[phenixbb] Rotamer restraints in phenix.real_space_refine

Pavel Afonine pafonine at lbl.gov
Thu Feb 28 20:16:13 PST 2019


Hi Ashley,

yes, the behavior is different now. Before we would restrain each 
residue to its nearest valid rotamer without considering density map, 
which works well in many cases but doesn't in some. In current version 
we optimize rotamers to minimize clashes (but we mind H-bonds) and 
maximize model-to-map fit. Clearly, it may happen that local model 
geometry is such that no optimal fit is possible, this is probably 
responsible for some outliers that remain after refinement.

If you have a clear example where this does no work please send it to us 
and we will investigate.

Pavel

On 2/28/19 09:59, Oleg Sobolev wrote:
> Hi Ashley,
>
> Indeed, apart from strategy defaults, the rotamer fitting algorithm 
> itself was changed between 1.13 and 1.14. It is possible it behaves 
> slightly different (we were hoping for the better behavior in most 
> cases). I'm not sure that it is possible to get exact 1.13 behavior in 
> 1.14. I would expect you get small amount of rotamer outliers (<1%). 
> Otherwise we probably should investigate. Sometimes inclusion of 
> hydrogen atoms could help get a better refined model.
>
> Best regards,
> Oleg Sobolev.
>
> On Thu, Feb 28, 2019 at 4:24 AM Ashley Pike <ashley.pike at sgc.ox.ac.uk 
> <mailto:ashley.pike at sgc.ox.ac.uk>> wrote:
>
>     Oleg,
>
>     Thanks - so behaviour has changed. Running with local_grid +
>     rotamer restraints=true seems to optimise rotamers (although does
>     not necessarily eliminate outliers). Is there an equivalent
>     combination that enforces rotamers (ie. doesn’t allow outliers)?
>     This is behaviour of 1.13 with just rotamer_restraints=true (and
>     no local_grid _search). In my case, I can't really justify having
>     rotamer outliers when refining against 4A cryo-em maps.
>
>     Many thanks,
>
>     Ashley
>
>     *From:*Oleg Sobolev <osobolev at lbl.gov <mailto:osobolev at lbl.gov>>
>     *Sent:* 28 February 2019 00:14
>     *To:* Ashley Pike <ashley.pike at sgc.ox.ac.uk
>     <mailto:ashley.pike at sgc.ox.ac.uk>>
>     *Cc:* phenixbb at phenix-online.org <mailto:phenixbb at phenix-online.org>
>     *Subject:* Re: [phenixbb] Rotamer restraints in
>     phenix.real_space_refine
>
>     Hi Ashley,
>
>     Do you have "local_grid_search" box checked in refinement
>     settings->strategy when running 1.14? This should be on by default
>     since 1.14, but it was not the case in 1.13. Sometimes people
>     using .eff files from 1.13 in later versions end up without old
>     defaults for refinement strategy, disabling rotamer fitting.
>
>     I can have a closer look if you send me the .log file (off-list).
>
>     Best regards,
>
>     Oleg Sobolev.
>
>     On Wed, Feb 27, 2019 at 10:56 AM Ashley Pike
>     <ashley.pike at sgc.ox.ac.uk <mailto:ashley.pike at sgc.ox.ac.uk>> wrote:
>
>         Hi - Am I right in thinking that rotamer restraints behaves
>         differently in latest official release (1.14-3260) compared to
>         last but one official release (1.13-2998)? Issuing
>         rotamer_restraints=True from command line or GUI used to give
>         0% rotamer outliers in the molprobity analysis at end of log
>         in 1.13-2998 but now in 1.14 gives non-zero for same input
>         model. It is as if they are not being applied.
>
>         In the old log file it used to clearly print - **Update
>         rotamer restraints*** after every macrocycle, in the 1.14 logs
>         this is omitted.
>
>         Many thanks,
>
>         Ashley
>
>         _______________________________________________
>         phenixbb mailing list
>         phenixbb at phenix-online.org <mailto:phenixbb at phenix-online.org>
>         http://phenix-online.org/mailman/listinfo/phenixbb
>         Unsubscribe: phenixbb-leave at phenix-online.org
>         <mailto:phenixbb-leave at phenix-online.org>
>
>
> _______________________________________________
> phenixbb mailing list
> phenixbb at phenix-online.org
> http://phenix-online.org/mailman/listinfo/phenixbb
> Unsubscribe: phenixbb-leave at phenix-online.org

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://phenix-online.org/pipermail/phenixbb/attachments/20190301/4168752c/attachment.htm>


More information about the phenixbb mailing list