xmipp_ctf_sort_psds (v3.0)


Evaluate the CTFs and PSDs of a set of micrographs. This process is strongly coupled to the output produced by the preprocessing micrographs step of the Xmipp protocols. For each input PSD, the program writes its enhanced version since it is used in the computation of some of the criteria. The different criteria for evaluating the PSDs are:

this is the envelope value at the border of the PSD. Micrographs with a high envelope value at border are either wrongly estimated strongly undersampled.

First zero average
this is average in Angstroms of the first zero. Normally, this value should be between 4x and 10x the sampling rate in Angstroms.

First zero disagreement
if the CTF has been estimated by two different methods (normally Xmipp and Ctffind), then this criterion measures the average disagreement in Angstroms between the first zero in the two estimates. Low disagreements are indicative of correct fit.

First zero ratio
this measures the astigmatism of the CTF by computing the ratio between the largest and smallest axes of the first zero ellipse. Ratios close to 1 indicate no astigmatism.

Fitting score
the CTF is computed by fitting a theoretical model to the experimentally observed PSD. This criterion is the fitting score. Smaller scores correspond to better fits.

Fitting correlation between zeros 1 and 3
the region between the first and third zeroes is particularly important since it is where the Thon rings are most visible. This criterion reports the correlation between the experimental and theoretical PSDs within this region. High correlations indicate good fits.

PSD correlation at 90 degrees
The PSD of non-astigmatic micrographs correlate well with itself after rotating the micrograph 90 degrees. This is so because non-astigmatic PSDs are circularly symmetrical, while astigmatic micrographs are elliptically symmetrical. High correlation when rotating 90 degrees is an indicator of non-astigmatism. This criterion is computed on the enhanced PSD. See ctf_enhance_psd.

PSD radial integral
this criterion reports the integral of the radially symmetrized PSD. This criterion can highlight differences among the background noises of micrographs. This criterion is computed on the enhanced PSD. See ctf_enhance_psd.

PSD variance
the PSD is estimated by averaging different PSD local estimates in small regions of the micrograph. This criterion measures the variance of the different PSD local estimates. Untilted micrographs have equal defoci all over the micrograph, and therefore, the variance is due only to noise. However, tilted micrographs have an increased PSD variance since different regions of the micrograph have different defoci. Low variance of the PSD are indicative of non-tilted micrographs

PSD Principal Component 1 Variance
when considering the local PSDs previously defined as vectors in a multidimensional space, we can compute the variance of their projection onto the first principal component axis. Low variance of this projection is indicative of a uniformity of local PSDs, i.e., this is another measure of the presence of tilt in the micrograph.

PSD PCA Runs test
when computing the projections onto the first principal component, as discussed in the previous criterion, one might expect that the sign of the projection is random for untilted micrographs. Micrographs with a marked non-random pattern of projections are indicative of tilted micrographs. The larger the value of this criterion, the less random the pattern is.


-i <selfile>
Selfile with micrographs, it needs the columns image, psd and ctfmodel This file is modified by the addition of the evaluation criteria

Enhancement filter parameters

-f1 <freq_low=0.02>
Low freq. for band pass filtration, max 0.5
-f2 <freq_high=0.2>
High freq. for band pass filtration, max 0.5
-decay <freq_decay=0.02>
Decay for the transition bands
-m1 <mfreq_low=0.01>
Low freq. for mask, max 0.5
-m2 <mfreq_high=0.45>
High freq. for mask, max 0.5

User's comments