On 29/05/12 19:49, Dayana Nisbar wrote:
I need help with finding water in my protein structure using Coot.
I tried to find water with peaks above 1.6 sigma for both the 2Fo-Fc and
mFo-DFc maps.
The result were 253 water molecules found for 2Fo-Fc map and 563 water
molecules found for mFo-DFc map.
My question is....
1) Why is it the number of water molecules are different?
Because the map you are searching are different. The peaks have
different positions, shape and distance to protein atoms.
2) Which map should we use to correctly place the water molecules?
I prefer the 2Fo-Fc style map. You could create a weighted average of
these maps if you wish.
3) What is the sigma cut-off for finding water?
How "deep" do you want to delve into you map? There is a lot of shape
at 0.1 rmsd - do you want to model that? (Most people do not). Most
people are happy to model density features around 1 rmsd for 2FoFc-style
map and around 3 rmsd for the difference map. (Not to say, of course,
that 1 rmsd has much meaning for 2FoFc-style map, but as a rule of
thumb, it is not bad at about 1.8-2.2A and solvent content 55+/-7% in my
experience).
4) How can I validate these water molecules?
Other than the water validation built into Coot of course, you could
also try Whatcheck.
HTH,
Paul.
