[phenixbb] Re: red blobs (Andrea Smith)

Jumping on an old thread… I have a similar situation to Andrea’s; optimize_mask converged to the default values of rsolve and rshrink, but running phenix.mosaic produced quite improved maps. I’m trying, using version 1.21.2-5419, to run phenix.refine with main.mosaic=True and can’t find the setting in the gui, nor is it recognized when I run from the command line. Is it implemented in this version, or in some other version I can download? Best wishes Kevin -- Kevin M. Jude, PhD Structural Biology Research Specialist, Garcia Lab Howard Hughes Medical Institute Stanford University School of Medicine Beckman B177, 279 Campus Drive, Stanford CA 94305 From: [email protected] on behalf of Pavel Afonine Date: Thursday, March 28, 2024 at 7:49 AM To: Andrea Smith Cc: [email protected] , Kay Diederichs Subject: Re: [phenixbb] red blobs (Andrea Smith) Hi Andrea, here is what you can do today.. phenix.mosaic model.pdb data.mtz (yes, command line only, so far). This will take a minute or less (or more if the model is very large) and create model_mosaic.mtz file that contains the following Fourier map coefficients for mFo-DFc and 2mFo-DFc kind of maps: mFo-DFc_whole mFo-DFc_main mFo-DFc_mosaic 2mFo-DFc_whole 2mFo-DFc_main 2mFo-DFc_mosaic (you can see the MTZ file content by using this command: phenix.mtz.dump model_mosaic.mtz) The suffixes 'whole', 'main' and 'mosaic' refer to the types of masks used in calculation of each map and are explained in the paper: https://onlinelibrary.wiley.com/doi/abs/10.1002/pro.4909 To reiterate: 'whole' refers to the standard default bulk-solvent mask; 'main' refers to the largest part of the standard default bulk-solvent mask (small mask droplets inside protein region are removed); 'mosaic' refers to the mosaic mask, which is essentially the 'main' mask plus only those smaller masks that contain measurable amounts of the bulk-solvent (the rest, empty ones, that are responsible for 'red blobs' are removed). Next, you load these maps and your atomic model into Coot and inspect. Artifacts such as 'red blobs' you reported earlier should be absent in 'mFo-DFc_mosaic' map. This may lead to improvement of corresponding '2mFo-DFc_mosaic' map. Now, as you can see, this is separate from refinement. Here is what you should be able to do tomorrow.. Today I will add a parameter to phenix.refine (main.mosaic=True/False), which means if you get and install tomorrow's nightly build of Phenix (dev-5285 and up) the mosaic maps, both mFo-DFc_main and 2mFo-DFc_mosaic, will be present in the MTZ file created by phenix.refine (which is available in both CL and GUI). Let me know if you have any questions! Good luck! Pavel On 3/28/24 06:24, Andrea Smith wrote:

Hi all,

can anyone please tell me how to use the mosaic model? I went through the phenix documentation and didn't find anything about how to use it.

Thank you, Andrea

On Friday, March 15, 2024 22:14 CET, "Andrea Smith" wrote:

...

Hi,

I went through the paper quickly during the day thinking I will have a thorough look at home only to realize I don't have acces to it.

From the quick look it seemed that my biological background will not be enough to understand all of it, but I remember that it said at the end that the mosaic model is implemented in phenix. However, I don't know where. I went through the parameters in GUI and didn't find anything that seemed to fit the description.

Could you please explain what setting I need to use in the refinement?

Thank you, best, Andrea

On Friday, March 15, 2024 16:16 CET, Pavel Afonine wrote:

...

Hi All,

The explanation of the reason for these blobs and the solution is both detailed in depth in this paper:

https://onlinelibrary.wiley.com/doi/abs/10.1002/pro.4909

Quick facts are:

- These blobs are artifacts of bulk-solvent modeling. - You can efficiently deal with them in Phenix. - In some cases (which I witnessed myself), it is important to deal with them for map improvements elsewhere (for example, in regions of interest, such as ligands).

Let me know if you have any questions!

All the best, Pavel

On 3/15/24 07:39, Mitchell D. Miller wrote:

...

Hi Andrea,

You can also put a few zero occupancy atoms in the negative density to force phenix.refine to exclude the region from the bulk solvent mask.

(You may also need to set refinement.mask.ignore_zero_occupancy_atoms = False so that the zero occupancy atoms are included in the mask)

Regards, Mitch

Quoting Kay Diederichs :

...

Hi Andrea,

hmm, did phenix.refine actually use optimize_mask=true ?

If you compare the logfiles of phenix.refine (for the default run with opimize_mask=false, and the new run with optimize_mask=true) side-by-side with xxdiff or vimdiff (yes this needs to be run from a command-line) then there should be a difference.

Making peace with the red blobs is somewhat unsatisfactory from a technical viewpoint, but probably not relevant from a biological one.

I'd guess that the authors of https://journals.iucr.org/a/issues/2024/02/00/pl5035/index.html would be interested to look at your case ...

Best wishes, Kay

Am 15.03.24 um 08:27 schrieb Andrea Smith:

...

Hi Kay,

I tried the mask optimization and there is no change in how the final map looks like.

Should I just make peace with it?

Best, Andrea

On Thursday, March 14, 2024 23:11 CET, Kay Diederichs wrote: > Hi Andrea, > > in your case, phenix.refine seems to fill bulk solvent into volumes > that > are not actually filled by solvent. > It might help to optimize the mask, see > https://phenix-online.org/documentation/reference/refinement.html#bulk-solve... > > "6. Mask parameters". > > Best, > Kay > -- > Kay Diederichs http://strucbio.biologie.uni-konstanz.de > email: [email protected] Tel +49 7531 88 4049 > Fachbereich Biologie, Universität Konstanz, Box M647, D-78457 Konstanz > > This e-mail is digitally signed. If your e-mail client does not > have the > necessary capabilities, just ignore the attached signature > "smime.p7s". > _______________________________________________ > phenixbb mailing list > [email protected] > http://phenix-online.org/mailman/listinfo/phenixbb > Unsubscribe: [email protected]

-- Kay Diederichs http://strucbio.biologie.uni-konstanz.de email: [email protected] Tel +49 7531 88 4049 Fachbereich Biologie, Universität Konstanz, Box M647, D-78457 Konstanz

This e-mail is digitally signed. If your e-mail client does not have the necessary capabilities, just ignore the attached signature "smime.p7s".

_______________________________________________ phenixbb mailing list [email protected] http://phenix-online.org/mailman/listinfo/phenixbb Unsubscribe: [email protected]

phenixbb mailing list [email protected] http://phenix-online.org/mailman/listinfo/phenixbb Unsubscribe: [email protected]