[phenixbb] Fitting ligand with DM map?

Dale Tronrud detBB at daletronrud.com
Sat Jun 3 12:18:23 PDT 2017

   Density modification tries to improve the map by introduction
restrictions on the flattness of the bulk solvent region and the details
of the shape of the density in the protein region (histogram matching).
Both kinds of information are included in normal crystallographic

   After one has constructed a model that covers much of the structure
one goes into refinement and examines the 2DFo-mFc map to see if the
previously uninterruptible bits of the map have become better.  This is
a much more powerful technique than DM alone.

   Personally, I would refine the model without any ligand built to get
the best possible map before making any interpretation of the bound
compound. I would also do several rounds of rebuilding of the protein
and add the obvious water.  Not only will the result be the best
possible (unbiased) image of your ligand the result will also be an
"omit" map that you can use in your publication to justify your final

Dale Tronrud

On 6/3/2017 11:33 AM, Sam Tang wrote:
> Dear community
> I guess it may not be a standard practice but I wonder if anyone has
> tried to use a density modified map for ligand building.
> My situation is that I am trying to fit a single strand DNA ligand into
> an enzyme model. The dataset is at 3 Ang, P1 space, with overall
> completeness of only about 88%. MR was done by the apo-enzyme and the
> density for the first two-third of the ligand is very good and could
> easily be modelled. Towards the last one-third the density becomes much
> weaker so nucleotides cannot be easily put. This is the only dataset so
> far, so multi-crystal averaging / map averaging is not possible at this
> stage. NCS averaging is also not possible. We did try to sharpen map
> during refinement.
> I understand that DM is usually used to improve phase at the initial
> stage esp. with experimental phasing. With MR would you think density
> modification could produce a better map to aid the modelling of the last
> one-third ligand? Can I use the denmod.mtz directly for manual model
> building?
> Any input would be very much appreciated.
> Best
> Sam
> Biochemistry Programme, CUHK
> _______________________________________________
> phenixbb mailing list
> phenixbb at phenix-online.org
> http://phenix-online.org/mailman/listinfo/phenixbb
> Unsubscribe: phenixbb-leave at phenix-online.org

More information about the phenixbb mailing list