Hi Mark,

phenix.refine finds potential water peaks looking at mFo-DFc map (primary map). Then it sorts these peaks by the distances (water-water, water-other). Then it checks for hydrogen bonding. Finally, it verifies that the peaks are present at 2mFo-DFc map (secondary map) at given sigma. Finally finally, the water B-factors (and occupancies, if requested) are refined separately from refinement of other parameters in order to get a reasonable starting point for the next refinement macro-cycle. The procedure is smart enough about cases when your structure has explicit hydrogen atoms.

What is crucial is that this procedure is built into refinement, so the water molecules are updated (added / removed / refined) during refinement and NOT as a separate step.

There are various options to refine water with isotropic or anisotropic B-factors, refine water occupancies, etc.

We still need to do a better job on NOT picking water into ligand density. This is in our to-do list and will be done at some point.

There are a few slides in one of the most recent phenix.refine presentation that outline most of what I wrote above:

http://cci.lbl.gov/~afonine/aca2008_knoxville_neutron/phenix_refine_2008_may_31_ACA_final.pdf

Please let us know if you have any other questions,
Pavel.

On 7/31/2008 2:03 PM, Mark Collins wrote:

Hi All

I was just going through the .help section and website documentation

for ordered_solvent. And wonder if somebody could enlighten me as too

the meanings of;

ordered_solvent {

primary_map_type = mFobs-DFmodel

secondary_map_type = 2mFobs-DFmodel

h_bond_min_mac = 1.8

h_bond_min_sol = 1.8

h_bond_max = 3.2

}

peak_search {

map_next_to_model {

use_hydrogens = False

}

max_number_of_peaks = None

peak_search {

peak_search_level = 1

max_peaks = 0

interpolate = True

min_distance_sym_equiv = None

general_positions_only = False

min_cross_distance = 1.8

}

What is the difference between the primary and secondary map? Is there

away to pick sites in one and remove in another? Eg. pick sites at 3 sigma

in FoFc and remove if 1.2 sigma in 2FoFc.

I thought h-bonding was generously about 2.4-3.6A so why the 1.8 and 3.2

defaults? Also if use_hydrogens = True, does the refinement use h_bond

parameters instead of model_peak_dist and peak_peak_dist parameters, or

both? Also what do all of the peak_search parameters do?

And finally does anyone have any recommendations for progessively changing

the ordered_solvent parameters? In CNS we used to gradually lower the

2FoFc peak picking (and deleting) and increase the bfactor_max for

deleting?

Thanks Mark
_______________________________________________
phenixbb mailing list
phenixbb@phenix-online.org
http://www.phenix-online.org/mailman/listinfo/phenixbb
  