How to use a colour-valued attribute?
Hi, This feels like a basic question but it has stumped me for a day now so here we go... I'm exploring ways of programmatically setting colours to find the most efficient method. This is for the 'custom coloured' case (residue-based colour schemes are not a problem). The simple version is to send as many 'color' commands as there are colours in my (Jalview) sequence: color #ffafaf #0:1.B|#0:5.B|#0:7-8.B|....;color #ffff00 #0:39.B|#0:44.B|#0:47.B|#0:77.B; This works, but doesn't seem to scale up well for large/many structures - I start to see Chimera using 100% CPU for minutes at a time. I'm trying instead to see whether setting a colour-value attribute is faster. I can set the attributes ok with commands like setattr r color green #0:1-84.B etc and confirm they are set with list residues spec #0:1-84.B attribute color but I can't figure out how to get Chimera to use the attribute value as the display colour. 'Render by Attribute' only seems to consider numeric (range-valued) attributes, not colour-valued. I thought perhaps 'my' colour was just hidden in the colour hierarchy, but removing other colours doesn't seem to help. What am I missing? Thanks, Mungo Carstairs Division of Computational Biology University of Dundee http://www.jalview.org The University of Dundee is a registered Scottish Charity, No: SC015096
You used two different ways to specify atoms for the "color" and "setattr" commands. Have you tried comparing the speed for color #ffafaf #0:1-84.B as opposed to color #ffafaf #0:1.B|#0:2.B|... to see if there are performance differences? The former specification requires testing each residue against a single sequence range, whereas the latter requires testing against many residue sequence numbers. So I would expect the former to work faster than the latter. Conrad On 12/2/2014 2:10 AM, Mungo Carstairs wrote:
Hi,
This feels like a basic question but it has stumped me for a day now so here we go...
I'm exploring ways of programmatically setting colours to find the most efficient method.
This is for the 'custom coloured' case (residue-based colour schemes are not a problem).
The simple version is to send as many 'color' commands as there are colours in my (Jalview) sequence:
color #ffafaf #0:1.B|#0:5.B|#0:7-8.B|....;color #ffff00 #0:39.B|#0:44.B|#0:47.B|#0:77.B;
This works, but doesn't seem to scale up well for large/many structures - I start to see Chimera using 100% CPU for minutes at a time.
I'm trying instead to see whether setting a colour-value attribute is faster.
I can set the attributes ok with commands like
setattr r color green #0:1-84.B
etc
and confirm they are set with
list residues spec #0:1-84.B attribute color
but I can't figure out how to get Chimera to use the attribute value as the display colour.
'Render by Attribute' only seems to consider numeric (range-valued) attributes, not colour-valued.
I thought perhaps 'my' colour was just hidden in the colour hierarchy, but removing other colours doesn't seem to help.
What am I missing?
Thanks,
Mungo Carstairs
Division of Computational Biology University of Dundee http://www.jalview.org
The University of Dundee is a registered Scottish Charity, No: SC015096
_______________________________________________ Chimera-users mailing list Chimera-users@cgl.ucsf.edu http://plato.cgl.ucsf.edu/mailman/listinfo/chimera-users
To add a bit to Conrad’s suggestion, you can specify non-contiguous residues in one command, like color red #0:1,4-6,8.A and that may be faster than the union “|” operator. Tom
On Dec 2, 2014, at 10:17 AM, Conrad Huang wrote:
You used two different ways to specify atoms for the "color" and "setattr" commands. Have you tried comparing the speed for color #ffafaf #0:1-84.B as opposed to color #ffafaf #0:1.B|#0:2.B|... to see if there are performance differences? The former specification requires testing each residue against a single sequence range, whereas the latter requires testing against many residue sequence numbers. So I would expect the former to work faster than the latter.
Conrad
On 12/2/2014 2:10 AM, Mungo Carstairs wrote:
Hi,
This feels like a basic question but it has stumped me for a day now so here we go...
I'm exploring ways of programmatically setting colours to find the most efficient method.
This is for the 'custom coloured' case (residue-based colour schemes are not a problem).
The simple version is to send as many 'color' commands as there are colours in my (Jalview) sequence:
color #ffafaf #0:1.B|#0:5.B|#0:7-8.B|....;color #ffff00 #0:39.B|#0:44.B|#0:47.B|#0:77.B;
This works, but doesn't seem to scale up well for large/many structures - I start to see Chimera using 100% CPU for minutes at a time.
I'm trying instead to see whether setting a colour-value attribute is faster.
I can set the attributes ok with commands like
setattr r color green #0:1-84.B
etc
and confirm they are set with
list residues spec #0:1-84.B attribute color
but I can't figure out how to get Chimera to use the attribute value as the display colour.
'Render by Attribute' only seems to consider numeric (range-valued) attributes, not colour-valued.
I thought perhaps 'my' colour was just hidden in the colour hierarchy, but removing other colours doesn't seem to help.
What am I missing?
Thanks,
Mungo Carstairs
Division of Computational Biology University of Dundee http://www.jalview.org
The University of Dundee is a registered Scottish Charity, No: SC015096
_______________________________________________ Chimera-users mailing list Chimera-users@cgl.ucsf.edu http://plato.cgl.ucsf.edu/mailman/listinfo/chimera-users
_______________________________________________ Chimera-users mailing list Chimera-users@cgl.ucsf.edu http://plato.cgl.ucsf.edu/mailman/listinfo/chimera-users
Another clarification, if you want only chain A, it would need to be: color red #0:1.A,4-6.A,8.A (- or -) color red #01.a,4-6.a,8.a … chain case doesn't matter unless there are both upper- and lowercase chain IDs present. Elaine On Dec 2, 2014, at 11:13 AM, Tom Goddard <goddard@sonic.net> wrote:
To add a bit to Conrad’s suggestion, you can specify non-contiguous residues in one command, like
color red #0:1,4-6,8.A
and that may be faster than the union “|” operator.
Tom
participants (4)
-
Conrad Huang -
Elaine Meng -
Mungo Carstairs -
Tom Goddard