Hi, I can't see why doing both: defining custom bonds between N-O pairs that stabilize the helical region and using Ramachandran plot restraints would not work. Could you please try it? If it still doesn't work, then please send me the data and model, indicate the region that you want to be helical, and tell which command you used; then I will have a look. Alternatively, as Nat suggested, you can use reference model restraints. Just create a PDB file containing idealized (the way you want it) helical region and supply it to phenix.refine for reference model restraints. Pavel. On 7/29/11 10:35 AM, Nathaniel Echols wrote:

On Fri, Jul 29, 2011 at 10:25 AM, Daqi Tu wrote:

...

I was trying to restrain a relatively mobile region of my structure to be helical using Phenix secondary structure restraints. The result is very bad. It forces the N-O distance to be about 3A, but the phi/psi angles are all ugly. What is the way in Phenix I can apply phi/psi restraints just like in the old CNS, which had always worked nicely? At present, I think your only options are

a) use the reference model feature (preferred, if you have a higher-resolution structure - this can also be done with a model optimized by another program, e.g. Rosetta) b) Ramachandran restraints (potentially dangerous, should be used with extreme caution) c) enable use of the monomer library phi/psi restraints (discard_psi_phi=False on the command line, or Settings->Advanced->PDB file interpretation in the GUI)

We've discussed restraining phi/psi in secondary structure instead of the hydrogen bonds, but nothing has been implemented yet. If you're willing to share your model and data with us for testing purposes, it might speed things up. (It sounds like maybe the existing secondary structure restraints aren't working properly either - although our tests have indicated that they can't do much with a poor input model.)

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

I downloaded the latest build (dev-833), but could not see a Ramachandran restraints option under Phenix.refine GUI /all parameters/atom selections. Am I missing something? Thanks. Daqi On 7/29/2011 1:53 PM, Nathaniel Echols wrote:

On Fri, Jul 29, 2011 at 10:45 AM, Pavel Afonine wrote:

...

I can't see why doing both: defining custom bonds between N-O pairs that stabilize the helical region and using Ramachandran plot restraints would not work. By default Ramachandran restraints are applied to all residues - this is very risky, especially when the goal is to keep secondary structure from unraveling. It would be prudent to define the parameter rama_selection (in the GUI, "Atom selection for Ramachandran restraints") to include only the helical regions.

What is the resolution of this structure?

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

Yes, applying Ramachandran restraints alone is able to maintain helical conformation. Cheers, Daqi On 7/29/2011 1:45 PM, Pavel Afonine wrote:

Hi,

I can't see why doing both: defining custom bonds between N-O pairs that stabilize the helical region and using Ramachandran plot restraints would not work. Could you please try it? If it still doesn't work, then please send me the data and model, indicate the region that you want to be helical, and tell which command you used; then I will have a look.

Alternatively, as Nat suggested, you can use reference model restraints. Just create a PDB file containing idealized (the way you want it) helical region and supply it to phenix.refine for reference model restraints.

Pavel.

On 7/29/11 10:35 AM, Nathaniel Echols wrote:

...

On Fri, Jul 29, 2011 at 10:25 AM, Daqi Tu wrote:

...

I was trying to restrain a relatively mobile region of my structure to be helical using Phenix secondary structure restraints. The result is very bad. It forces the N-O distance to be about 3A, but the phi/psi angles are all ugly. What is the way in Phenix I can apply phi/psi restraints just like in the old CNS, which had always worked nicely? At present, I think your only options are

a) use the reference model feature (preferred, if you have a higher-resolution structure - this can also be done with a model optimized by another program, e.g. Rosetta) b) Ramachandran restraints (potentially dangerous, should be used with extreme caution) c) enable use of the monomer library phi/psi restraints (discard_psi_phi=False on the command line, or Settings->Advanced->PDB file interpretation in the GUI)

We've discussed restraining phi/psi in secondary structure instead of the hydrogen bonds, but nothing has been implemented yet. If you're willing to share your model and data with us for testing purposes, it might speed things up. (It sounds like maybe the existing secondary structure restraints aren't working properly either - although our tests have indicated that they can't do much with a poor input model.)

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

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

Yes, I did only apply to that helical region. The point is as long as there is reasonable density there the Ramachandran restraints are able to hold the secondary structure conformation. Daqi On 8/3/2011 2:15 AM, Pavel Afonine wrote:

Hi Daqi,

oh good! Then I guess you simply need to apply Ramachandran restraints to that helical region only (using rama_selection keyword).

Pavel.

On 8/2/11 8:28 AM, Daqi Tu wrote:

...

Yes, applying Ramachandran restraints alone is able to maintain helical conformation.

Cheers, Daqi

On 7/29/2011 1:45 PM, Pavel Afonine wrote:

...

Hi,

I can't see why doing both: defining custom bonds between N-O pairs that stabilize the helical region and using Ramachandran plot restraints would not work. Could you please try it? If it still doesn't work, then please send me the data and model, indicate the region that you want to be helical, and tell which command you used; then I will have a look.

Alternatively, as Nat suggested, you can use reference model restraints. Just create a PDB file containing idealized (the way you want it) helical region and supply it to phenix.refine for reference model restraints.

Pavel.

On 7/29/11 10:35 AM, Nathaniel Echols wrote:

...

On Fri, Jul 29, 2011 at 10:25 AM, Daqi Tu wrote:

...

I was trying to restrain a relatively mobile region of my structure to be helical using Phenix secondary structure restraints. The result is very bad. It forces the N-O distance to be about 3A, but the phi/psi angles are all ugly. What is the way in Phenix I can apply phi/psi restraints just like in the old CNS, which had always worked nicely? At present, I think your only options are

a) use the reference model feature (preferred, if you have a higher-resolution structure - this can also be done with a model optimized by another program, e.g. Rosetta) b) Ramachandran restraints (potentially dangerous, should be used with extreme caution) c) enable use of the monomer library phi/psi restraints (discard_psi_phi=False on the command line, or Settings->Advanced->PDB file interpretation in the GUI)

We've discussed restraining phi/psi in secondary structure instead of the hydrogen bonds, but nothing has been implemented yet. If you're willing to share your model and data with us for testing purposes, it might speed things up. (It sounds like maybe the existing secondary structure restraints aren't working properly either - although our tests have indicated that they can't do much with a poor input model.)

-Nat

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