[phenixbb] mr_rosetta question

Tom Terwilliger tterwilliger at newmexicoconsortium.org
Sat Jul 18 11:17:45 PDT 2020


Hi Wei,

Normally just set

  resolution=3.8

or whatever it is.  Leave place_model.mr_resolution alone (set
automatically) and leave rosetta_modeling.map_resolution alone (always 3 A
to match rosetta expectations).

All the best,
Tom T

On Sat, Jul 18, 2020 at 10:08 AM Wei Wang <ww2283 at columbia.edu> wrote:

> Thanks Tom! This is very helpful and informative. One more question: if my
> data resolution is worse than 3Å, should I modify the following parameters
> to replace the default values of 3 for the mr_rosetta run?
>
> mr_rosetta.rosetta_modeling.map_resolution=
> mr_rosetta.crystal_info.resolution=
> mr_rosetta.place_model.mr_resolution=
> Wei
>
>
>
>
> On Jul 17, 2020, at 10:18 PM, Tom Terwilliger <
> tterwilliger at newmexicoconsortium.org> wrote:
>
> Hi Wei,
>
> Your sequence file should be 1 copy of the unique sequence.
>
> Your search model should be either your original search model, or probably
> better, any one copy of a placed and already refined model (just take your
> A chain from your existing solution and call it a search model).
>
> Your fixed_model consists of the 5 placed copies you already have.
>
> Your value of ncs_copies is 1 if you are using a fixed_model to hold the 5
> existing copies.
>
> This approach assumes that the 6th copy is not that similar to the other 5
> and NCS averaging will not be useful. Note that there is an alternative
> approach to do all this that involves specifying the translations and Euler
> angles of the 5 placed copies.  This alternative method will apply NCS to
> all 6 copies and will work much better than the way described above if all
> copies actually are very similar.  Probably the way described here is
> better for your case because you get the advantage of using the refined
> positions of the first 5 copies to get an accurate description of most of
> the cell contents and as mentioned below the 6th copy ma not be similar.
>
> Note also that the chances of success are low because you already tried to
> find the 6th copy with Phaser and it didn't work.  MR-rosetta could
> possibly work if the 6th copy is just in a different conformation.  If it
> is disordered or otherwise not visible...it is not likely to work.
>
> You might also just look very carefully in your 2mFo-DFc map for any
> indication of where the 6th molecule may be located.  If you can see a hint
> of it, a real-space search may be worth trying instead of further molecular
> replacement.  Also extensive auto-building could possibly show you where
> the 6th copy is located.
>
> If you want to try the alternative approach, see here:
> https://www.phenix-online.org/documentation/reference/mr_rosetta.html
>
> under this description:
>
> fixed_ensemblesIf you already know the placement of one or more molecules
> you can specify them as fixed ensembles. NOTE 1: you are specifying
> location and orientation of one or more copies of the search model NOTE 2:
> you cannot specify use_all_plausible_sg if you have fixed ensembles
>
>
>
> fixed_ensembleID_list = NoneEnter the word 'ensemble_1' to indicate that
> you want to specify a copy of your search model that is to be fixed. To
> specify more than one placement just say 'ensemble_1' more than once. For
> example if you
>
>
>
> ....
>
>
> Let me know if that doesn't do it or if I gave you any of the parameters
> incorrectly.
>
> All the best,
> Tom T
>
> On Fri, Jul 17, 2020 at 7:07 PM Wei Wang <ww2283 at columbia.edu> wrote:
>
>> Dear all,
>>
>> I would like to get help on a mr_rosetta run. I have a case that 5 copies
>> of searching model was found by Phaser, and I would like to find the sixth
>> one. In addition I would like to improve the building of the MR solution.
>>
>> In such case I’m considering a mr_rosetta run, with some parameters like
>> this: phenix.mr_rosetta seq_file=seq.dat search_models=placed_5_copies.pdb
>> already_placed=true ncs_copies=1 …
>>
>> My question is, should I use the already placed 5 copies model as
>> search_models, with ncs_copies=1, or should I use the original searching
>> model with ncs_copies=6 (in this case I guess I should get rid of the other
>> four copies from the result of Phaser)? For seq.dat, should I use the
>> sequence from one copy of searching model, or paste 5 copies into the
>> seq.dat?
>>
>> Thanks,
>> Wei
>>
>>
>>
>>
>> _______________________________________________
>> phenixbb mailing list
>> phenixbb at phenix-online.org
>> http://phenix-online.org/mailman/listinfo/phenixbb
>> Unsubscribe: phenixbb-leave at phenix-online.org
>
>
>
> --
> Thomas C Terwilliger
> Laboratory Fellow, Los Alamos National Laboratory
> Senior Scientist, New Mexico Consortium
> 100 Entrada Dr, Los Alamos, NM 87544
> Email: tterwilliger at newmexicoconsortium.org
> Tel: 505-431-0010
>
>
>

-- 
Thomas C Terwilliger
Laboratory Fellow, Los Alamos National Laboratory
Senior Scientist, New Mexico Consortium
100 Entrada Dr, Los Alamos, NM 87544
Email: tterwilliger at newmexicoconsortium.org
Tel: 505-431-0010
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://phenix-online.org/pipermail/phenixbb/attachments/20200718/ec066c33/attachment.htm>


More information about the phenixbb mailing list