Re: [phenixbb] coordinates and map shifted after phenix.refine without simulated annealing
Hi Hari, What is your space group? I am guessing that the shift is along an axis where the origin is not fixed, for example the b axis in P21. This could happen with rigid-body refinement where a molecule can move as a whole so that the center of mass could move along that axis. Perhaps someone else can comment as to whether there is or could be an optional term in phenix.refine to maintain the center of mass at the same coordinate in such cases. All the best, Tom T
Hi i am using phenix version 1.4 release tag 70
hari@charlie:~$ phenix.refine --version ------------------------------------------------------------------------------- PHENIX: Python-based Hierarchical ENvironment for Integrated Xtallography Version: 1.4 Release tag: 70 Platform: intel-linux-2.6-x86_64 linux User: hari -------------------------------------------------------------------------------
I am seeing a strange shifting of my coordinates "out of my experimental phases map" before and after refinement without simulated annelaing
I am at the final stages of my refinement. When I refine without simulated annealing the resulting model is shifted relative to my experimental phases map into which I build my model The phenix command line strategy i am using are rigid_body+individual_sites+individual_adp+tls .or the same strategy with simulated_annealing=true
The output map coefficients fortunately match my model in both cases . But I am wondering why there is a difference. I am using the mlhl target for refinement to 3.1 A , My R and rfree is 28/31 .
Any ideas why my model shifts without simulated annealing . All pdbs have the same CRYST record . I am using coot version 0.61 for my building and map display using fft inside of coot .
hari _______________________________________________ phenixbb mailing list [email protected] http://www.phenix-online.org/mailman/listinfo/phenixbb
Hello Tom ,
Thanks for your speedy reply
My space group is P1 , and the shift is by around an 1.2 A away and very
much in the vicinity of the input model and experimental map
I suspected the rigid body step (
rigid_body+individual_sites+individual_adp+tls) might be causing this and so
I am trying a refinement without the rigid body.
Also I am quite close to convergence and that step should not be necessary
as a first step.
But I am still confused as to why it shifts the molecule very close by , but
slightly off , like a off kilter MR solution.
Hari
On Sat, Jun 6, 2009 at 10:40 AM, Thomas C. Terwilliger wrote: Hi Hari,
What is your space group? I am guessing that the shift is along an axis
where the origin is not fixed, for example the b axis in P21. This could
happen with rigid-body refinement where a molecule can move as a whole so
that the center of mass could move along that axis. Perhaps someone else
can comment as to whether there is or could be an optional term in
phenix.refine to maintain the center of mass at the same coordinate in
such cases.
All the best,
Tom T Hi i am using phenix version 1.4 release tag 70 hari@charlie:~$ phenix.refine --version PHENIX: Python-based Hierarchical ENvironment for Integrated
Xtallography
Version: 1.4
Release tag: 70
Platform: intel-linux-2.6-x86_64 linux
User: hari I am seeing a strange shifting of my coordinates "out of my experimental
phases map" before and after refinement without simulated annelaing I am at the final stages of my refinement. When I refine without simulated annealing the resulting model is shifted relative to my experimental
phases
map into which I build my model
The phenix command line strategy i am using are
rigid_body+individual_sites+individual_adp+tls .or the same strategy
with
simulated_annealing=true The output map coefficients fortunately match my model in both cases .
But
I
am wondering why there is a difference.
I am using the mlhl target for refinement to 3.1 A , My R and rfree is
28/31
. Any ideas why my model shifts without simulated annealing . All pdbs
have
the same CRYST record .
I am using coot version 0.61 for my building and map display using fft
inside of coot . hari
_______________________________________________
phenixbb mailing list
[email protected]
http://www.phenix-online.org/mailman/listinfo/phenixbb _______________________________________________
phenixbb mailing list
[email protected]
http://www.phenix-online.org/mailman/listinfo/phenixbb
I think I just understood your explanation for the shift after rigid body.
The rigid body phenix.refine picked an origin very close by in my P1 Cell ,
since the choice is arbitrary anyways . And so the output was off .
And I can see how an option to maintain centre of mass will solve that
problem
Thanks again
Hari
On Sat, Jun 6, 2009 at 10:51 AM, hari jayaram
Hello Tom , Thanks for your speedy reply
My space group is P1 , and the shift is by around an 1.2 A away and very much in the vicinity of the input model and experimental map I suspected the rigid body step ( rigid_body+individual_sites+individual_adp+tls) might be causing this and so I am trying a refinement without the rigid body. Also I am quite close to convergence and that step should not be necessary as a first step.
But I am still confused as to why it shifts the molecule very close by , but slightly off , like a off kilter MR solution.
Hari
On Sat, Jun 6, 2009 at 10:40 AM, Thomas C. Terwilliger < [email protected]> wrote:
Hi Hari, What is your space group? I am guessing that the shift is along an axis where the origin is not fixed, for example the b axis in P21. This could happen with rigid-body refinement where a molecule can move as a whole so that the center of mass could move along that axis. Perhaps someone else can comment as to whether there is or could be an optional term in phenix.refine to maintain the center of mass at the same coordinate in such cases. All the best, Tom T
Hi i am using phenix version 1.4 release tag 70
hari@charlie:~$ phenix.refine --version
PHENIX: Python-based Hierarchical ENvironment for Integrated Xtallography Version: 1.4 Release tag: 70 Platform: intel-linux-2.6-x86_64 linux User: hari
I am seeing a strange shifting of my coordinates "out of my
experimental
phases map" before and after refinement without simulated annelaing
I am at the final stages of my refinement. When I refine without simulated annealing the resulting model is shifted relative to my experimental phases map into which I build my model The phenix command line strategy i am using are rigid_body+individual_sites+individual_adp+tls .or the same strategy with simulated_annealing=true
The output map coefficients fortunately match my model in both cases . But I am wondering why there is a difference. I am using the mlhl target for refinement to 3.1 A , My R and rfree is 28/31 .
Any ideas why my model shifts without simulated annealing . All pdbs have the same CRYST record . I am using coot version 0.61 for my building and map display using fft inside of coot .
hari _______________________________________________ phenixbb mailing list [email protected] http://www.phenix-online.org/mailman/listinfo/phenixbb
_______________________________________________ phenixbb mailing list [email protected] http://www.phenix-online.org/mailman/listinfo/phenixbb
participants (2)
-
hari jayaram -
Thomas C. Terwilliger