[phenixbb] Low resolution refinement

[Randy John Read]

For the most part, the AlphaFold models tend to be locally pretty good, so at low resolution it will be a good idea to use the original AlphaFold model as a reference model. Because the restraints are in terms of torsion angles, things like domain reorientations can take place even with the restraints present.

Tristan Croll has been playing with AlphaFold models for rebuilding low-resolution structures in his ISOLDE plugin to ChimeraX, and he’s finding that they make low resolution work much easier in general.

Best wishes,

Randy

> On 17 Nov 2021, at 12:54, Armando Albert <xalbert at iqfr.csic.es> wrote:
> 
> Thank you all, 
> The MR solution comes from an alphaFold model. It packs well, it provides a sensible (biological) oligomer and a reasonable electron density (tubes for alpha helices). In any case, I am not sure about the solution. Phaser refinement seems to work and I wonder if I can progress with phenix.refine (or any other). 
> Armando
> 
> 
> 
>> El 17 nov 2021, a las 12:51, Randy John Read <rjr27 at cam.ac.uk> escribió:
>> 
>> Also, these days I would recommend using an AlphaFold2 model as a reference model, unless you have the structure of a very closely-related homologue.
>> 
>> Randy
>> 
>>> On 16 Nov 2021, at 20:58, Pavel Afonine <pafonine at lbl.gov> wrote:
>>> 
>>> Hi Armando,
>>> 
>>>> I have got a very promising MR solution at 5A resolution. Can I refine this structure? Any hint would be very much appreciated
>>> 
>>> this is a low resolution indeed, so modeling and refinement is likely to be challenging. All the standard 'low-resolution refinement' advises apply here. Since there are few data (low-res) you need to employ as much 'a priori' known information as possible, in form of restraints or/and constraints. Those typically include:
>>> 
>>> - If NCS is present, use NCS constraints (yes, constraints, not restraints, as at this resolution you are highly unlikely to see data-supported genuine differences between NCS copies!);
>>> 
>>> - if similar homologous models obtained at higher resolution are available -- try using them as reference models;
>>> 
>>> - Use secondary structure restraints. Important: make sure to obtain secondary structure annotation as accurately as possible.
>>> 
>>> - Use Ramachandran plot restraints (to avoid appearing outliers during refinement, not to fix existing ones!);
>>> 
>>> - Avoid stretching your imagination by adding model in the absence of density;
>>> 
>>> - At this resolution refining group B factors (two B per residue) may be a good idea.
>>> 
>>> Let me know if you have any more specific questions!
>>> 
>>> Good luck!
>>> Pavel
>>> 
>>> 
>>> _______________________________________________
>>> phenixbb mailing list
>>> phenixbb at phenix-online.org
>>> http://phenix-online.org/mailman/listinfo/phenixbb
>>> Unsubscribe: phenixbb-leave at phenix-online.org
>> 
>> -----
>> Randy J. Read
>> Department of Haematology, University of Cambridge
>> Cambridge Institute for Medical Research     Tel: +44 1223 336500
>> The Keith Peters 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
>> 
> 

-----
Randy J. Read
Department of Haematology, University of Cambridge
Cambridge Institute for Medical Research     Tel: +44 1223 336500
The Keith Peters 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