Changing NCS groups

older
Xtriage systematic results is not...

Marie Elizabeth Fraser

22 Jan 2016 22 Jan '16

10:54 p.m.

Hello, I would like to edit the NCS information for a model with two chains. I've tried running find_ncs and editing the phil file, but I get an error when I try to use my edited file in the next cycle of refinement. I've tried using the utility to detect NCS groups, then edited in the little windows - also no success. How are other people defining NCS when they aren't content with the equivalences the program finds? Thank you for your insight, Marie

Attachments:

attachment.html (text/html — 988 bytes)

Show replies by date

Oleg Sobolev

22 Jan 22 Jan

11:35 p.m.

Dear Marie, I would suggest to upgrade to the latest available nightly build http://www.phenix-online.org/download/nightly_builds.cgi because the NCS code is being under active development. If this doesn't solve your problem, please be more specific about what selections are provided by automatic detection utilities and why you are not happy with them (what you are trying to input). Also specific error message from Phenix could be very helpful. Best regards, Oleg Sobolev. On Fri, Jan 22, 2016 at 2:54 PM, Marie Elizabeth Fraser wrote:

...

Hello, I would like to edit the NCS information for a model with two chains. I've tried running find_ncs and editing the phil file, but I get an error when I try to use my edited file in the next cycle of refinement. I've tried using the utility to detect NCS groups, then edited in the little windows - also no success. How are other people defining NCS when they aren't content with the equivalences the program finds? Thank you for your insight, Marie

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

Edward A. Berry

23 Jan 23 Jan

6:13 a.m.

On 01/22/2016 05:54 PM, Marie Elizabeth Fraser wrote:

...

Hello, I would like to edit the NCS information for a model with two chains. I've tried running find_ncs and editing the phil file, but I get an error when I try to use my edited file in the next cycle of refinement. I've tried using the utility to detect NCS groups, then edited in the little windows - also no success. How are other people defining NCS when they aren't content with the equivalences the program finds? Thank you for your insight, Marie

If you really do want to tweak the NCS restraints manually, and if you want cartesian restraints, In an ancient version of phenix.refine (dev1745), it used to be possible to specify in the .eff file with the below syntax (duplicate the blank group which is already present). I think if you first use auto detection (and cartesion ncs) these lines will be generated in the output .eff and .def file. You can then edit them in the .def file and run, perhaps going back to the original pdb for input in case inappropriate restraints destroyed the output of that first run. (And turn off find_automatically) You can look at the .geo file from the first run to find the biggest violations in your orig pdb, then look at the input and output files in the density, and decide in each case if it is a true violation that should be released from NCS. You can duplicate the restraint_group section as many times as you like in case you have multiple chains in the monomer or chains that don't behave as a single rigid body. refinement { ncs { restraint_group { reference = None selection = None coordinate_sigma = 0.05 b_factor_weight = 10 } restraint_group { reference = chain A and ((RESID 12:255 and not (resid 205 or resid \ 184)) or RESID 1001 ) selection = chain N and ((RESID 12:255 and not (resid 205 or resid \ 184)) or RESID 1001 ) coordinate_sigma = 0.05 b_factor_weight = 0 } restraint_group { reference = chain A and ((RESID 256:365 and not (resid 332 or resid \ 9225)) or ( RESID 1002) ) selection = chain N and ((RESID 256:365 and not (resid 332 or resid \ 9225)) or ( RESID 1002) ) coordinate_sigma = 0.05 b_factor_weight = 0 }

Pavel Afonine

8:05 a.m.

You can have NCS related groups defined automatically or manually for restraints (torsion or Cartesian) or constraints, in current version. Pavel On 1/22/16 22:13, Edward A. Berry wrote:

...

On 01/22/2016 05:54 PM, Marie Elizabeth Fraser wrote:

...
Hello, I would like to edit the NCS information for a model with two chains. I've tried running find_ncs and editing the phil file, but I get an error when I try to use my edited file in the next cycle of refinement. I've tried using the utility to detect NCS groups, then edited in the little windows - also no success. How are other people defining NCS when they aren't content with the equivalences the program finds? Thank you for your insight, Marie

If you really do want to tweak the NCS restraints manually, and if you want cartesian restraints, In an ancient version of phenix.refine (dev1745), it used to be possible to specify in the .eff file with the below syntax (duplicate the blank group which is already present).

I think if you first use auto detection (and cartesion ncs) these lines will be generated in the output .eff and .def file. You can then edit them in the .def file and run, perhaps going back to the original pdb for input in case inappropriate restraints destroyed the output of that first run. (And turn off find_automatically)

You can look at the .geo file from the first run to find the biggest violations in your orig pdb, then look at the input and output files in the density, and decide in each case if it is a true violation that should be released from NCS.

You can duplicate the restraint_group section as many times as you like in case you have multiple chains in the monomer or chains that don't behave as a single rigid body.

refinement { ncs { restraint_group { reference = None selection = None coordinate_sigma = 0.05 b_factor_weight = 10 } restraint_group { reference = chain A and ((RESID 12:255 and not (resid 205 or resid \ 184)) or RESID 1001 ) selection = chain N and ((RESID 12:255 and not (resid 205 or resid \ 184)) or RESID 1001 ) coordinate_sigma = 0.05 b_factor_weight = 0 } restraint_group { reference = chain A and ((RESID 256:365 and not (resid 332 or resid \ 9225)) or ( RESID 1002) ) selection = chain N and ((RESID 256:365 and not (resid 332 or resid \ 9225)) or ( RESID 1002) ) coordinate_sigma = 0.05 b_factor_weight = 0 } _______________________________________________ phenixbb mailing list [email protected] http://phenix-online.org/mailman/listinfo/phenixbb Unsubscribe: [email protected]

Eckhard Hofmann

25 Jan 25 Jan

11:51 a.m.

New subject: Cannot activate NCS in 1.10.1

Hi all, during finalization of a structure I wanted to quickly check the effect of cartesion vs. torsion ncs. Surprisingly changing settings in the GUI did not effect the stats at all (numerically identical runs). I did a quick check with Release 2155 (1.10.1) Release 1692 (1.9) Release 2294 (dev) on the commandline with just the pdb and mtz and with and without main.ncs=true (old version) ncs_search.enabled=True (new versions) Again nothing changed in the stats in the new versions, whereas the old version correctly stated the identified torsion NCS-pairs in the log and gave diverging stats. Maybe I'am triggering a bug here? I know the ncs code is under reconstruction, but no change (especially in log output) is troublesome. Cheers, Eckhard -- Prof. Dr. Eckhard Hofmann Ruhr-Uni Bochum AG Proteinkristallographie, LS Biophysik, ND04/318 44780 Bochum Tel: +49-(0)234/32-24463, Sekr. -24461, FAX: -14762 ORCID: 0000-0003-4874-372X

Oleg Sobolev

6:06 p.m.

New subject: Cannot activate NCS in 1.10.1

Hi Eckhard, My guess would be that newer versions fail to find NCS in your structure. There should be some indication of this fact in the log files. I'm happy to check what is going on and introduce any fixes if needed. Can you please send me (off-list) model and data files to reproduce the behavior you described? The files will be kept in confidence. Best regards, Oleg Sobolev. On Mon, Jan 25, 2016 at 3:51 AM, Eckhard Hofmann < [email protected]> wrote:

...

Hi all, during finalization of a structure I wanted to quickly check the effect of cartesion vs. torsion ncs. Surprisingly changing settings in the GUI did not effect the stats at all (numerically identical runs). I did a quick check with

Release 2155 (1.10.1) Release 1692 (1.9) Release 2294 (dev)

on the commandline with just the pdb and mtz and with and without main.ncs=true (old version) ncs_search.enabled=True (new versions)

Again nothing changed in the stats in the new versions, whereas the old version correctly stated the identified torsion NCS-pairs in the log and gave diverging stats.

Maybe I'am triggering a bug here? I know the ncs code is under reconstruction, but no change (especially in log output) is troublesome.

Cheers, Eckhard

-- Prof. Dr. Eckhard Hofmann Ruhr-Uni Bochum AG Proteinkristallographie, LS Biophysik, ND04/318 44780 Bochum Tel: +49-(0)234/32-24463, Sekr. -24461, FAX: -14762 ORCID: 0000-0003-4874-372X _______________________________________________ phenixbb mailing list [email protected] http://phenix-online.org/mailman/listinfo/phenixbb Unsubscribe: [email protected]

3259

Age (days ago)

3262

Last active (days ago)

List overview

Download

5 comments

5 participants

participants (5)

Eckhard Hofmann
Edward A. Berry
Marie Elizabeth Fraser
Oleg Sobolev
Pavel Afonine