Re: [ccp4bb] "reset" a structure before re-refinement

[Matthias Barone]

Thanks James for your reply. I my naive way, I thought that when doing paired 
refinement, it would be sufficient to run several cycles of refinement with 
simulated annealing followed by a thorough real space rebuilding. What is your 
thought about using  the "jiggling" of additional cartesian annealing steps 
(phenix) to check the influence of, say, an additional 0.1A outermost shell? 
 I used paired refinement over a couple of cycles and thoroughly deleted what 
was not visible anymore. Less for the removal of bias, but more to increase 
degrees of freedom... But now that I read your answer, I have the impression 
that this was not enough to remove bias of the sidechains?

matthias



>>> James Holton  08/19/17 7:23 PM >>>
 
 Yes, B factors are indeed a super-absorbent sponge for model bias.  
Best to re-set those before re-refinement.  But you should also do 
something with the coordinates, and probably the occupancies too.  
 
 My jiffy to add to Graeme's list of options is this script:
 http://bl831.als.lbl.gov/~jamesh/scripts/jigglepdb.awk
 It has a variety of options, including separate rms shifts for 
coordinates, B factors and occupancies (shift=, Bshift=, and Oshift=), and 
a special option called "shift=byB" that moves coordinates in accordance to 
the atom's B factor.  The distribution of the shifts can be Gaussian 
(default), Lorentzian or uniform sphere.  The latter is useful if you want 
to avoid very large shifts (see below).  By default, one conformer (i.e. A, 
B, C) is randomly given an occupancy of "1" for all residues and all the 
other conformers are given  zero.  This is an attempt to simulate the 
cell-to-cell variation of the structure, which can only have one conformer 
at a time.  If you specify "keepocc=1" this behavior gets turned off.  Any 
non-zero "Oshift" will add random noise to all occupancies.
 
 I'm sure all this functionality would be easy to re-create using CCTBX 
as well.
 
 
 I've actually played around with "jiggling" pdb files a fair bit.  
Turns out that in order to remove "bias" completely you need to kick the 
atoms by an rms distance comparable to the relevant resolution.  That is, 
1.5 A for 1.5 A spots, but also 6 A if you want to un-bias 6 A spots.  This 
tends to be outside the radius of convergence of most refinement programs.  
In fact, kicking atoms by as little as 0.5 A can often result in some side 
chains falling into alternate rotamers and the like.  Waters can also fall 
out of place and drift into nearby pockets of liberated density.  You can 
do a "cleanup" after re-refinement to correct for such gross errors.  
Looking for the largest differences between starting and re-refined models  
   generally lights these up.  But before you "fix" these things you have 
to start asking questions about what "bias" really is.  
 
 It seems a popular "bias-reduction" approach is to re-run Phaser or 
other molecular replacement job, but all that really does is put your model 
onto an alternative origin or asu choice.  These choices are arbitrary, of 
course, and crystallographically equivalent.  Once you have corrected for 
the origin shift using something like "reforigin" you can see that all your 
Phaser run has produced is a slight rigid-body shift relative to where you 
started.  Hardly a bias-removal technique.
 
 So, what is "bias"?  We tend to think of "bias" the same way we think 
of "twinning": as a synonym for "evil".  Something to be kept out of our 
models and data at all costs.  But if you think about it a bit you may 
realize that even having the backbone structure itself intact is a form of 
"bias".  As in you are "biasing" your model to resemble the overall fold 
you originally assigned to the structure.  You may or may not be all that 
worried about this.  And if you are certain that the main chain trace is 
correct, then enforcing it in further analysis is not "bias", it is "prior 
knowledge".  I suppose knowing the difference between these two things is 
what science is all about.
 
 So, I'd say that "resetting" a structure depends on what aspect of the 
structure you're trying to test.  If you made a mistake in the backbone 
trace, or even a rotamer assignment, then doing a 0.5A jiggle isn't going 
to reset that mistake.  But if your trying to test the influence of data 
between 1.5 A and 1.4 A, then I'd say do a jiggle of at least half that 
distance.
 
 -James Holton
 MAD Scientist
 
 
 On 8/17/2017 8:40 AM, Robbie Joosten   wrote:
 
 *.EmailQuote {
margin-left: 1.0pt;
padding-left: 4.0pt;
border-left: rgb(128,0,0) 2.0px solid;
}
  p.x_MsoNormal, li.x_MsoNormal, 

[ccp4bb] Antw: Re: [ccp4bb] Ramachandran oultliers increase upon restrained refinement

[Matthias Barone]

Beside the good propositions of fixing the environment, flip the peptide, or 
using PDB_REDO, I would suggest the following additional attempts: 
Try
1) setting the OCC of the sidechain (CG and further) below 0.3 or mutate 
directly to Ala. See which ones remain visible in a diff map and bring them 
back in during the upcoming ref cycles (refine OCCs in Phenix)
2) use simulated CART annealing to get the side chains out of your way
3) use a FEM map against misinterpreted side chains 
4) In very nasty cases, I use MOLOC to do real space refinement and give a lot 
of weight to the H-Bond network. This usually helps to keep weakly defined 
regions in place.

good luck, matthias




>>> Vipul Panchal  02.07.17 19.31 Uhr >>>
I do agree with Eleanor. When I was solving structure at 2.16A resolution, 
outlier residues were having stearic clash or required side chain flipping. 


At 2.76A resolution, hydrogen bonds would be difficult to visulize. 


I found phenix.refine more user friendly. You may too find it useful.

Vipul Panchal,
Ph.D. student
CSIR-IGIB
(M)- 091 7678617949

On 02-Jul-2017 9:14 PM, "Eleanor Dodson"  wrote:
Just remember - most Ramachandran outliers are due to errors in the environment 
- eg: maybe some side chains clash hence refinement tries to move things to 
accommodate that bad feature, etc etc etc... At that resolution it is 
inevitable you will have some level of mis-interpretation .. 
4% outliers does not surprise me. 

EDSTATS is a useful tool to find things such as pep flips - you can access it 
most simply from CCP4 GUI2 .

But unless the outliers are in a important part of the structure I would 
suggest checking, then living with them.

Eleanor









On 2 July 2017 at 16:31, Rajesh Kumar  wrote:
Dear Meytal,

PHENIX-REFINE has an option for Ramachandran outlier refine. If you check this 
on, it will do the work. But after this refinement, you must check every 
residues to make sure residues are in the density.


Thank you
Rajesh




 ---x
With regards
Rajesh K. Harijan, Ph.D.
Schramm Laboratory
Albert Einstein College of Medicine
1300 Morris Park Ave., Bronx, NY 10461Tel: 718.430.2777  |  Fax: 718.430.8565






On Sun, Jul 2, 2017 at 7:26 AM, Meytal Galilee  wrote:
Dear all,
I am refining a 2.76A structure using refmac5.
I keep fixing the Ramachandran outliers down to 18 (1.9%), however, upon 
restrained refinement, the outliers increase back to 35.
Do you have any suggestions how can I fix my structure?
Thanks,
Meytal Galilee
 
 
 


 



 


 





 

[ccp4bb] Antw: Re: [ccp4bb] What are acceptable Rwork/Rfree for publication

[Matthias Barone]

If you dont use Phenix, print the figures of 
https://doi.org/10.1016/S0969-2126(02)00743-8
and use them beside your screen. Helps a lot if you wanna have a quick look at 
Rfact distributions..




>>> Paul Emsley  14.06.17 15.25 Uhr >>>
On 14/06/2017 14:09, Khoa Pham wrote:
> Dear CCP4 members,
> 
> I am refining a structure with a resolution of ~ 2.9 A and the Rwork/Rfree 
> are are 
> 0.34/0.37, respectively.
> My question is that these Rwork/Rfree are acceptable for publication. 

If you have Phenix, try POLYGON. That will give you a good clue.

 >If not, how can I
> reduce them?
> 

That is more complicated to answer.


Paul.

[ccp4bb] Antw: Re: [ccp4bb] When should one add solvents

[Matthias Barone]

Hi Abhishek
Im by far not as experienced as others in the bulletin board, but I made the 
experience that automated water update with Phenix leads to an aggressive 
placement of solvent molecules, which can be damped a bit by restricting H-bond 
distances of newly placed waters. As Robbie and Eleanor mentioned, manual 
inspection is a must. 
In your situation, I would first check for the optimal refinement strategy 
(amount of TLS groups for example) and the use FEM maps to rebuild missing 
residues. If this does not help, start filling waters slowly.
At this point, I always run paired refinement of models with manually placed 
waters and one model where I unleash phenix water update. Usually, Phenix is 
doing a good job there and produces reasonable water placements.
I regularly let PDB_REDO have a look at waters and cross check those desities 
with a FEM map. It could be that one model reveals density and information that 
you then can transfer to the other model. Its worth a try, but no guarantee 
this helps..
Especially the PDB_REDO *_final.scm turned out to be a handy script for coot, 
allowing for manual inspection of the water molecules removed by PDB_REDO.
Please correct me if Im wrong, but if it comes to water placement, a model with 
reduced Rfree is not neccessarily a better model. Rfree will most probalbly go 
down while loading more waters, simply as you reduce the degrees of freedom 
(see Merritt, To B or not to B, ActA D 2012.). 
best, matthias




>>> Eleanor Dodson  18.04.17 13.34 Uhr >>>
I routinely use coot " difference map peaks" to inspect maps to see errors - 
fit residues - find alternate conformations- etc. it is often obvious then that 
you have found a solvent molecule; good hydrogen bonds, clear density, and then 
I add it. i know this is old fashioned, but honestly I think no slower than 
invoked automated searches then having to correct the errors they introduce.. 
Eleanor


On 18 April 2017 at 10:35, Robbie Joosten  wrote:
   Assuming you do this by hand (which I highly recommend), you can add waters 
as soon as it becomes obvious that they are not something else (particularly 
missing  parts of the protein). It is best to be conservative at start.
  
 Cheers,
 Robbie
  
   From: CCP4 bulletin board [mailto:CCP4BB@JISCMAIL.AC.UK] On Behalf Of Dr 
A.A. Jalan
 Sent: Monday, April 17, 2017 23:46
 To: CCP4BB@JISCMAIL.AC.UK
 Subject: [ccp4bb] When should one add solvents
 
 
  
 Dear All,
 I have a very elementary question. At what stage of refinement is it 
appropriate to start adding solvent (water). For example, during refinement of 
a structure, I was able to build ~ 90% of residues. The density of the 
remaining  residues is not complete. If I take a guess and add residues based 
on my knowledge of the sequence, the r-free increases. Since I see no further 
way to improve the maps (any suggestions?), should I start adding water in the 
hope of improving the overall density  and revealing the missing bits. I would 
really appreciate any inputs.
 thank you
 Abhishek
   
 
 


 
 


 

[ccp4bb] Antw: Re: [ccp4bb] R/Rfree values

[Matthias Barone]

Dear Rohit
Additionally to the very good suggestions, I'd feed the structure into
the PDB_REDO server. Its a fast and easy way to detect if the refinement
strategy is correct. The server gives you a good feeling how tight the
x-ray weights need to be. 
If however PDB_REDO does not manage to decrease the Rfactor gap of 7%, I
would suggest to switch to Phenix and give it a try.
If this does bring the factors down and relax the structure, let
pointless have a look into the XDS_ASCII file. Pointless and Xtriage are
two fast programs to detect if you are correct with the choice of the
space group. They give you also a first hint about twinning and the
amount of twinning.
good luck,
matthias




>>> Prem Prakash  14.12.16 19.20 Uhr >>>
Dear Rohit, I am totally agree with Mark, that just R free and R work
does not decide the structure solved, you have to see other parameters
as Mark already suggested. What is and redundancy and also what is the
mosaicity? please look at these parameters too. 

Best 
Prem 


On Wed, Dec 14, 2016 at 10:07 PM, Ashok Nayak 
wrote:
Dear Rohit,
Look at what percentage of the reflections are in the Rfree set, if its
less say 5 %; and you have a good redundancy you can use 10 %. You might
like to use the other Rfree set[1 if you have already used the 0 set] or
consider using the whole set of reflections as a free set if need be if
you suspect a bias in your Rfree set.
As suggested earlier I would also look at how the Fobs and Fcalc
correlate as a function of resolution from the sigma A plot or CC*
values.

Sometimes overestimating the resolution would add up more noise and it
would give you bad R values.

and If the molecule is elongated you might have diffraction anisotropy
and in that case The R values don't come down.

I would also like to use Refmac along with Phenix; that helped in my
case.

So you would like to take them one by one and figure out carefully.

Best Wishes

Ashok Nayak
Post-Doctoral Fellow,
Department of Physiology and Biophysics
VCU Medical Centre,1101 E Marshall ST, 
Richmond,VA
USA


  


On Wed, Dec 14, 2016 at 11:12 AM, Morais, Marc C. 
wrote:
  In addition to what others have mentioned, you might want to consider
the unpleasant possibility that part of your structure is simply built
incorrectly. Sometimes rebuilding a  short loop or even a few residues
in that loop can do the trick. Look carefully for regions of your map
with poor density and/or regions of your model with poor geometry, and
see if there might be an alternate way to build that region. There is
often a psychological  barrier to attempting these rebuilds, because
they are difficult (if they weren't we would have built correctly the
first time). However, once begun, it is often less painful than
expected. SA/composite omit maps can help.
   From: CCP4 bulletin board [CCP4BB@JISCMAIL.AC.UK] on behalf of
Eleanor Dodson [eleanor.dod...@york.ac.uk]
 Sent: Wednesday, December 14, 2016 9:49 AM
 To: CCP4BB@JISCMAIL.AC.UK
 Subject: Re: [ccp4bb] R/Rfree values
 
 
 
 es - look carefully at your data quality indicators - batch scales,
Wilson plot - moments etc.. tHese can show up if there is a problem with
ice rings or crystal decay or whatever. 
 
 Then I always look at the REFMAC plot of  v  
 
 If they overlap well - good - but problems with scaling will show up
there.
 
 
 Eleanor
 
 
 On 14 December 2016 at 15:21, Mark J van Raaij 
 wrote:
  Dear Rohit,
 
 I wouldn’t judge a structure just by the Rwork and Rfree values, but
also by the validation and other statistics (bond lengths, angles,
Ramachandran plot, map quality, fit to map, average B values). If these
are all ok, you should be able to “get away with”  an Rfree of 33%.
 In your email you state that you have already made a significant effort
in different refinement strategies, so perhaps there is no improvement
to be made there.
 The reason for good, and reprocessing might help. Although the most likely 
outcome is
that you can’t significantly improve it, but at least trying will put
your mind more at ease.
 In the end, the only way to improve the structure, and R-factors, may
be to grow a better crystal, cryo-protect better and/or collect better
data - this particular crystal may just have some kind of disorder.
 
 Greetings,
 
 Mark J van Raaij
 Dpto de Estructura de Macromoleculas
 Centro Nacional de Biotecnologia - CSIC
 calle Darwin 3
 E-28049 Madrid, Spain
 tel.  (+34) 91 585 4616
 http://wwwuser.cnb.csic.es/~mjvanraaij
  
 


 
 
 
 > On 14 Dec 2016, at 16:02, rohit kumar  wrote:
 >
 > Dear All,
 >
 > I am solving a data of 2.5 A (C121 space group). Right now the
R/Rfree values are 26/33, after many cycles of refinements (With or With
out water) the R/Rfree values still same. Zanuda suggests that the space
group seems to be correct and the model is looking  fine in coot.
 > Some one suggest what is the main problem.
 > Should I 

[ccp4bb] Antw: Re: [ccp4bb] Antw: Re: [ccp4bb] High B factor

[Matthias Barone]

In my feeling, lowering the occ explains much more such unordered loops (or 
solvent exposed sidechains..). 
B-factors are a measure of uncertainty of the atom position xyz: Letting 
B-factors fly is like broadening a distribution to an extent where the 
expectation value looses its purpose. That in mind, I have no problem reducing 
the occ of a solvent-exposed lysine to 30% or even 10% if the atoms are not 
even visible in a FEM map...
best, matthias




>>> Tim Gruene <tim.gru...@psi.ch> 17.10.16 10.06 Uhr >>>
Dear Carlos and Phoebe,

the B-value describes the movement of an atom at first order approximation, 
i.e. as a harmonic motion. It's very unlikely that an atom that is not visible 
in density, discribes a harmonic motion. 
It never occurred to me, but modelling disorder by strong B-factor restraints 
and refining the occupancy seems quite a good idea and would be a much better 
model than an explosion of B-factors.
Coot marks atoms with reduced occupancy, hence there is no extra work required 
to visualise reduced occupancy. I am aware that some people use graphics 
rendering machinery to work with PDB files instead of a dedicated program like 
Coot, but you can never suit everyone.

Best,
Tim

On Monday, October 17, 2016 09:35:02 AM Carlos CONTRERAS-MARTEL wrote:
> Even that occupancy refinement seems to be very interesting for
> crystallographers, I complete agree with Phoebe.
> 
> On 10/14/16 17:38, Phoebe A. Rice wrote:
> > Interesting way to look at it.  But those loop residues are really in
> > the crystal with an occupancy of 1, so wouldn't letting the B factor
> > fly give a clearer description of what's in the crystal?  Especially
> > as many people know to color the structure by B factors to get a feel
> > for which bits are wiggly, but they'll never think to color it by
> > occupancy.
> 
> Let them fly ... at least for protein atoms ...
> 
> Carlos
> 
> > ----
> > *From:* CCP4 bulletin board [CCP4BB@JISCMAIL.AC.UK] on behalf of
> > Matthias Barone [bar...@fmp-berlin.de]
> > *Sent:* Friday, October 14, 2016 8:00 AM
> > *To:* CCP4BB@JISCMAIL.AC.UK
> > *Subject:* [ccp4bb] Antw: Re: [ccp4bb] High B factor
> > 
> > Picking up the mail of Pavel, Phenix refines occupancies..
> > If you expect the loops to be disordered, did you try to lower the
> > occupancy of these residues, following Ethan Merritt statement that
> > "general uncertainty [...] is represented better by occupancy <1
> > rather than an arbitrary large B factor" (To B or not to B,
> > doi:10.1107/S0907444911028320).
> > If this attempt does not bring the B-factors down, it will surely make
> > the model more accurate, as the atom coordinates of the loops may not
> > be correct at all, no?
> > 
> > matthias
> > 
> > >>> Pavel Afonine <pafon...@gmail.com> 14.10.16 9.36 Uhr >>>
> > >>> 
> > If you are still worry about your Bfactor, you could try TLS,
> > 
> > Or NCS, but SA with MLHL might be better.
> > 
> > (A joke).
> 
> --
>   Carlos CONTRERAS MARTEL, Ph.D.
>   (CR1 CNRS)
> 
>   carlos.contreras-mar...@ibs.fr
> 
>   "Bacterial Pathogenesis Group"
>  Institut de Biologie Structurale
>   UMR5075 CEA-CNRS-UGA
> 
>IBS
>Campus EPN
>71, avenue des Martyrs
>CS 10090
>38044 Grenoble CEDEX 9
>FRANCE
> 
> 
>   tel : (+33) (0)4 57 42 86 41
> 
> http://www.ibs.fr/groupes/groupe-pathogenie-bacterienne/?lang=fr
> http://www.ibs.fr/groups/bacterial-pathogenesis-group/?lang=en
-- 
--
Paul Scherrer Institut
Dr. Tim Gruene
- persoenlich -
Principal Investigator
Biology and Chemistry
OFLC/102
CH-5232 Villigen PSI

Phone: +41 (0)56 310 5297

GPG Key ID = A46BEE1A

[ccp4bb] Antw: Re: [ccp4bb] High B factor

[Matthias Barone]

Picking up the mail of Pavel, Phenix refines occupancies..
If you expect the loops to be disordered, did you try to lower the 
occupancy of these residues, following Ethan Merritt statement that 
"general uncertainty [...] is represented better by occupancy <1 
rather than an arbitrary large B factor" 
(To B or not to B, doi:10.1107/S0907444911028320). 
If this attempt does not bring the 
B-factors down, it will surely make the model more accurate, as the atom
 coordinates of the loops may not be correct at all, no?

matthias



>>> Pavel Afonine  14.10.16 9.36 Uhr >>>

  If you are still worry about your Bfactor, you could try 
TLS,


Or NCS, but SA with MLHL might be better. 

(A joke). 




 

[ccp4bb] Antw: Re: [ccp4bb] Offtopic: Software to closely visualize interacting partnets in protein complex

[Matthias Barone]

Hi,
I just checked the PIC sever tim suggested. very nice indeed. If 
you only want to map different interfaces and the amino acids involved 
in, I suggest to run the pisa server, too. http://www.ebi.ac.uk/pdbe/pisa/ . I 
used it extensively to find out whether a certain set of crystal contacs leads 
to a certain crystal packing. 
best, matthias

 Tim tim.schu...@rub.de 11.03.15 18.25 Uhr 
Hi,
Molprobity is also a nice software to do this kind of analysis - if you 
use it as implemented in phenix you also get good visualization in coot.
I also asked the pymol community to create an implementation for pymol, 
but I did not follow if somebody took that up.

Also this 'protein interactions calculation' server is very neat:
http://pic.mbu.iisc.ernet.in/

/Tim


On 10/03/15 11:25, Debasish Kumar Ghosh wrote:
 Dear All,

 Apologies for this little off-topic inquiry. I want to closely visualize the 
 interacting residues in an multimeric protein complex to understand the 
 nature of interactions. Is there any good software to give this information 
 with good clarity.
 Any suggestion is highly appreciated.

 Thanks,
 Best !!!

 Debasish Kumar Ghosh

 CSIR- Junior Research Fellow (PhD Scholar)
 C/o: Dr. Akash Ranjan
 Computational and Functional Genomics Group
 Centre for DNA Fingerprinting and Diagnostics
 Hyderabad, INDIA

 Email(s): dkgh...@cdfd.org.in, dgho...@gmail.com
 Telephone: 0091-9088334375 (M), 0091-40-24749396 (Lab)
 Lab URL: 
 http://www.cdfd.org.in/labpages/computational_functional_genomics.html

[ccp4bb] Data Fitting for protein-ligand interaction.

[Matthias Barone]

Dear All
I've been reading several mails that adress the problem of acetylated N-termini 
when refining peptide ligands with refmac. I managed to include LINKR records 
after running refmacs review restraints as suggested by Eleanor Dodson in one 
of the mails I found:

LINKRC   ACE I   0 N   TRP I   1ACE_C-N

But the records themself are obviously not sufficient to maintain the ACE 
linked to TRP during refinement and ACE moves away.

Eleanor Dodson wrote about that topic:
If you run [...] review restraints, it will detect and make a LINK entry for 
you
Then you will need to use the GUI task - merge monomer library to combine your 
corrected MAL with the new LINK
Run refmac again with XYZIN the output from review restraints  task (that 
will include a LINK record) and it should/might! work...

If I read that correctly, Review restraints should produce a LIBOUT that can be 
loaded as .cif in a later step for refinement? Such a file was 

not produced.

I tried afterwards: 

1) To regularize and safe an ACE-TRP monomer in JLIgand and load the .cif as 
LIBIN for refmac. The link is found, but ACE still moves away with 

the same messages:

  WARNING : link:ACE_C-N  is found dist = 1.361 ideal_dist= 1.329
ch:II   res:   1  TRP  at:N   .-Ia   res:   0  ACE 
 at:C   .

Even though XYZOUT still contains the LINKR records.

2) As I suspected, refmac needs except for the LINKR record a LIBIN that should 
be produced when running Review Restraints. As this .cif file 

is not produced, I sent the whole ligand to prodrug and safed the .cif from 
there. But that did not work out either.

3) I also tried to change the LINKR record into LINK


I failed until now to tell refmac to maintain ACE linked to the TRP. I know 
this issues have been discussed before, but none of the suggestions helped yet. 
Any help would be highly appriciated.