Hi Ashu,

2mFobs-DFmodel synthesis that phenix.refine outputs is not sharpened. High overall atomic B (maximal common B-factor that can be subtracted from all individual atomic B-factors that will still leave them non-negative) is what makes the map blurred.
If we subtract this common B from individual atomic B-factors, recompute Fcalc and then recompute Fmodel, which is

Fmodel = k_total * (Fcalc + Fbulk)

and then compute the new synthesis with amplitudes (2mFobs-DFmodel)/k_total it will be sharpened (and likely similar to what you used to build the model). Note, noise will be upscaled too (this is the main reason why it is not done by default!).

We are thinking of making this map available as part of phenix.refine output.

For now try this:

phenix.pdbtools shift_b_iso=SOME_NEGATIVE_VALUE model.pdb

where SOME_NEGATIVE_VALUE is negative B-factor, like -10 or so, chosen such that subtracting it from atomic B will not leave them negative.

Then compute maps using modified PDB file:

phenix.maps model.pdb data.mtz

Pavel

On 9/8/14 7:57 AM, Ashu S wrote:

Dear all,

I am working on a 3.6A Se SAD data. Autosol followed by autobuild gave a decent map (as expected at this resolution) with a moderate trace. After building poly ala trace and assigning few resides in and around Se positions when I refine the model, all the sharp features of the electron density map are lost. For example, a nice looking helical density becomes a tube. I presume that this happens because wilson B is high (~110 A2) and the map requires modification, sharpening etc. Is there a way to get the map with similar sharp features after model building and refinement as auto build gives? How could I follow exactly the same steps of density modification etc which auto build does?

Thanks,

Ashu
_______________________________________________
phenixbb mailing list
[email protected]
http://phenix-online.org/mailman/listinfo/phenixbb