[phenixbb] refinement of an ensemble of structures -> cryoEM variability

Thu Jan 13 12:07:12 PST 2022

Combining all maps from a map series from 3DVA would be equivalent to taking the "consensus" map before 3DVA was applied, I believe (it depends what one means by "combine"). This map is blurry, since it results from aligning images of particles that all have slightly different conformations (and therefore can never align perfectly). So, doing this would cancel all the benefits of running 3DVA.

The same applies with cryoDRGN, it would not be useful to average together all the different states it found in a heterogeneous set of particles.

And to answer this earlier question:

Also.. just curious -- what is "3D variability of this map"? Is this one map that is a composition of several map or an ensemble of maps?

The big picture is that 3DVA is a PCA trying to find the major ways in which 2D particle images differ. Once principal components of variability are identified, it can then generate maps that are linear interpolations between observed particles, taken along the principal components. What is most often done in practice is to generate a series of maps traversing the first one or two principal components (or as many as you want until you explain most of the variability in the data). This can often identify hinge motions in a complex, or similar large-scale conformational changes. Now, because these maps are linear interpolations, they can sometimes show unphysical features, so I am not sure refining a model against these maps would work well. But it would with maps from cryoDRGN, since it does no such interpolation and can even generate maps that are guaranteed to be supported by observed data.

This is very simplified, I am no expert in computational methods. If you want to understand better what is reasonable to do with these maps, you need to know how they are produced, and the two papers I pointed to in my previous message explain this in detail.

I hope this helps,

Guillaume

On 13 Jan 2022, at 17:40, Pavel Afonine <pafonine at lbl.gov<mailto:pafonine at lbl.gov>> wrote:

Hi,

good discussion, thanks!

Instead of having a series of maps would it make sense to combine all these maps into one composite map, and then refine one atomic model against this composite map, and model variability using alternative conformations (very much like in crystallography)?

If that's a possible route then all is needed is a tool to combine the maps and phenix.real_space_refine can already handle models with alternative conformations (can refine occupancies in real space).

Pavel

On 1/13/22 08:17, Oliver Clarke wrote:
Hi,

Just to add my two cents, I agree this would be really useful for a lot of folks. Analysis of continuously distributed variability is very common these days in cryoEM, and having a way to jointly refine an ensemble of models against a series of maps would be very handy. Cryodrgn, 3D-VA in cryosparc, ManifoldEM, multibody refinement in relion - there are many tools now for generating a series of density maps potentially corresponding to conformational modes, so having the capacity in phenix to refine models against each map would be very helpful.

Cheers
Oli

Message: 1
Date: Wed, 12 Jan 2022 10:36:27 +0100
From: vincent Chaptal <vincent.chaptal at ibcp.fr<mailto:vincent.chaptal at ibcp.fr>>
To: Guillaume Gaullier <guillaume.gaullier at icm.uu.se<mailto:guillaume.gaullier at icm.uu.se>>, Pavel Afonine
<pafonine at lbl.gov<mailto:pafonine at lbl.gov>>
Cc: PHENIX user mailing list <phenixbb at phenix-online.org<mailto:phenixbb at phenix-online.org>>
Subject: Re: [phenixbb] refinement of an ensemble of structures ->
cryoEM variability
Message-ID: <ef1a3a20-a0dc-57b0-fcc3-48a9426ce2c3 at ibcp.fr<mailto:ef1a3a20-a0dc-57b0-fcc3-48a9426ce2c3 at ibcp.fr>>
Content-Type: text/plain; charset="utf-8"; Format="flowed"

Hi Guillaume,

thanks for the backup.
It's exactly my feeling also.

Best
Vincent

Le 12/01/2022 ? 10:09, Guillaume Gaullier a ?crit?:
Hi,

I am guessing what we are talking about here are the maps generated by
cryoSPARC 3D variability analysis. See: Punjani A & Fleet DJ (2021) 3D
Variability Analysis: Resolving continuous flexibility and discrete
heterogeneity from single particle cryo-EM.?Journal of Structural
Biology: 107702 https://doi.org/10.1016/j.jsb.2021.107702

But this is not the only program that generates series of maps to
describe continuous heterogeneity from single-particle cryoEM images,
see also cryoDRGN:?Zhong ED, Bepler T, Berger B & Davis JH (2021)
CryoDRGN: reconstruction of heterogeneous cryo-EM structures using
neural networks.?Nature Methods: 1?10
https://doi.org/10.1038/s41592-020-01049-4

Except some ideally rigid particles like apoferritin, pretty much
everything shows some degree of flexibility that generates continuous
heterogeneity in cryoEM images. So, my feeling is that a user-friendly
program to fit a series of models (ideally, auto-generated from a
single starting model) to a map series is probably going to be a
standard requirement pretty soon for typical single-particle cryoEM
projects.

Cheers,

Guillaume

On 11 Jan 2022, at 17:16, Pavel Afonine <pafonine at lbl.gov> wrote:

Hi Vincent,

this looks like a very specialized task that I've never heard of
before! We can add a tool to do that if this becomes something that
more than one person does more than once. Meanwhile, a simple script
in a language of your?preference (python, linux shell, etc) should do
the job. I can help with a script if needed, let me know!

Also.. just curious -- what is "3D variability of this map"? Is this
one map that is a composition of several map or an ensemble of maps?

Pavel

On 1/11/22 01:48, vincent Chaptal wrote:
Hi Phenix-ers,

I thought to ask for something that I believe you have already
implemented, but I'm not sure of the best tool to use.

I have a cryoEM map where I refine my "high resolution" structure. I
also have the 3D variability of this map that shows?several maps
varying around the consensus high-res map. I want to refine an
ensemble (20) of structures, one for every 20?maps around the
consensus map.
Is there a tool in phenix to do this?

I could refine individually the high-res structure into each map
incrementally; since every map differs a little from the?original
one, Real-space-refinement could move the structure a little at a
time. Then I could combine all the PDBs in an?ensemble?
A tool to refine variability would be very useful. Input could be a
PDB and an ensemble of maps, and output would be all the?PDBs combined?

Thank you.

All the best
Vincent

--
Vincent Chaptal, PhD
Director of GdR APPICOM
Drug Resistance and Membrane Proteins Lab

MMSB -UMR5086
7 passage du Vercors
69007 LYON
FRANCE
+33 4 37 65 29 01
http://www.appicom.cnrs.fr
http://mmsb.cnrs.fr/en/

_______________________________________________
phenixbb mailing list

phenixbb at phenix-online.org
http://phenix-online.org/mailman/listinfo/phenixbb

Unsubscribe:
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
Page Title

N?r du har kontakt med oss p? Uppsala universitet med e-post s?
inneb?r det att vi behandlar dina personuppgifter. F?r att l?sa mer om
hur vi g?r det kan du l?sa h?r:
http://www.uu.se/om-uu/dataskydd-personuppgifter/

E-mailing Uppsala University means that we will process your personal
data. For more information on how this is performed, please read here:
http://www.uu.se/en/about-uu/data-protection-policy
--

Vincent Chaptal, PhD

Director of GdR APPICOM

Drug Resistance and Membrane Proteins Lab

MMSB -UMR5086

7 passage du Vercors

69007 LYON

FRANCE

+33 4 37 65 29 01

http://www.appicom.cnrs.fr

http://mmsb.cnrs.fr/en/

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://phenix-online.org/pipermail/phenixbb/attachments/20220112/e891a7ad/attachment-0001.htm>

------------------------------

Message: 2
Date: Wed, 12 Jan 2022 18:11:02 +0530
From: Viney Singh <vineysingh22 at gmail.com>
To: phenixbb at phenix-online.org
Subject: [phenixbb] regarding number of unique reflections
Message-ID:
<CALO_hTx1dtmm2nfkd4+9iPUb+jmAq4kM2ObB=jwrQ5VvCuUTmQ at mail.gmail.com>
Content-Type: text/plain; charset="utf-8"

Dear all,

First of all sorry for the novice question.

I *processed* one of my datasets using xds. Since I was not expecting an
anomalous signal, I kept Friedel's Law = True.
The number of unique reflections I got: *17,000*
I used xds_ASCII.HKL after correct.LP and converted it to mtz using Phenix
with 10% reflections for Rfree calculations.

Now, when I am trying to* refine *the structure in Phenix, the structure is
being refined against *32,300* reflections with *3230* reflections for
Rfree calculation. Looks like refinement is treating Friedel's pair as two
different reflections.

When I am trying to upload PDB on rcsb, on the refinement tab, the number
of reflections used for refinement and Rfree calculations are shown as
32,300 and 3220 respectively, while in the validation report, no. of
reflections used for Rfree calculation are shown as *1700*.

I would really appreciate if someone can guide me to resolve this
discrepancy.

Thanks in advance.

Viney
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://phenix-online.org/pipermail/phenixbb/attachments/20220112/f1d16873/attachment-0001.htm>

------------------------------

Message: 3
Date: Wed, 12 Jan 2022 15:03:49 +0000
From: "Tanner, John J." <TannerJJ at missouri.edu>
To: Viney Singh <vineysingh22 at gmail.com>, "phenixbb at phenix-online.org"
<phenixbb at phenix-online.org>
Subject: Re: [phenixbb] regarding number of unique reflections
Message-ID:
<CH0PR01MB7154E88CABD14AE39C4FD543A7529 at CH0PR01MB7154.prod.exchangelabs.com>

Content-Type: text/plain; charset="us-ascii"

This is a common problem. I use ccp4i CAD to remove all the columns from the .mtz file except for F, SIGF, and FreeR_flag, and then use this new .mtz file in refinement before depositing. For the next structure, remember to do this in the early stages.

--
John J. Tanner
Professor of Biochemistry and Chemistry
Associate Chair of Biochemistry
Department of Biochemistry
University of Missouri
117 Schweitzer Hall
503 S College Avenue
Columbia, MO 65211
Phone: 573-884-1280
Email: tannerjj at missouri.edu<mailto:tannerjj at missouri.edu>
https://cafnrfaculty.missouri.edu/tannerlab/
Lab: Schlundt Annex rooms 3,6,9, 203B, 203C
Office: Schlundt Annex 203A

From: phenixbb-bounces at phenix-online.org <phenixbb-bounces at phenix-online.org> on behalf of Viney Singh <vineysingh22 at gmail.com>
Date: Wednesday, January 12, 2022 at 6:52 AM
To: phenixbb at phenix-online.org <phenixbb at phenix-online.org>
Subject: [phenixbb] regarding number of unique reflections
WARNING: This message has originated from an External Source. This may be a phishing expedition that can result in unauthorized access to our IT System. Please use proper judgment and caution when opening attachments, clicking links, or responding to this email.
Dear all,

First of all sorry for the novice question.

I processed one of my datasets using xds. Since I was not expecting an anomalous signal, I kept Friedel's Law = True.
The number of unique reflections I got: 17,000
I used xds_ASCII.HKL after correct.LP and converted it to mtz using Phenix with 10% reflections for Rfree calculations.

Now, when I am trying to refine the structure in Phenix, the structure is being refined against 32,300 reflections with 3230 reflections for Rfree calculation. Looks like refinement is treating Friedel's pair as two different reflections.

When I am trying to upload PDB on rcsb, on the refinement tab, the number of reflections used for refinement and Rfree calculations are shown as 32,300 and 3220 respectively, while in the validation report, no. of reflections used for Rfree calculation are shown as 1700.

I would really appreciate if someone can guide me to resolve this discrepancy.

Thanks in advance.

Viney
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://phenix-online.org/pipermail/phenixbb/attachments/20220112/70d938d1/attachment-0001.htm>

------------------------------

Message: 4
Date: Wed, 12 Jan 2022 21:28:37 +0530
From: Viney Singh <vineysingh22 at gmail.com>
To: "Tanner, John J." <TannerJJ at missouri.edu>
Cc: "phenixbb at phenix-online.org" <phenixbb at phenix-online.org>
Subject: Re: [phenixbb] regarding number of unique reflections
Message-ID:
<CALO_hTyJj3bSg_rev1hZVEvr=K91+C9x1AtbdrpiOajsSKk-Hg at mail.gmail.com>
Content-Type: text/plain; charset="utf-8"

Thanks a lot, Dr .Tanner. I will try the suggestion. Currently, I tried
running xds_ASCII.HKL on pointless followed by truncate. Seems like mtz is
now all good. But I am wondering if this is the right way to go about it?

On Wed, Jan 12, 2022 at 8:33 PM Tanner, John J. <TannerJJ at missouri.edu>
wrote:

This is a common problem. I use ccp4i CAD to remove all the columns from
the .mtz file except for F, SIGF, and FreeR_flag, and then use this new
.mtz file in refinement before depositing. For the next structure, remember
to do this in the early stages.

--

John J. Tanner

Professor of Biochemistry and Chemistry

Associate Chair of Biochemistry

Department of Biochemistry

University of Missouri
117 Schweitzer Hall

503 S College Avenue
Columbia, MO 65211
Phone: 573-884-1280

Email: tannerjj at missouri.edu <tannerjj at missouri.edu>
https://cafnrfaculty.missouri.edu/tannerlab/

Lab: Schlundt Annex rooms 3,6,9, 203B, 203C

Office: Schlundt Annex 203A

*From: *phenixbb-bounces at phenix-online.org <
phenixbb-bounces at phenix-online.org> on behalf of Viney Singh <
vineysingh22 at gmail.com>
*Date: *Wednesday, January 12, 2022 at 6:52 AM
*To: *phenixbb at phenix-online.org <phenixbb at phenix-online.org>
*Subject: *[phenixbb] regarding number of unique reflections

*WARNING:* This message has originated from an External Source. This may
be a phishing expedition that can result in unauthorized access to our IT
System. Please use proper judgment and caution when opening attachments,
clicking links, or responding to this email.

Dear all,

First of all sorry for the novice question.

I *processed* one of my datasets using xds. Since I was not expecting an
anomalous signal, I kept Friedel's Law = True.

The number of unique reflections I got: *17,000*

I used xds_ASCII.HKL after correct.LP and converted it to mtz using Phenix
with 10% reflections for Rfree calculations.

Now, when I am trying to* refine *the structure in Phenix, the structure
is being refined against *32,300* reflections with *3230* reflections for
Rfree calculation. Looks like refinement is treating Friedel's pair as two
different reflections.

When I am trying to upload PDB on rcsb, on the refinement tab, the number
of reflections used for refinement and Rfree calculations are shown as
32,300 and 3220 respectively, while in the validation report, no. of
reflections used for Rfree calculation are shown as *1700*.

I would really appreciate if someone can guide me to resolve this
discrepancy.

Thanks in advance.

Viney

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://phenix-online.org/pipermail/phenixbb/attachments/20220112/094a7f44/attachment.htm>

------------------------------

_______________________________________________
phenixbb mailing list
phenixbb at phenix-online.org
http://phenix-online.org/mailman/listinfo/phenixbb

End of phenixbb Digest, Vol 194, Issue 2
****************************************

_______________________________________________
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
_______________________________________________
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

N?r du har kontakt med oss p? Uppsala universitet med e-post s? inneb?r det att vi behandlar dina personuppgifter. F?r att l?sa mer om hur vi g?r det kan du l?sa h?r: http://www.uu.se/om-uu/dataskydd-personuppgifter/

E-mailing Uppsala University means that we will process your personal data. For more information on how this is performed, please read here: http://www.uu.se/en/about-uu/data-protection-policy
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://phenix-online.org/pipermail/phenixbb/attachments/20220113/34eb27e4/attachment.htm>