Project

General

Profile

1D Component CTF Plot Resolution Limits

Added by Jason van Rooyen over 9 years ago

Hi team,

Is there a way to extend the high-resolution limit of the 1D CTF-fit plot?

I am calculating CTFs on binned K2 images with a final pixel size of 1.7 A/pixel and I am using no binning in CTF estimation (bin = 1). The problem is that the upper limit of the plot is given at 5.1 A. This is written as 1/5.1 A.

I would like to use these plots to sort our micrographs.

Is it possible to set a higher value?

Thanks,
Jason


Replies (4)

RE: 1D Component CTF Plot Resolution Limits - Added by Jason van Rooyen over 9 years ago

This limit also seems to set the 0.5 cut-off value estimation.

RE: 1D Component CTF Plot Resolution Limits - Added by Jason van Rooyen over 9 years ago

I found the following lines in ctftools.py:

def drawPowerSpecImage(self, origpowerspec, maxsize=1200, outerresolution=7.7):

def CTFpowerspec(self, imgdata, ctfdata, fftpath=None, fftfreq=None, outerbound=5e-10, twod=True):

...and changed the outerresolution and outerbound values to 3.4 and 3e-10 respectively.

This has extended the CTF plots and allowed resolution estimatino past 5.

RE: 1D Component CTF Plot Resolution Limits - Added by Anchi Cheng over 9 years ago

jason,

If you are on 3.1 branch, you may want to look into the changes made in trunk of appion/appionlib/apCtf since 3.1 release.

Quite a lot has changed there to address this problem as in Issue #2970. For example, your change is in r18376 and the change was in ctfdisplay.py not ctftools.py,

RE: 1D Component CTF Plot Resolution Limits - Added by Neil Voss over 9 years ago

Hi,

It is much easier to change/find in trunk. I know in the trunk version, we have changed it to:

self.outerAngstrom1D = 3.0
# plotlimit2DAngstrom trims the power spectrum generated
#from self.outerAngstrom1D limit for the 2D plot
self.plotlimit2DAngstrom = 5.0

But the myami-3.1 is still at outerbound=5e-10 and outerresolution=7.7; I think I will check in the above limits to myami-3.1

I think the 2D power spectra is hard to read past 5A, but definitely the 1D could go further. I have made improvements to the "trisection" fit from my paper into a "N-section" fit, which should work better at these higher resolution, but I have done enough testing to check it in yet.

Neil

    (1-4/4)