armandb
Hello everybody,
I have an initial model with fractional coordinates (not rigid body) of my structure.
I would like to set constant values for some angles during refinement.
As I know one of the way is macro:
[m]Angle_Restrain(sites, t, t_calc, tol, wscale)[/m]
eg.:
[m]Angle_Restrain(O1 P1 O1 8, 112, 112.92311, 0, 0.001)[/m]
where [m]t=112[/m] is expected angle,
[m]t_calc =112.92311[/m] is calculated angle from strucutre.
[m]tol=0[/m]
[m]wscale=0.001[/m]
Some time ago Alan explained me how it works personally. I red his explanantions
and analyzed examples several times. And I thought that it is simple, but in practice
I could not do, what I want.
How I should change [m]tol[/m] and [m]wscale[/m] to force the desired angle [m]t_calc=t[/m]?
I my case: [m]Angle_Restrain(O1 S1 O2, 111.06653, 109.25200, 0, 0.001).[/m]
Mayby you know some another way? Have you got idea how to do it?
I spend several days to prepare rigid body of this part of structure, although I know initial model in fractional coordinates.
Thank you for your advices and your help.
johnsoevans
Armand,
What you're trying to do is a soft restraint not a constraint. i.e. you're introducing a piece of information that's treated in the least squares in a similar way to the experimental data. Thus it may not be obeyed precisely during the refinement (e.g. if the data pulls the structural model in a different way). You can increase the weighting of restraints relative to the data using the command:
penalties_weighting_K1 1
If you increase the value from 1 you'll find restraints are obeyed more closely. Try e.g. a value of 10. Watch out for how this effects convergence.
The wscale weights restraints relative to each other. If you restrain some distances and some angles you might like the distances to be more heavily weighted than an angle. e.g. you might like a distance to be within 0.01 A of ideal value but a bond angle only within 5 degrees. You'd then have a higher wscale on the distance restraint than the angle.
Have a look in topas.log at the end of refinement and you'll see the types of penalty equation topas applies.
John
armandb
John,
Thanks for the explanations and advices.
I see that it is a wrong way to constrain.
I forgot to add in previous post, that my idea was to
divide penalties between bonds, angles and others.
In this way I thought about the way like this:
Angle_Restrain, with [m]tol=10[/m] and very high [m]wscale[/m] (for constrains)
Distance_Restrain, with [m]tol=0[/m] and low [m]wscale[/m] (for refinement).
Now I'm sure that it is a wrong way.
Could you advise me another idea do constains?
Is there only one way... rigid body?
johnsoevans
Armand,
You ought to be able to do what you need with restraints (but the constraints of a rigid body will always be mathematically the most precise).
I don't know the system/data quality etc.
My first guess would be to put all tols at 0.0 to start. Try penalties_weighting_K1 5. Try using a distance weighting of 1.0 and an angle of 0.001. You can then see how the various restraints are contributing by looking at the "ideal" and "refined" values. If you look in topas.log you can see how to modify macros or write out penalty equations to help you monitor penalty values more quantitatively.
John
marishvcard
Your posting list is Excellent! Thanks for sharing with the community.
Can i use them all at once in a search? But i know maximam search result finding Error):
nprietveld
Dear All,
I can see that changing the weighting of a restraint during the refinement is quite important; if I want to decrease the influence of weighting in the refinement it is anyway difficult for me to decide if I have to change the weighting value from say 5 to 4 or to 3.45 or whatever else...Namely, is there any way to evaluate the influence of "weighting" restraints? Is there any way to express it analitically?
Another interesting thing would be to allow the weighting value of a restraint to decrease until Rwp_dash decreases, stopping the process when a minimum is reached. Maybe there is a way to do this?
Thanks for your attention.
Natale