Hi Sjors,
Yes, that makes sense, you don't want to load the full size images only
to calculate the FSC during refinement. My concern was that if you
calculate an FSC between these two maps (or you look at the FSC in the
model.star file) there
is a strange increase in correlation at very high frequencies, which is
not there when you do not use the masked fsc option. In the beginning I
thought it was a masking problem, but now I realize that it simply comes
from calculating the fsc at frequencies that are not longer present in
the maps.
Best
Felipe
On 03/11/16 09:21, Sjors Scheres wrote:
> Hi Felipe,
> During the iterations, these unfil.mrc are indeed unfiltered. However, the
> behaviour you see comes from the fact they are not necessarily calculated
> all the way out to Nyquist (to save computer resources). That's why it
> appears as 'low-pass filtered'.
> HTH,
> Sjors
>
>> Dear all,
>>
>> I've been recently testing the new option in Relion to use the masked
>> half maps during refinement to calculate the resolution
>> (--solvent_correct_fsc). During the processing the program writes many
>> _unfil.mrc files which I assumed are non-filtered volumes, similar to
>> the final one. However, I noticed that they have some sort of low-pass
>> filter applied, which can be seen in the model.star files and the 1D
>> power spectrum (see the 1D power spectrum attached). Is this intended
>> and the name _unfil.mrc is just unfortunate or are these supposed to be
>> really unfiltered volumes?
>>
>> Best
>>
>> Felipe
>>
>>
>> --
>> Felipe Merino
>> Max Planck Institute for Molecular Physiology
>> Department of Structural Biochemistry
>> Otto-Hahn-Str. 11
>> 44227 Dortmund
>> Phone: +49 231 133 2315
>>
>>
>
--
Felipe Merino
Max Planck Institute for Molecular Physiology
Department of Structural Biochemistry
Otto-Hahn-Str. 11
44227 Dortmund
Phone: +49 231 133 2315
|