[phenixbb] FAQ high res refinement

Eike Schulz eike.schulz at embl.de
Thu Feb 27 10:39:25 PST 2014


> 
> Dear all, 
> 
> I am currently refining a rather well resolved protein at 0.97Å. However, even
> after many cycles of refinement the R-factors are stuck at around Rwork/Rfree
> 14.5/16.5. Considering number of aa, Rmerge (3.5%) and resolution this seems
> to be too high. I assume 10 ­ 25 macrocycles should be enough for effective
> weight refinement in phenix, however no improvement of the R-factors.
> 
> A quick run with default settings in competitor software 'R' yields >2% better
> R-factors. 
> 
> I would appreciate suggestion how to improve my refinement protocol, values
> differing from default are listed below.
> 
> Thank you very much in advance
> 
> Eike
> 
> 
> 
> 
>     
> 
> refine {
>     adp {
>       individual {
>         isotropic = element H
>         anisotropic = not element H
>       }
>       group_adp_refinement_mode = one_adp_group_per_residue \
>                                   *two_adp_groups_per_residue group_selection
>     }
>   }
>   main {
>     apply_overall_isotropic_scale_to_adp = False
>     nqh_flips = False
>     ordered_solvent = True
>     place_ions = True
>     number_of_macro_cycles = 10
>     hydrogen_bonds = True
>     use_convergence_test = True
>     target = auto *ml mlhl ml_sad ls
>     random_seed = 3326496
>     wavelength = 0.885611
>     nproc = 4
>   }
>   hydrogens {
>     refine = individual *riding Auto
>   }
>   tls {
>     find_automatically = False
>     one_residue_one_group = False
>   }
>     ordered_solvent {
>     mode = second_half filter_only every_macro_cycle \
>            *every_macro_cycle_after_first
>     h_bond_min_mac = 1
>     h_bond_min_sol = 1
>     h_bond_max = 6
>     refine_occupancies = True
>     new_solvent = isotropic *anisotropic
>     b_iso_min = 0
>   }
>   peak_search {
>     max_number_of_peaks = 1303
>   }
>   target_weights {
>     optimize_xyz_weight = True
>     optimize_adp_weight = True
>     wxc_scale = 0.2
>     wxu_scale = 0.2


-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://phenix-online.org/pipermail/phenixbb/attachments/20140227/390a4b0e/attachment.htm>


More information about the phenixbb mailing list