Optimize map grid size based on fit of atomic model

[olibclarke@…]

Begin forwarded message:

From: Oliver Clarke
Subject: ChimeraX sugestion - auto refine pixel size of map
Date: February 16, 2016 at 7:29:33 AM PST
To: Tom Goddard

Hi Tom,

Another suggestion:

One of the main annoyances when dealing with EM maps for model building is that the magnification (and hence pixel size) is often not accurate - the nominal pixel sizes on many EM maps are wrong, which substantially affects model fitting at high res.

The way I have been correcting for this in Chimera is to take a known crystal structure of one of the domains of the target and rigid body fit it in the appropriate region of density across a range of pixel sizes centered around the nominal pixel size, recording the correlation in each case. Usually, there will be a distinct maximum in this distribution, which I take as the "true" pixel size. I wonder whether an automated version of such a procedure, perhaps using ideal secondary structural elements rather than crystal structures (which may not be available in all cases) might be worth incorporating into ChimeraX?

Cheers,
Oli.