Yes, it can certainly be error-prone and subjective. A nice survey of issues 
and scoring functions (including use of Laplacian filter) is in 
http://www.ncbi.nlm.nih.gov/pubmed/21296161

You might find some useful tips in our recent workshops:

http://www.ccpem.ac.uk/training/icknield_2016/icknield_2016.php
http://i2pc.cnb.csic.es/hands-on-course-may-2016.html

In practice, most EMers use Chimera with manual placement followed by local 
optimisation. Still plenty of scope of over-imaginative fitting...

HTH
m

From: CCP4 bulletin board [mailto:[email protected]] On Behalf Of Filip Van 
Petegem
Sent: 08 August 2016 20:09
To: ccp4bb
Subject: Re: [ccp4bb] Docking crystal structure into EM model

Hi Zbyszek,

I'm referring to the fitting contrast increase in the Situs methods slide:
http://situs.biomachina.org/fft.pdf

Perhaps this is what you meant anyway, but there's no explicit mention of 
"image sharpening".  The main advantage, as I understand it, is to increase 
fitting contrast by making the inside density negative, and the contour 
positive.  This allows both contour and volume matches to contribute to the 
correlation coefficient.  From experience using this on many occasions, the 
results are far superior compared to using standard cross-correlation 
coefficients (important, this is with cryo-EM maps at resolutions worse than 
10A). Where standard correlation coefficients failed miserably, the filtered 
method allowed the correct location an orientation to be found, as later 
verified experimentally.


Filip Van Petegem



On Mon, Aug 8, 2016 at 11:41 AM, 
<[email protected]<mailto:[email protected]>> wrote:
Laplacian filter is a particular form of image sharpening, being a
high-pass filer with coefficient increasing quadratically with resolution.
If it produces more interpretable map image, the map density definitely
needs some form of sharpening.

For experimental images of atomic structures, there is no theoretical
justification why Laplacian filter should be better than Gaussian
sharpening, with adjustable B-factor, as present in Coot, for example.

Zbyszek Otwinowski

> Hi Sam,
>
> I highly recommend the 'colores' routine in Situs (
> http://situs.biomachina.org/).  It has the advantage of being able to
> apply
> a Laplacian filter to enhance surface matching. This feature is quite
> crucial when the resolution of the map is not very high, and when the
> crystal structure only represents a small fraction of the entire
> structure.
>
> Best regards,
>
> Filip Van Petegem
>
>
>
> On Mon, Aug 8, 2016 at 10:00 AM, Sam Tang 
> <[email protected]<mailto:[email protected]>> wrote:
>
>> Dear all
>>
>> Sorry for this thread which is slightly off-topic.
>>
>> We have recently got a crystal structure of a protein which forms part
>> of
>> a higher order complex, the cryo-EM model of which has been solved and
>> published. In order to visualize some structural features we try to
>> place
>> this protein into the EM map of the complex in Pymol manually. However
>> limitation in resolution of the EM map renders this process error-prone
>> and
>> indeed subjective.
>>
>> I wonder if anyone has similar experience and could advise a better
>> strategy in doing so.
>>
>> Thanks in advance.
>>
>> Sam
>>
>> Biochemistry Programme, School of Life Sciences, CUHK
>>
>>
>
>
> --
> Filip Van Petegem, PhD
> Associate Professor
> The University of British Columbia
> Dept. of Biochemistry and Molecular Biology
> 2350 Health Sciences Mall - Rm 2.356
> Vancouver, V6T 1Z3
>
> phone: +1 604 827 4267<tel:%2B1%20604%20827%204267>
> email: [email protected]<mailto:[email protected]>
> http://crg.ubc.ca/VanPetegem/
>

Zbyszek Otwinowski
UT Southwestern Medical Center at Dallas
5323 Harry Hines Blvd.
Dallas, TX 75390-8816
Tel. 214-645-6385<tel:214-645-6385>
Fax. 214-645-6353<tel:214-645-6353>



--
Filip Van Petegem, PhD
Associate Professor
The University of British Columbia
Dept. of Biochemistry and Molecular Biology
2350 Health Sciences Mall - Rm 2.356
Vancouver, V6T 1Z3

phone: +1 604 827 4267
email: [email protected]<mailto:[email protected]>
http://crg.ubc.ca/VanPetegem/

Reply via email to