For the resulting FSC curve(s) to be valid, the averages in each pair must share a common position and orientation. When enforcing symmetry, for example, transforms may have been applied bringing each average to some standard orientation. In this case, calcUnbiasedFSC can be applied directly.
Alternatively, a separate PEET run can be done to align the 2 averages from each pair. To perform such an alignment, create a PEET project in a new directory, using the first average as volume 1 and the reference, the second average as volume 2, and specifying a model with a single point at the volume center for each. Set "flgNoReferenceRefinement = 1" in the parameter file and choose appropriate angular and translation search ranges, using as many iterations as necessary. When the alignment is complete, read the necessary Euler angles, Z1, X, and Z2 from columns 17, 19, and 18, respectively, of the output motive list for volume 2 and shifts, DX, DY, and DZ from columns 11-13. Change your working directory back to the project where alignment search for the second member of the pair was done, and for each final iteration motive list present execute:
cp MOTL1.csv MOTL1.csv.orig
modifyMotiveList MOTL1.csv.orig MOTL1.csv "Z1,X,Z2" "DX,DY,DZ" 0 1
where Z1, X, Z2, DX, DY, and DZ are to be replaced by the corresponding values, and MOTL1 by the actual name of the motive list.
CalcUnbiasedFSC can also be used to compute the Fourier Shell Correlation between a subvolume average and a crystal structure or other estimate of the true structure in a similar manner.
In most regards, calcUnbiasedFSC behaves similarly to calcFSC(1), accepts similar arguments, and produces output suitable for plotting with plotFSC(1). Two differences are worth noting. First, calcFSC's repCount is ignored by calcUnbiasedFSC, with the number of parameter file pairs provided taking its place. When multiple pairs are provided, the resulting FSC will be the mean of results from the individual pairs, and confidence limits will be stored in 'arrSigma.txt' for optional plotting. Second, results produced by calcUnbiasedFSC will often produce more legible plots when correlation is plotted versus spatial frequency rather than resolution. See the plotFSC(1) man page for the appropriate options.