 cannot do what you want. You could modify the program to read in power spectra instead. Alternatively, you can use one of the other available programs that come with EMAN2, Spider or IMAGIC.

A word about the classification method you are describing: Although you end up with nice power spectra in the class averages that you can use to determine the defocus, the accuracy of your measurements for individual particles can only be as good as your classification. If particles end up in the wrong class they will be assigned the wrong defocus. This may be a fundamental limitation of this procedure.

In reply to by niko

I am trying to compile ctffind3 and will change the program later to handle the power spectrum. Even though I could compile the program (ctffind3 as given), I got the error message "Segmentation fault". I have tried every Makefile options given. Also, I put the unlimited in the script and tried two different linux clusters from two separate groups. But I am getting same error message in all the trial. Do you have any suggestions for me to try?

In reply to by bghan

We have not had this problem. Is there any output before the program crashes? You could try an older version from our web page to see if the bug goes away.