[Chimera-users] Custom presets

Tue May 20 13:10:38 PDT 2014

Great, thanks Tom. You’re absolutely right, it should be RMS not RMSD, I misspoke. I’ll look forward to trying these options out, they should be very handy for dealing with X-ray data.

Best,
Oliver.
On May 20, 2014, at 3:59 PM, Tom Goddard <goddard at sonic.net> wrote:

> Hi Oliver,
> 
>  Ok, in tonight’s daily build there will be two new options to “vop scale”.  One “rms” scales the map so that sqrt(sum(v*v)/N) is 1 (“vop scale #0 rms 1”), and the other “sd” shifts the map so that the new map has mean 0 and then scales to make sqrt(sum(v*v)/N) equal to 1 (“vop scale #0 sd 1”).
> 
>  In my earlier email I incorrectly said the standard deviation value did not subtract the mean before squaring.  The standard deviation (SD) does subtract the mean before squaring.  The RMS (root mean square) value does not subtract the mean.  Also the terminology is RMS, not RMSD.  The term root mean square deviation (RMSD) is used for a comparison between two data sets and is not applicable to this map scaling discussion.
> 
>  Oddly, I find Uppsala electron density server maps have mean values different from 0.  This is probably because Chimera fetches the O format map (.omap) which has only one byte per value.  For instance Chimera reports 4pfd.omap (2fo-fc) to have mean -0.01 and SD 0.33, so the mean is off from 0 by about 3% of the SD.  The deviations from 0 are small and probably do not make a significant difference.  But that is where my idea came from that the mean was not 0.
> 
> 	Tom
> 
> 
> On May 20, 2014, at 12:06 PM, Oliver Clarke <olibclarke at gmail.com> wrote:
> 
>> It seems sensible to do the shift. 
>> 
>> I think Coot does the same think actually (although you’d need to check with Paul Emsley to be certain) - the map statistics output by Coot for every map that I’ve looked at, both from the EDS and elsewhere, have means that are arbitrarily close to zero (the highest I could find was 1e-6, and most are closer to 1e-11, too small to make any difference to the normalization).
>> 
>> Oliver.
>> 
>> 
>> On May 20, 2014, at 2:54 PM, Tom Goddard <goddard at sonic.net> wrote:
>> 
>>> Hi Oliver,
>>> 
>>> Ok, if I add a vop scale “rmsd" option it seems sensible that it should both shift the map to have mean 0 and scale so RMSD is 1.  Or should I not do the shift?  The reason I think the shift to make mean 0 is probably best is that without the shift you could not easily set a threshold at 2 sigma (2*RMSD above the mean) because you wouldn’t know the mean value.  But I’m uncertain about this since I don’t understand what the 0 value of the x-ray map represents in cases where it is not 0.  I could additionally have an option “mean 0” that would shift the map values to make the mean 0, if that is useful.
>>> 
>>> 	Tom
>>> 
>>> 
>>> On May 20, 2014, at 11:32 AM, Oliver Clarke <olibclarke at gmail.com> wrote:
>>> 
>>>> Hi Tom,
>>>> 
>>>> Good question, and I suspect if you asked different crystallographers you may receive conflicting responses. There was a good discussion on this point on the ccp4bb some time ago, but I can’t seem to find it right now.
>>>> 
>>>> I put sigma level in quotation marks precisely for this reason - people refer to the sigma level, but generally mean multiples of the RMSD. I know that Coot refers to rms values, and these values are often quoted as sigma levels. 
>>>> 
>>>> Would it be possible to include rmsd as a separate option for vop scale, so we can choose?
>>>> 
>>>> Best,
>>>> Oliver.
>>>> 
>>>> 
>>>> On May 20, 2014, at 2:13 PM, Tom Goddard <goddard at sonic.net> wrote:
>>>> 
>>>>> Hi Oliver,
>>>>> 
>>>>> I’m unclear about the meaning of “sigma level” in x-ray maps, and also uncertain now whether we agree on the definition of standard deviation.  By standard deviation I mean sqrt(sum(v_ijk*v_ijk)), in words, the squareroot of the sum of the square of the map values.  If instead, you first subtracted the mean map value then you would have the root-mean-square-deviation RMSD which will be a different value if the mean is not 0.   X-ray density maps from the Uppsala electron density server don’t have mean 0, so the difference is significant.  Is sigma level defined as RMSD units above the mean, or SD units above 0?
>>>>> 
>>>>> 	Tom
>>>>> 
>>>>> 
>>>>> On May 20, 2014, at 10:01 AM, Tom Goddard  wrote:
>>>>> 
>>>>>> Hi Oliver,
>>>>>> 
>>>>>> I added to tonight’s Chimera daily build a vop option to scale by one over the standard deviation: “vop scale #0 sd 1”  If you want the new map to have standard deviation 5 you could instead use “vop scale #0 sd 5”.  So the sd option value specifies the standard deviation of the new map.
>>>>>> 
>>>>>> 	Tom
>>>>>> 
>>>>>> 
>>>>>> On May 19, 2014, at 9:10 PM, Oliver Clarke wrote:
>>>>>> 
>>>>>>> Using # doesn’t seem to work with vop if there are any non-volume models, as it complains that they are not surfaces. Using the selection ~@ (everything that isn’t an atom) seems to work fine though.
>>>>>>> 
>>>>>>> In case it is of use to anyone else, here are the presets I came up with for X-ray maps. Seems to work okay as a starting point to then tweak by adjusting clipping, using vop zone on an atom selection, etc. Would like to normalize to RMS automatically but that will require a little more thinking.
>>>>>>> 
>>>>>>> window @
>>>>>>> set dcstart 0.4
>>>>>>> set dcend 1.0
>>>>>>> sop cap off
>>>>>>> vop zone ~@ @ 5
>>>>>>> volume # style mesh
>>>>>>> volume # squaremesh false
>>>>>>> volume # linethickness 1.5
>>>>>>> volume # meshlighting false
>>>>>>> volume # capfaces false
>>>>>>> background solid black
>>>>>>> scolor # color blue
>>>>>>> disp #
>>>>>>> ~ribbon #
>>>>>>> represent stick #
>>>>>>> modelcolor gold #
>>>>>>> color byhet #
>>>>>>> 
>>>>>>> On May 19, 2014, at 8:06 PM, Tom Goddard  wrote:
>>>>>>> 
>>>>>>>> For commands that want a specifier, I think you can always specify “all models” using “#” — just the # sign without a number.
>>>>>>>> 
>>>>>>>> 	Tom
>>>>>>>> 
>>>>>>>> 
>>>>>>>> On May 19, 2014, at 4:45 PM, Elaine Meng wrote:
>>>>>>>> 
>>>>>>>>> Hi Oliver,
>>>>>>>>> You could "hard-code" a model number or list of model numbers into a preset by simply including them in the commands in the preset definition file (e.g., one line in the file could be something like: color red,r #0).  Choosing the preset is the same as executing all the commands in the file.  For many commands, if you don't include a command-line atom specification, they will apply to all applicable items (e.g.: color red).  Similarly, the volume command takes specific model numbers or the word "all" to mean all currently open volume models.
>>>>>>>>> 
>>>>>>>>> If you mean you want to specify model number on the fly at the time of invoking a macro, that can only be done with aliases.  See the "alias" documentation for how to specify arguments to your macro:
>>>>>>>>> <http://www.rbvi.ucsf.edu/chimera/docs/UsersGuide/midas/alias.html>
>>>>>>>>> 
>>>>>>>>> One semi-exception to the above is that you could use the word "sel" in many commands to mean the currently selected models or atoms.  Then, you can interactively select only the parts of interest to be subsequently acted upon by the commands in your alias or preset that have "sel" as the atom/model specification.  See the individual command man pages for how atoms/models are specified, as sometimes the specification needs to be preceded by a keyword.
>>>>>>>>> <http://www.rbvi.ucsf.edu/chimera/docs/UsersGuide/selection.html>
>>>>>>>>> <http://www.rbvi.ucsf.edu/chimera/docs/UsersGuide/framecommand.html>
>>>>>>>>> 
>>>>>>>>> You can test your commands in the command line and then save them to a file using the command history, and/or test Chimera command files (*.com,*.cmd) and Python files (*.py) by simply opening them with File… Open.
>>>>>>>>> <http://www.rbvi.ucsf.edu/chimera/docs/UsersGuide/history.html>
>>>>>>>>> <http://www.rbvi.ucsf.edu/chimera/docs/UsersGuide/filetypes.html>
>>>>>>>>> 
>>>>>>>>> I hope this helps,
>>>>>>>>> Elaine
>>>>>>>>> ----------
>>>>>>>>> Elaine C. Meng, Ph.D. 
>>>>>>>>> UCSF Computer Graphics Lab (Chimera team) and Babbitt Lab
>>>>>>>>> Department of Pharmaceutical Chemistry
>>>>>>>>> University of California, San Francisco
>>>>>>>>> 
>>>>>>>>> On May 19, 2014, at 4:18 PM, Oliver Clarke wrote:
>>>>>>>>> 
>>>>>>>>>> Hi Elaine,
>>>>>>>>>> Thanks for the info! How do I specify the model or map number in the preset command? Is there a wildcard I can use? 
>>>>>>>>>> Usually I would be specifying the model or map number explicitly in any commands that I use, but presets seem to be applied to all open models or maps - how do I do that?
>>>>>>>>>> Best,
>>>>>>>>>> Oliver.
>>>>>>>>>> On May 19, 2014, at 1:00 PM, Elaine Meng wrote:
>>>>>>>>> 
>>>>>>>>> 
>>>>>>>>> _______________________________________________
>>>>>>>>> Chimera-users mailing list
>>>>>>>>> Chimera-users at cgl.ucsf.edu
>>>>>>>>> http://plato.cgl.ucsf.edu/mailman/listinfo/chimera-users
>>>>>>>>> 
>>>>>>>> 
>>>>>>> 
>>>>>>> 
>>>>>>> _______________________________________________
>>>>>>> Chimera-users mailing list
>>>>>>> Chimera-users at cgl.ucsf.edu
>>>>>>> http://plato.cgl.ucsf.edu/mailman/listinfo/chimera-users
>>>>>>> 
>>>>>> 
>>>>> 
>>>> 
>>>> 
>>> 
>> 
>> 
> 