[phenixbb] Phaser-EP log file

Randy Read rjr27 at cam.ac.uk
Sun Feb 28 06:28:40 PST 2016

Dear Nishant,

I suspect that you’re using an older version of Phenix and Phaser, because I’m not seeing this in recent logfiles.

Because the refinement of the anomalous scatterer model in Phaser is done using a minimizer, whereas what we want to do is to maximize the log-likelihood-gain, the program minimizes minus log-likelihood-gain.  At times, we have accidentally printed out the target in places in the log file, and it looks like you’re running a version where we still had one of those output errors.

Even if you hadn’t carried out building and refinement, the relatively high FOM would have been a good sign.  A better indicator for a correct substructure and good phases would be the scores that AutoSol looks at to compare different solutions, such as the predicted correlation to the true map and the skewness of the density map.  Of course, the best indicator is the ability to build a good model, which AutoBuild has done!  So clearly this structure is basically solved.

As for your second question, it’s fairly arbitrary which symmetry copy is built in the map, and if you ran the whole procedure again slightly differently, you might end up with a substructure on a different choice of origin.  I think you’ll find that the two models are equivalent, differing only possibly in choice of origin and in symmetry copy.  You could check this by using phenix.famos (which compares models to see if they are the same after checking alternative origins and symmetry mates).

Best wishes,

Randy Read

Randy J. Read
Department of Haematology, University of Cambridge
Cambridge Institute for Medical Research    Tel: +44 1223 336500
Wellcome Trust/MRC Building                         Fax: +44 1223 336827
Hills Road                                                            E-mail: rjr27 at cam.ac.uk
Cambridge CB2 0XY, U.K.                               www-structmed.cimr.cam.ac.uk

> On 27 Feb 2016, at 21:43, Nishant Pandey <nishant.pandey_85 at yahoo.co.in> wrote:
> Hi,
> I have a confusion regarding solving iodine SAD-data in Phaser-EP.
> 1. GUI result shows Log-likelihood gain: 115037.70 and FOM:  0.527. But in log file
>     SAD Refinement Table (Sorted)
>    -----------------------------
>    SET    SpaceGroup    FOM           Log-Likelihood   Rfac  Composition  I   
>    # 1      P 21 21 2          0.53            -115038                69.4          23  
> Why there is negative value of Log-Likelihood in log file? Is Log-Likelihood and Log-likelihood gain is two different things?
> Has Phaser solved it? 
> When I feed these results to Autobuild, the model was ~80 % complete with Rfree and Rwork of 0.3171 and 0.2754 and Model Map correlation 0.74. Model was further refined in phenix.refine upto Rfree = 0.27.
> 2. When I am solving same dataset with Autosol and Autobuild, the dimer orientation have changed? Is this 'normal' or one of the solution is wrong  and if it then which one should choose to refine further?
> Thanks in advance.
> _______________________________________________
> phenixbb mailing list
> phenixbb at phenix-online.org
> http://phenix-online.org/mailman/listinfo/phenixbb
> Unsubscribe: phenixbb-leave at phenix-online.org

More information about the phenixbb mailing list