color by atom/res attribute should automatically also color molecular surface caps

[Elaine Meng]

...if target includes surface, which it does by default. Presumably this would require refining the surface-cap triangulation as needed to fix related ticket #2242. Currently the cap is ignored and stays a solid color.

[Tom Goddard]

There are some difficulties with this. Color by attribute colors only molecular surfaces which have an association of each surface point to some atom. Clip caps don't have an association to atoms. But color zone could be used to color the caps. It is not clear what the range of the color zone would be, but maybe a fixed default like 2 Angstroms would be ok. Another problem is that if the surface is colored when clipping is off, and then clipping is turned on, current code has no memory that a color by attribute was done in the past, so it would not know how to color the cap. Another problem is that session saving does not have a way to remember how the cap was colored, because the cap is not saved in sessions. To solve these problems some new color by attribute code that remembers how a surface was colored and can save its state in sessions is needed.

The same issues apply with mlp coloring, although it is slightly different since it is basically a coloring by map value using a computed mlp map (usually not shown).

Even the case of "color sample" that colors a surface by map values has trouble with caps now. It works on caps, but session saving does not restore the cap color because surface caps are not save in sessions.

There are a lot of problems to solve with coloring caps and they probably will not be fixed anytime soon.

[Elaine Meng]

If “color zone” is used to color caps on molecular surfaces to match the underlying atoms, the range distance would need to be at least the diameter of the probe sphere used for surface calculation because atoms could be that far apart without inducing a bubble within the surface that wouldn’t be capped.  In other words, points on the surface cap can be up to that distance from any atomic center.  I also tested this theory in practice and results agreed: with cutoff 2-2.6 I could see a giraffe-like pattern where the interstices still had the original cap color rather than any atom color.

For now, I’ll explain this in the documentation for any advanced users who actually try to color the caps by attribute.  From the Chimera images I’ve seen in papers, however, most people just give up and leave caps uncolored.  I’ll also mention color zone in the color-by-attribute section, in case anybody actually reads it.