[Chimera-users] Calculating sigma level for EM map

[Thomas Goddard]

Hi Andy,

   I don't know any convention defining the sigma value for an EM map. 
For x-ray maps there is a convention because the density is periodic you 
can just take the root-mean square density value across a whole unit 
cell.  Note that that isn't a measure of noise level -- it's just a 
reproducible value describing the typical density levels across the unit 
cell.  That isn't useful in EM maps because the value depends on how big 
the box is that contains the non-periodic structure.  A bigger box 
contains more grid points outside the structure (values near zero) that 
make the root-mean square value (RMS) smaller.  So such an RMS value 
isn't a reproducible measure of density values since it depends on the 
box size.  I think when sigma is mentioned in EM papers they tend to 
mean the noise level.  I don't know of any standard protocol for 
determining the noise level.

   If you knew the sigma level you wished to use, Chimera is still not 
able to display density values in sigma units.  That is on our feature 
request list.  Currently you have to take the values Chimera displays 
and divide by the sigma value yourself.  I haven't added this capability 
yet because the notion of sigma does not seem to be well defined in EM 
work.  Since Chimera map analysis is primarily used for EM, not x-ray, 
showing sigma units has not been a high priority.

	Tom


-------- Original Message --------
Subject: Re: Threshold
From: Anindito Sen
To: Tom Goddard
Date: 2/4/10 1:09 AM

> Hi Tom
> 
> I should have been more specific- my reconstruction is a EM density map 
> of a helical biopolymer. In that case what should be  the correct 
> approach to calculate the threshold. Can you site any reference which 
> may be useful in this regard? 
> 
> Do we have the option of directly obtaining  sigma value for the 
> threshold in Chimera ?
> 
> Dr. Anindito Sen (Ph.D)
> 
> --- On *Thu, 4/2/10, Tom Goddard* wrote:
> 
> 
>     From: Tom Goddard
>     Subject: Re: Threshold
>     To: "Anindito Sen" 
>     Date: Thursday, 4 February, 2010, 9:29
> 
>     Hi Andy,
> 
>       That may be right.  But there are many issues.  If your map is for
>     a crystal unit cell then that seems reasonable.  In other cases
>     where the map is not a crystal unit cell then the RMS of the whole
>     map will depend on what the boundaries are of the map.  For
>     instance, for an EM single particle reconstruction the RMS will
>     depend on how much padding of zeros there is, ie how big the box
>     is.  If it is an x-ray map but not a unit cell again the RMS will
>     vary depending on what part of the map is taken.
> 
>         Tom
> 
> 
>     -------- Original Message --------
>     Subject: Threshold
>     From: Anindito Sen
>     To: Tom Goddard
>     Date: 2/3/10 7:23 PM
>>     *
>>
>>     Hi Tom 
>>
>>      I am "cropping" out high electron dense region from my 
>>      density map to determine the volume of that region. The value of
>>     the Level in my Volume viewer is 438 (Range -618- 1024). The other
>>     factors are :
>>
>>     f15_volume035.brix: mean = 23.123, SD = 137.14, RMS = 139.07
>>
>>     Therefore following your email, can I say, threshold is 438/139.07
>>     = 3.1s
>>
>>     Thanks for your help.
>>
>>     Best Wishes
>>
>>     Andy
>>
>>
>>
>>     Tom Goddard* 
>>     /Tue May 20 15:54:10 PDT 2008/
>>     ------------------------------------------------------------------------
>>     Hi Dave,
>>
>>       Chimera can report the mean, standard deviation and root mean square 
>>     map value using menu entry
>>
>>         Tools / Volume Data / Volume Mean, SD, RMS
>>
>>     The values using whatever normalization the map file has.  If you divide 
>>     the threshold level you set in the volume dialog by the RMS value you 
>>     will get a sigma value.  Actually I'm not sure if that is correct -- 
>>     depends on the definition of sigma.  Maybe you have to take the volume 
>>     dialog number, subtract the mean and divide by the SD.  The reported SD 
>>     and RMS differ only in that the SD is the standard deviation of the map 
>>     about the mean while the RMS is the standard deviation about 0.
>>
>>       Allowing the volume dialog to show the sigma values directly is on our 
>>     requested features list (item 88) but has not yet been implemented.
>>
>>       http://www.cgl.ucsf.edu/chimera/plans.html
>>
>>         Tom
>>
>>
>>     David Chenoweth wrote:
>>     >/ Hi!
>>     />/
>>     />/ Can you please let me know if there is a way to report or set the  
>>     />/ sigma value for an electron density map from a high resolution crystal  
>>     />/ structure. I am using a ccp4 FWT.map and DELFWT.map file for  
>>     />/ structures <1.2 angstrom resolution. Can I somehow find the sigma  
>>     />/ value that corresponds to the Level setting in the Volume viewer?
>>     />/
>>     />/ Thanks,
>>     />/ Dave Chenoweth
>>     />/   
>>     /
>>             
>>     ------------------------------------------------------------------------