I have used many methods to do this over the years, but now I would
never consider using anything but the COOT interface.
The NCS operators that COOT uses are based on the SSM (secondary
structure matching) program written by Eugene Krissenel.
You read in your coordinates into coot and it recognises there are 2
copies of the molecule and determines the NCS operator.
When you ask for NCS maps it moves density B to density A and averages them
So you can look at density A, density B-overlapping-A, and Average.
In general I rebuild everything easy into the averaged map
Then output rebuiltA molecule and make a copy of A onto B and output that
Then I have to edit back together A rebuilt and A-onto-B rebuilt to make
a new pair. There is probably a better way but I cannot remember it!
Then re-refine, and do it all again until you can see that A and B
have real differences..
Be careful applying the NCS restraints in REFMAC - I usually chose tight
MC, medium sidechains.
Eleanor
Kolstoe S.E. wrote:
Dear ccp4bb,
I am about to start model building a fairly low resolution structure (3A) with
ten identical monomers in the asu, solved using MR. What I am thinking of doing
is to build one monomer using some form of averaged NCS map, replicate and
translate my one built monomer into the other nine positions, and then refine
in refmac with tight NCS restraints. But I have a couple questions:
1) Is this a sensible approach at this resolution?
2) Which programs would people suggest I use for the ncs calculations, map averaging and final
replication of my built monomer into the other nine positions? I see coot has an "ncs
maps" function, but I also see a number of usf and ccp4 programs that all seem to do similar
things. I am also confused as to the different formats that ncs operators seem to come in, and
which programs read which formats. Could anyone suggest a "road-map" of programs I could
use that would accept each others outputs and hence accomplish this feat?
Thanks,
Simon
