Notes from Helen S on EM

1
Notes from Helen S on EM models Where to get
initial models Homologous structures, homology
models, deformation of related structures by
normal mode analysis constrained by EM map, or
deformed on the basis of other biochemical or
biophysical information (see pneumolysin example
on next slide) How to assess quality of fitted
models Normalized correlation coefficient is
not an absolute measure e.g. depends on whether
there are regions of density not filled by model,
or if they are masked out. Examine fitted
structure for clashes or gaps? Test fit to
opposite hand, if unknown? Validation ?? How
good does it look? Rerun automated fitting with
random displacement of starting position? Compare
to other information.. Archiving As previously
discussed database front end should make it
look as if map and fit are stored together, even
if they are in different places. The position and
orientation must be correctly stored so that the
fit always ends up in the correct place in the
map. There should be an easy way to generate all
the symmetry related molecules. There should be a
category that includes fits to negative stain EM
maps.
2
Example of a fit which is clearly not correct in
detail but is a useful first approximation to a
very large conformational change in a
pore-forming toxin
EM maps and manual fits of prepore and pore forms
of pneumolysin in phospholipid vesicles Automated
fitting was not possible.
Crystal structure of soluble monomer -
perfringolysin
Membrane insertion from fluorescent probe analysis
1
1
1
3
2
2
2
3
3
4
4
4
Fits were deposited as a-carbons
Tilley et al, Cell 2005