Elise - after looking into a related problem for some time, I've come
to the conclusion it's a bit harder than it seems. Two obstacles:
(1) there is no good coordinate system for partitioning the overall
solvent volume (fraction of unit cell not occupied by protein density)
into suitable "neighborhoods"
(2) experimental support for water placement in deposited structures
varies from sound to wishful thinking - with no single useful
credibility metric ( a helpful review :
http://www.ncbi.nlm.nih.gov/pubmed/8081736
)
The path I've gone down for (1) is to : (a) compute the distance field
from the molecular surface and skeletonize it. This will give you the
outer boundary of surrounding solvent (eg if you blew up the molecular
surfaces like balloons, the surface defined by where the balloons
touch); (b) carve up the resulting volume as appropriate, eg radial
shells from the molecular surface for outer regions modeled as bulk
solvent, atom-specific expansion of the molecular surface for regions
occupied by modeled waters/ions.
This is quite a bit of work (or so it seemed) but it enables you to
segment the overall solvent volume into regions of interest and then
calculate density of ordered waters/ions across structures (bearing in
mind that much of modeled solvent may not be real). For sorting out
non-bonding contacts there is a very nice and fast class, MAtomNonBond,
in Kevin Cowtan's clipper library. For the distance field and
skeletonization/medial-axis computation you might want to look at
fast-marching implementations.
Good luck!
Alastair
---
On 10/29/2013 04:43 PM, Elise B wrote:
Hello, I am working on a project with several (separate) structures of
the same protein. I would like to be able to compare the solvent
molecules between the structures, and it would be best if the waters
that exist in roughly the same position in each PDB share the same
residue number. Basically, I want to compare solvent molecule
coordinates and assign similar locations the same name in each
structure. What would be the best strategy for re-numbering the water
molecules such that those with similar coordinates in all the structures
receive the same residue number? I'd appreciate any suggestions. Elise
Blankenship
