[phenixbb] Improving Ramachandran plot

[Oleg Sobolev]

Hi Reza,

I assume that you run Phenix via command line. We have a new enhanced
"ramachandran_plot_restraints" scope to define these (similar) parameters.
The previous one is there (for a short while) for backward compatibility
(so users can restore old GUI jobs). The new scope looks like:

    ramachandran_plot_restraints {
      enabled = True
      favored = *oldfield emsley emsley8k
      allowed = *oldfield emsley emsley8k
      outlier = *oldfield emsley emsley8k
      selection = None
      inject_emsley8k_into_oldfield_favored = True
      oldfield {
        weight = 0.
        weight_scale = 0.01
        distance_weight_min = 2.0
        distance_weight_max = 10.0
        plot_cutoff = 0.027
      }
      emsley {
        weight = 1.0
        scale_allowed = 1.0
      }
      emsley8k {
        weight_favored = 5.0
        weight_allowed = 10.0
        weight_outlier = 10.0
      }
    }
So please use this one. Notable change is that now you can restrain
favored/allowed/outliers separately (or not restrain one or another).
Please let me know if you need to know exactly how to convert old-style
parameters into new ones.

In case you are going to perform a similar study as you are mentioning, I
would like to suggest adding the Rama-Z metric to the set of validation
metrics. It was included particularly to judge the distribution of residues
on Ramachandran plot: https://doi.org/10.1016/j.str.2020.08.005

Let me know if you have any questions.

Best regards,
Oleg Sobolev.

On Sun, Sep 6, 2020 at 9:12 PM Reza Khayat <rkhayat at ccny.cuny.edu> wrote:

> Hi,
>
>
> I'm trying to follow the protocol described in White et al., Structural
> principles of SNARE complex recognition by the AAA+ protein NSF
>  e-life 2018
>
> https://elifesciences.org/articles/38888
>
>
> Here is the pertinent paragraph:
>
>
> "To improve the Ramachandran statistics and geometry of the models, a
> two-parameter grid search of real space refinements was performed in which
> 3–5 macrocycles of global minimization and local grid search were performed
> in the presence of secondary structure restraints. First, a grid refinement
> search was performed for both target functions with around 1000 refinements
> each. For the *emsley* target function, *rama_weight* and *scale_allowed* were
> varied from 0.01 to 300; for the *oldfield* target function, a grid of
> refinements was performed over *plot_cutoff* values from 0.01 to 1.0 and
> *weight_scale* values from 0.1 to 300. Results were judged empirically
> and based primarily on a balance between *CCmask* and a minimal fraction
> of residues flagged by the program CaBLAM (Richardson et al., 2018
> <https://elifesciences.org/articles/38888#bib37>) because focusing on the
> fraction of residues with favored *CCmask* and Ramachandran statistics
> alone often resulted in unrealistic models with serious problems (Figure
> 12 <https://elifesciences.org/articles/38888#fig12>, Table 2
> <https://elifesciences.org/articles/38888#table2>)."
>
>
> There are no differences in the refined structures regardless of what
> parameters I change. Attached is the input file I use for
> phenix.real_space_refine (1.18.2​). Altering the oldfield weight_scale (0.1
> to 300) and plot_cutoff (0.1 to 0.5) makes no difference. Any help is
> highly appreciated.
>
>
> Best wishes,
> Reza
>
>
> Reza Khayat, PhD
> Associate Professor
> City College of New York
> Department of Chemistry and Biochemistry
> New York, NY 10031
> _______________________________________________
> phenixbb mailing list
> phenixbb at phenix-online.org
> http://phenix-online.org/mailman/listinfo/phenixbb
> Unsubscribe: phenixbb-leave at phenix-online.org
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://phenix-online.org/pipermail/phenixbb/attachments/20200908/1840f1ca/attachment.htm>