[Chimera-users] Higher Order

Tom Goddard goddard at cgl.ucsf.edu
Wed Jan 21 11:30:43 PST 2009

Hi Hernando,

  From your image it looks like your surface has a lot of detail -- a 
lot of triangles.  A fast graphics card will allow you to rotate it more 
smoothly.  It won't speed up calculating the surface the first time you 
show it.  The Chimera graphics benchmark page is the best info I have 
about cards:


The top two cards in the table differ by a factor of 3 in price but by 
just a tiny amount in surface rendering speed.  So it depends how much 
money you want to spend.  The top half of the table are all pretty 
fast.  And many standard cards like Geforce 8800 GTX are not even listed.

A second probably more important consideration is how buggy is the 
graphics driver for the card.  Our information on this is incomplete and 
out of date.  Here's that info:



hsosa at aecom.yu.edu wrote:
> Dear Tom , Elaine David
>   Thanks a lot for the help.  As suggested we placed helical symmetry 
> information in the BIOMT entries of the  PDB file  (3EDL). A 
> representation of the molecule made with chimera using the Higher 
> Order Structures tool and PovRay is attached.
> As mentioned making these type of figures in my computer was 
> relatively slow.  What would be the best graphic card choice to 
> accelerate the display and manipulation of surfaces ? (I need to 
> replace the graphic card anyway).  A graphic card  with the  biggest 
> amount of graphics  memory   ?.  From your benchmark scores it seems 
> that memory is not the most important parameter, right ?.  Would the 
> Quadro FX 1700 be the best choice to display  many surfaces ?.
> Thanks again
> Hernando
> Tom Goddard wrote:
>> Hi Hernando,
>>  If you make multiscale surfaces totaling millions of triangles it 
>> can take a while to create the surfaces and it can be slow rotating 
>> them.  To see how many triangles you've got press the select "All" 
>> button on the multiscale dialog and then use Actions / Inspect, set 
>> the Inspect menu to Surface (probably already set that way) and the 
>> last line says "triangles 522960" when I try with the biological unit 
>> of 1uf2 and default surface resolution 8 Angstroms.  That rotates 
>> fast but took several seconds to create the surfaces on my Mac 
>> laptop.  If you get into millions of triangles it can be quite slow.  
>> To speed it up you can set the resolution to a larger value in the 
>> multiscale dialog, select all chains and press Resurface.  Or set the 
>> resolution value larger before you make the model in the first 
>> place.  With resolution 15 Angstroms I get 121000 triangles for 1uf2.
>>    Tom
>> hsosa at aecom.yu.edu wrote:
>>> Thanks Tom, Now I can change the colors on each chain.
>>> Another issue is that after displaying the surfaces of a big complex 
>>> (33 copies as BIOMT entries) chimera seems to freeze for very long 
>>> periods of time.  It may be related to not having enough computer 
>>> power (Dual Xeon 1.4 GHz,, 1GB RAM,  Windows XP,  Graphics ATI 
>>> Radeon 9550, Chimera 1.3) ?.
>>> Hernando
>>> Tom Goddard wrote:
>>>> Hi Hernando,
>>>>  Multiscale models assigns colors by chain by default.  All copies 
>>>> of chain B will initially get the same color.  All copies of chain 
>>>> C get another color.  If you wish to change those colors, select a 
>>>> chain surface, click the select Copies button, then use the Color 
>>>> button or the Actions / Color menu (in Chimera 1.3 or newer).
>>>>  Maybe the case you are thinking of is where chains B and C are 
>>>> really copies of the same protein with different conformations.  If 
>>>> you select the surface for chain B you can then select all chains 
>>>> which have identical amino acid sequences with shortcut xc, then 
>>>> color those.  Now if you have several chains that are the same 
>>>> protein but have different sequences (maybe some residues not 
>>>> observed in some of the copies) then they are not identical.  Then 
>>>> you'd just have to hand-select one copy of each chains known to be 
>>>> the same, press the Copies button and color.  This is the situation 
>>>> with rice dwarf virus 1uf2.  Might be nice to have the ability to 
>>>> extend selection to all chains that have the same chain name (given 
>>>> in PDB file though not currently used in Chimera).
>>>>    Tom
>>>> hsosa at aecom.yu.edu wrote:
>>>>> Is it possible to assign colors by chain  to the surfaces created 
>>>>> my the multiscale models tool ?
>>>>> Thanks
>>>>> Hernando
> ------------------------------------------------------------------------
> ------------------------------------------------------------------------
> _______________________________________________
> Chimera-users mailing list
> Chimera-users at cgl.ucsf.edu
> http://www.cgl.ucsf.edu/mailman/listinfo/chimera-users

More information about the Chimera-users mailing list