Identify regions of unoccupied density

[tic20@…]

Begin forwarded message:

From: Tristan Croll
Subject: Re: ISOLDE and ChimeraX plan for the future
Date: May 4, 2020 at 3:24:17 PM PDT
To: Tom Goddard

Hi Tom,

...

Another would be an unfilled-density tool for cryo-EM. Even something as simple as masking out everything within ~1.5A of the existing atoms is a great start... better, of course, would be to generate a proper simulated density map based on the model and then subtract that, but that's a much more challenging problem that would first require me to have some level of B-factor refinement in place.

[Tom Goddard]

At low resolutions (3A) won't there always be lots of unoccupied density (loops, mobile stuff near periphery)? Seems like some idea of how to identify unoccupied density that should be occupied is needed. Some testing with real structures would allow assessing how hard this is.

[Tristan Croll]

Yes, a tool to identify and prioritise the biggest unfilled blobs would 
certainly be nice - but simply being able to show everything unfilled in 
a different colour (similar to the X-ray difference map) is already a 
big help. The big cryoEM mitochondrial complex I've been helping with is 
a good example - the maps are now a little better than 3A resolution, 
just good enough to start showing waters in the best-ordered regions. 
The other day I tried manually making a residual map by making a mask:

m = session.isolde.selected_model
atoms = m.atoms[m.atoms.element_names != 'H']
from chimerax.clipper.maps.mask_handler import VolumeMask
mask = VolumeMask(session, atoms.coords, 0.5, 1.2, 10)
session.models.add([mask])

... and then combining that with the map using the existing `volume 
multiply` and `volume subtract` commands. The result was actually really 
useful - showing the residual in a different colour and being able to 
independently adjust its contour really helped with picking out these 
sites. The ideal, of course, would be for this too to be dynamic with 
the residual map updating as the model changes.

On 2020-05-05 01:02, ChimeraX wrote: