"cartoon style nucleic" changes protein coil cartoon if model has nucleic residue(s) ligand

[Elaine Meng]

at least that's my theory of when it happens.

open 5mv5
cartoon style nucleic thickness 1.6

... makes all the protein coil fat. You might say "why would you use cartoon style nucleic" on this structure when it only has one nucleic residue as a ligand, but it is because this command appears in the cylinders/stubs preset. So if I choose that preset from the menu, all the coil suddenly and mysteriously becomes fat, even though it is NOT classified as nucleic but as coil.

I think this happens when the structure has a nucleotide ligand but no nucleotide chains. Another example besides 5mv5 is 121p.

[Conrad Huang]

Fixed in e8f65f3c2.

Code testing for coil is "not helix and not sheet and not nucleic". The single nucleotide wasn't classified as "nucleic", so it qualified as coil. Since a style applies at the model level, all coils got the new style. Fix is to change "not nucleic" to "must be protein".

[Elaine Meng]

I don’t really understand your explanation of the problem, since using those built-in terms in the command line was already working (“sel nucleic” and “sel coil” both select what they are supposed to select in 5mv5 and 121p).  However, I guess that’s that not important as long as it’s fixed!

[Conrad Huang]

For "cartoon style" to determine whether a style should apply, it checks which classes are included in the selected atoms.  Helices, sheets and nucleic acids all have explicit flags associated with them in the atom data structure; unfortunately, coils do not.  That is, there is no flag that I can check when given an atom to see whether it is a coil or not; this is independent of the coil selector which has its own code for determining what is or is not a coil atom.  Before the bug fix, the "cartoon style" coil test is "not helix and not sheet and not nucleic".  A single nucleotide is not marked as nucleic (not my code), so the selected atoms were considered "coil" and the coil style got updated.  The new coil test is "not helix and not sheet and protein".  Since the selected atoms are (also) not marked as protein (again, not my code), the selected atoms are not considered "coil" and the coil style remains unchanged.  I think the new test is more correct (if the residue is not part of a protein, it cannot be part of a coil), so it's the right fix.

Conrad

June 10, 2019 3:23 PM, "ChimeraX" <ChimeraX-bugs-admin@cgl.ucsf.edu> wrote: