From: REX PALMER
Date: 9 September 2011 16:20
----------
From: Ian Tickle
Hi Rex
I would assume it's the RMS deviation of the rotation function value from the mean. If so then it's defined in the same way as the "population standard deviation" as given here: http://en.wikipedia.org/wiki/Standard_deviation
Cheers
-- Ian
----------
From: Ian Tickle
Rex, sorry, addendum to my last post: the grid points in the Eulerian space of the cross-RF (which is what I assume you're talking about) do not occupy equal volumes in angular space as they would do in a Euclidean space (e.g. a normal electron density map). For example the volume of any grid point on the beta=0 & 180 sections is zero (these sections actually each consist of a line of points, and a line of course occupies no volume). Therefore it's necessary to weight the values in both the mean and the RMSD by the volume of each grid point (= sin(beta)). Whether ALMN actually does this or not would require an examination of the source code, though it's probably quicker to ask Eleanor!
Cheers
-- Ian
----------
From: Eleanor Dodson
Yes - the RMS for the map is weighted by sin beta
Here is the comment:
C Mean & RMS need to be weighted by sin beta, to allow for distortions
c in Eulerian space.
c On the last section, if beta = 90, the weight is 1/2 because of
c symmetry. Here we only need to test for last section, because
c (a) if last section = 180deg, the weight is zero anyway
c (b) if the last section is not 90 or 180, the rms is not valid anyway
Eleanor
Date: 9 September 2011 16:20
Dear CCP4'ers
In the input instructions for ALMN it says:
FIND peak maxpek [RMS] [OUTPUT filename]
Read peak threshold and maximum number of peaks. MAXPEK is the maximum number of peaks to find (default = 20). Up to MAXPEK peaks above PEAK will be found, and all symmetry related peaks generated.
If the keyword RMS is present, then the peak threshold is PEAK * RMS density, otherwise PEAK is the absolute threshold in the scaled map.
Question: RMS density of what and what is the formula used?
----------
From: Ian Tickle
Hi Rex
I would assume it's the RMS deviation of the rotation function value from the mean. If so then it's defined in the same way as the "population standard deviation" as given here: http://en.wikipedia.org/wiki/Standard_deviation
Cheers
-- Ian
----------
From: Ian Tickle
Rex, sorry, addendum to my last post: the grid points in the Eulerian space of the cross-RF (which is what I assume you're talking about) do not occupy equal volumes in angular space as they would do in a Euclidean space (e.g. a normal electron density map). For example the volume of any grid point on the beta=0 & 180 sections is zero (these sections actually each consist of a line of points, and a line of course occupies no volume). Therefore it's necessary to weight the values in both the mean and the RMSD by the volume of each grid point (= sin(beta)). Whether ALMN actually does this or not would require an examination of the source code, though it's probably quicker to ask Eleanor!
Cheers
-- Ian
----------
From: Eleanor Dodson
Yes - the RMS for the map is weighted by sin beta
Here is the comment:
C Mean & RMS need to be weighted by sin beta, to allow for distortions
c in Eulerian space.
c On the last section, if beta = 90, the weight is 1/2 because of
c symmetry. Here we only need to test for last section, because
c (a) if last section = 180deg, the weight is zero anyway
c (b) if the last section is not 90 or 180, the rms is not valid anyway
Eleanor
No comments:
Post a Comment