[phenixbb] DEN refinement is crashing
Melania Strycharska
melania at berkeley.edu
Mon Apr 2 14:46:25 PDT 2012
I've been having the same problem. No matter what settings I use my final R factors are a lot higher than the starting ones. Using default settings they blow up from 31/32 to 50/60 :/ I've gotten them back down to 35 with cartesian annealing but they are still higher than the starting ones and the maps are not any better. With the torsion annealing I can lower them by lowering the starting temperature but even if I go down to 1000K they are still higher than what I start with. I usually optimize both gamma and weight automatically too...
I tried running it with individual sites/individual adp with/and without NCS (trying both cartesian and torsion). I also tried using homologous reference models solved at higher resolution...
If you come up with suggestions for what else to try I would love to hear them :)
cheers,
Melania
On Apr 2, 2012, at 8:45 AM, Mario Sanches wrote:
> Hi Jeff,
>
> I ran it with XYZ and Individual B-factors refinement. I am sending you the inputs right now so you can do a test.
>
> Thanks again,
>
> Mario Sanches
>
>
> On Mon, Apr 2, 2012 at 11:39 AM, Jeff Headd <jjheadd at lbl.gov> wrote:
> Hi Mario,
>
> Did you run your DEN refinement with optimization? If not, that could explain the behavior. It could also be the default refinement settings.
>
> I would be interested in your input files to see why your R values are going up so much, but hopefully with better defaults this won't happen.
>
> Jeff
>
> On Apr 2, 2012 11:23 AM, "Mario Sanches" <mariosan at gmail.com> wrote:
> An update on the DEN job I was running last Friday:
>
> It went to completion. However, the Rwork/Rfree where higher than what I had before (original R + 20%). Since it is stil under development I will not read to much into it and wait for a newer version :-)
>
> I will be happy to provide the inputs if the developers want to use it on a test.
>
> Thank you again for your help,
>
> Mario Sanches
>
> On Fri, Mar 30, 2012 at 11:13 AM, Mario Sanches <mariosan at gmail.com> wrote:
> Thank you Nat and Jeff for your responses.
>
> I was using the latest stable PHENIX version but now downloaded the latest nightly build. It seems to be running fine now, even thought he job is still running. I will let you know of the output once it is finished.
>
> Thank you again,
>
> Mario Sanches
>
>
> On Fri, Mar 30, 2012 at 10:20 AM, Jeff Headd <jjheadd at lbl.gov> wrote:
> Hi Mario,
>
> I've not encountered that bug before, but as Nat said I would suggest trying a more recent installer if you have not.
>
> If you still encounter crashes, could you send me your input files off-list so I could try to replicate the problem here? Everything will be kept confidential.
>
> As an aside, DEN refinement will soon be separated out as its own program, with GUI, to allow for more sensible default parameters and directed control.
>
> Thanks,
> Jeff
>
> On Mar 30, 2012 5:42 AM, "Mario Sanches" <mariosan at gmail.com> wrote:
> I am running phenix.refine using the graphical interface. Under "global refinement paramerters..." I am selecting "DEN". The problem is that Python crashes all the time, always at the same point (see portion of the log output below). Is it a known bug? Is DEN incompatible with some of the options in refinement? My resolution is 2.9A by the way.
>
> Thank you for your inputs.
>
> Mario Sanches
>
> ___________________________
> Log output final portion:
>
> tardy dynamics:
>
> number of bodies: 3678
>
> number of degrees of freedom: 3713
>
> number of atoms * 3: 38220
>
> ratio: 0.097 = 1/10.29
>
> temperature degrees of freedom: cartesian (38220)
>
> kinetic energy sensitivity to generalized velocities:
>
> n: 3713
>
> min: 3.61771e-06
>
> max: 0.0300001
>
> mean: 0.00410663
>
> time step: 0.00100 pico seconds
>
> velocity scaling: True
>
> suppressing allowed origin shifts: False
>
> new target temperature: 2500.00 K
>
> step= 1 temperature= 2500.0 rmsd=0.0446 r_work=0.3010 r_free=0.3214
>
>
> --
> Mario Sanches
> Postdoctoral Researcher
> Samuel Lunenfeld Research Institute
> Mount Sinai Hospital
> 600 University Ave
> Toronto - Ontario
> Canada
> M5G 1X5
> http://ca.linkedin.com/in/mariosanches
>
> _______________________________________________
> phenixbb mailing list
> phenixbb at phenix-online.org
> http://phenix-online.org/mailman/listinfo/phenixbb
>
>
> _______________________________________________
> phenixbb mailing list
> phenixbb at phenix-online.org
> http://phenix-online.org/mailman/listinfo/phenixbb
>
>
>
>
> --
> Mario Sanches
> Postdoctoral Researcher
> Samuel Lunenfeld Research Institute
> Mount Sinai Hospital
> 600 University Ave
> Toronto - Ontario
> Canada
> M5G 1X5
> http://ca.linkedin.com/in/mariosanches
>
>
>
> --
> Mario Sanches
> Postdoctoral Researcher
> Samuel Lunenfeld Research Institute
> Mount Sinai Hospital
> 600 University Ave
> Toronto - Ontario
> Canada
> M5G 1X5
> http://ca.linkedin.com/in/mariosanches
>
> _______________________________________________
> phenixbb mailing list
> phenixbb at phenix-online.org
> http://phenix-online.org/mailman/listinfo/phenixbb
>
>
> _______________________________________________
> phenixbb mailing list
> phenixbb at phenix-online.org
> http://phenix-online.org/mailman/listinfo/phenixbb
>
>
>
>
> --
> Mario Sanches
> Postdoctoral Researcher
> Samuel Lunenfeld Research Institute
> Mount Sinai Hospital
> 600 University Ave
> Toronto - Ontario
> Canada
> M5G 1X5
> http://ca.linkedin.com/in/mariosanches
> _______________________________________________
> phenixbb mailing list
> phenixbb at phenix-online.org
> http://phenix-online.org/mailman/listinfo/phenixbb
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://phenix-online.org/pipermail/phenixbb/attachments/20120402/ea02f9c8/attachment.htm>
More information about the phenixbb
mailing list