if I understood your question correctly, then one of 13 scenarios described
fits your situation and you should be able to handle it in refinement all
For more specific advice please
BraggNet: integrating Bragg peaks using neural networks
B. Sullivan, R. Archibald, J. Azadmanesh, V. G. Vandavasi, P. S. Langan, L.
Coates, V. Lynch and P. Langan
J. Appl. Cryst. (2019). 52, 854-863
On Tue, Aug 3, 2021 at 4:53 AM Thorn, Dr. Andrea <
you may want to include this one:
Using support vector machines to improve elemental ion identification in
macromolecular crystal structures. Morshed N, Echols N, Adams PD Acta
Cryst. D71, 1147-58 (2015).
On Tue, Aug 3, 2021 at 4:53 AM Thorn, Dr. Andrea <
phenix.pdb.biomt_reconstruction command should do it.
On Tue, May 25, 2021 at 12:44 PM Frank von Delft <
> Hello all - this presumably has a really simple solution:
> For a PDB file with a (correct) biomolecular assembly record (REMARK
I have structure with bent DNA. I am trying to refine the structure
> using phenix. do I need to turn off the DNA secondary structure restraints
> during refinement?
probably not, unless you have reasons to do so otherwise.
P.S.: There is a Phenix mailing list for Phenix
as others pointed out electron rich elements tend to amplify imperfections
visible in Fo-Fc maps. Consider:
- refining occupancy of Fe, the site may be partially occupied;
- refining f' and f'' if data are anomalous;
- surrounding histidines may 'see' this Fe as a nonbonded interaction and
first off, there is a dedicated mailing list for Phenix specific questions (
You need to run more than one refinement cycle before you try making sense
of R factors.
On Wed, Jan 20, 2021 at 10:05 AM Wajahat ali khan
> Dear all,
Rama-Z is implemented in CCTBX meaning it is available to CCP4 and Phenix.
Also it is reported in validation and refinement in Phenix (you get the
number every time you run real- or reciprocal-space refinement!).
On Mon, Jan 18, 2021 at 1:58 PM Boaz Shaanan wrote:
> Will it be
you can do it in Phenix PDBTools: GUI->Model Tools-> load files then
Options->Other modifications look for Rename chain ID.
On Mon, Dec 7, 2020 at 9:49 AM Christian GALICIA <
> I'm trying to swap the chain IDs of a
I would choose to not do the real space refinement in phenix.refine during
> the last rounds of refinement of a model, when sidechain positions are
> essentially correct.
by design it is supposed to place and fit side chains as good as possible,
satisfying both map fit and geometry
> Hi, the question may be a bit weird, but how do you define 'over-fitting'
> in the context of structure refinement? From users' perspective the
> practical aspect is to 'fit' the model into the density. So there comes
> this question from our juniors: fit is fit, how is a model
there was a bug at some point that could potentially lead to this.
This all should be fixed in current Phenix nightly builds:
If not, get back to us (phenixbb or Phenix help lists or reply directly to
On Mon, Oct 19, 2020
in Phenix it is:
Refinement settings -> Select Atoms -> Custom Geometry Restraints : you can
define bonds, angles, torsions, planes, etc...
On Thu, Jul 2, 2020 at 8:11 AM Cristina Machon wrote:
> Dear all,
> I am writing regarding a problem we are facing with the
many of us have probably experienced that, in the
> diffraction of protein ligands containing heavy atoms (cisPt, I3C, etc),
> the overwhelming electron density of the metal can totally flatten that of
> the light atoms around (or rather make it look insignificant).
"Like comparing these map regions, excluding
intrusion of a solvent mask, etc.":
You didn't say much about the context.. So I'd say Polder map approach
comes to mind first based on these keywords. Next is "map comparison" (
If none of
phenix.match_maps can overlay model B and map B onto model A and map A. A
and B can be any symmetry and box (unit cell) dimensions. Model A and map A
stay in its original frame of reference. Let me know should you have
On Mon, May 11, 2020 at 3:19 PM Murpholino Peligro
Randy Read's paper in latest Acta D:
Measuring and using information gained by observing diffraction data
seems very relevant to this discussion!
On Fri, Mar 6, 2020 at 8:44 AM James Holton wrote:
> Thank you Kay,
Clearly, it is a good idea to keep hydrogens:
Not sure why this keeps coming up as a topic given how much it was said
about it in the past, all the MolProbity arguments, etc..
Issue of missing side chains and loops is tricker indeed.
Interesting conversation! I see the 2017 paper is on bioRxiv. I wonder if
it ever made into a peer reviewed journal (couldn't find quickly)?
@Tim Gruene : have a look at d_model in
https://www.ncbi.nlm.nih.gov/pubmed/30198894 which is sort of along similar
lines of what you are hinting here.
please make a note of upcoming Phenix workshop focusing on crystallography
and Cryo-EM tools for structure solution, August 2nd 2020 in San Diego,
California. This is a day-long satellite workshop prior to ACA meeting. For
schedule and registration see ACA 2020 web site:
did you use correct model parameterization and optimal refinement strategy
for the resolution? Such as:
- Add H atoms;
- Refine all but H atoms with anisotropic ADPs;
- Model alternative conformations (that one'd expect many at this
- Add solvent (water, crystallization
a possible programmatic approach may be a loop over all (model, map,
resolution) or (model, map, half_map1, half_map2, resolution) from PDB/EMDB
phenix.validation_cryoem model. emdb_.map resolution=value
phenix.validation_cryoem model. emdb_.map
Phenix reads and writes TLS records, both PDB and mmCIF format. It can also
read (but not write) TLS records in REFMAC and BUSTER format.
In ATOM records Phenix outputs complete B factors, which includes both
individual and TLS components (this is why they have ANISOU).
Yet another way is:
phenix.superpose_pdbs fixed.pdb moving.pdb selection_fixed="chain A and and
resseq 1:10 and name CA" selection_moving="chain B and resseq 1:10 and name
or using the GUI.
On Tue, Sep 17, 2019 at 8:06 AM Folmer Fredslund wrote:
> Dear Kyle,
> As other non-CCP4
for obvious issues like this one!
On Fri, Aug 30, 2019 at 7:59 AM Pavel Afonine wrote:
> Please send me log file off-list and that may be a start. -Thanks!
> On Fri, Aug 30, 2019 at 6:48 AM Tung Thanh Dinh wrote:
>> After phasing with phenix, i clicked on the
Please send me log file off-list and that may be a start. -Thanks!
On Fri, Aug 30, 2019 at 6:48 AM Tung Thanh Dinh wrote:
> After phasing with phenix, i clicked on the phenix.refine ribbon. Since
> then for every macrocycle of refinement, r-work and r-free values are
> always the same. Is
> My conclusion is that "no fill-in" option might be tried at some
> stages, but with caution, especially for datasets with poor low res
I'm guessing you really meant *high*, not low. More or less repeating what
others said already.. Correcting for low res
For the record, phenix.refine always produces two versions of 2mFo-DFc
maps, with and without filling in missing Fobs. The one that opens in Coot
by default is the "filled" map, but you can always load the other one for
On Tue, Aug 6, 2019 at 8:34 AM Ivan Shabalin
Hard to see from a static image, but could it be an alternative
On Tue, Jul 9, 2019 at 2:32 AM Lumbini Yadav wrote:
> Dear all,
> We have found a huge Fo-Fc density close to cysteine residue (see attached
> image) in the structure with resolution of 1.2A. In the
Hi Sam Tang,
Sorry for a naive question. Is there any circumstances where one may wish
> to refine to a lower resolution? For example if one has a dataset processed
> to 2 A, is there any good reasons for he/she to refine to only, say 2.5 A?
yes, certainly. For example, when information
send me files off list and I will have a look. From your description it
isn't clear to me what the problem is. You need to add H or D or H and D
only once (and phenix.ready_set is the right tool to do it!), then just do
refinement and all should work. Coot indeed may not play well
this may also be a result of too strong SS bond restraints or/and
inaccurate SS bond restraints parameters or/and disorder (in some fraction
of unit cells there is no SS bond). More on the topic:
"Disulfide bond restraints" here:
here is the link to cctbx-based code that computes Uhkl according to your
formula below, using model mean B and F000 that accounts for atomic model
I leave it up to you to read and understand
> - Is there any macromolecular crystallography software that can compute
> Uhkl as above, or equivalent?
I estimate this can take about 10 minutes to script in CCTBX. I can write a
script for you, if interested, and send off list.
> - If not, would it be more correct to use the
- How can F(000) be best estimated from the final model, which is not
> necessarily always the most complete or best refined? Should we simply add
> together the number of electrons for all the atoms refined in the
> asymmetric unit (protein + ligands + solvent)?
the text here
perhaps there are as many answers to this as many subscribers to this list,
but personally "Cysteine with attachment" seems more logic and clear to me
than calling the whole thing a different name. Although I would also
understand arguments like if it is a CYS with an attachment it is not
Also, values in parentheses (high-res shell) depend on how you (the program
you use) do binning. Different programs do it differently and so these
values can vary quite substantially. With a little of trial-and-error
effort choosing the binning one can make these values farther or closer to
P.S.: while on the topic, it might be helpful to you to have a look at this
article ("On the analysis of residual density distributions on an absolute
scale", page 43) available here:
On Tue, Mar 5, 2019 at 10:02 AM Pavel Afon
As Pavel mentioned, phenix.f000 will give you F(0,0,0) value, but I don't
> see that information stored and easily calculated from .ccp4 data.
phenix.f000 requires atomic model (PDB or mmCIF file) as input, not a map.
If you want bulk-solvent to be added, then you need to give it mean
> Unfortunately not all structure
> factor programs will give you that F000.
phenix.f000 will give you F(0,0,0) value based on atomic model alone or
atomic model plus bulk-solvent.
To unsubscribe from the
> Or, alternatively, if anyone has used a different program to refine small
> molecule structures determined by ED, we would be happy to hear about that
> program too.
phenix.refine has an option to use electron scattering factors. I can
provide further assistance off-list or on phenix
this is why Polder map tool also includes analysis of the map in question
to determine whether it looks like bulk-solvent or something else, as
described in paragraph 5 here:
This analysis tells you in plain English what you are
In the current version of phenix.refine contains the separate option for
> refinement: xyz (reciprocal space) and xyz (real space ). What does it
> mean and how it differs from the previous versions which had xyz and real
> space instead.
All the same, different name. One refines
> Any time you do a thought experiment you make a fake-data data set, the
> "true" phases and "true" amplitudes become the ones you put into the
> simulation process. This is by definition. Is there potential for
> circular reasoning? Of course! But you can do controls:
this is so
I can only guess without seeing files (and link to files seems to be
broken).. So my guess is that the ligand density is weak enough so FEM
treats it as "near noise" and it wipes it. Polder decides it is likely
ligand because it uses correlations, and so even of two maps are weak but
I wonder to which extent I can trust the positive blobs or polder maps that
> I am generating...
the answer to this question is given in corresponding paper that describes
Based on the analysis described
Now I see the value of storing data in plain text files even more (mind
Shelx or X-plor formats, for example) -;)
On Fri, Nov 9, 2018 at 9:47 PM Clemens Vonrhein
> Hi Eleanor,
> You could try running the oldest MTZ2VARIOUS binary you can find -
phenix.pdbtools model.pdb rotate=... translate=...
which should work with any PDB or mmCIF file.
On Thu, Nov 8, 2018 at 11:36 PM Tim Gruene wrote:
> Dear Kaushik,
> you could try moleman2 from the Uppsala Software Factory,
> http://xray.bmc.uu.se/usf/moleman2_man.html - maybe
I'm guessing this is because it isn't clear what CCwork/CCfree can tell you
that Rwork/Rfree can not. Needless to say we all more or less have a good
idea about what the ok values for Rwork, Rfree and Rfree-Rwork (as function
of resolution) while it is much less clear (to me at least)
I always feel some people are too "greedy" with the resolution they want to
> achieve. I mostly find that extremely high density is a pain to work with
> as it's usually accompanied by many dual, triple conformers, a lot of noise
> in the solvent phase that is often difficult to
how complete the data set? What's completeness across resolution zones?
Systematically missing reflections can have systematic impact in real space
(maps). I've seen entire ligands or domains disappear due to missing
Just another check-point to consider..
Nonbonded interactions? (if you approach this from classic geometry
restraints used in refinement programs)
On Mon, Sep 17, 2018 at 2:09 PM Joel Tyndall
> Polar interactions seems to make the most sense. This is what Pymol uses
> as I don’t think it differentiates
P.S.: all questions are welcome of course, no labeling. It's just some of
them are so orthogonal to common sense that answers my be such as well.
On Sun, Sep 9, 2018 at 9:38 PM Pavel Afonine wrote:
> Is there any sever available to create electron density maps fo
Is there any sever available to create electron density maps for cryo-em
The questions are nonsensical. Here is why:
1) In cryo-EM maps are not electron density maps but surfaces representing
2) Creating such a map is essentially carrying on from cryo-EM
> is there some kind of general trend if not consensus as how to refine
> cryo-EM structures?
Not really consensus, but my clearly biased contribution to the topic:
I'd say basic common sense for refinement applies:
- optimally sharpen the map (
also, it helps to keep in mind that some clashes may actually be valid
interactions that are labeled as 'clashes' by validation software that is
simply not sophisticated enough to distinguish between bad steric clashes
and chemically/physically favorable valid interactions. For an example,
> I (personally) think the best answer from these was to look at the
> TLS-subtracted residuals (ie. total B-factor - TLS component) — can’t
> remember who sent it, off the top of my head.
TLS is just an approximation, sometimes good and sometimes not. If TLS
parameters are refined along with
There is an option in phenix.refine to do this, described here:
Automated identification of elemental ions in macromolecular crystal
structures. Echols N, Morshed N, Afonine PV, McCoy AJ, Miller MD, Read RJ,
Richardson JS, Terwilliger TC, Adams PD Acta Cryst. D70, 1104-1114 (2014).
Phenix has its own way to do it which is "Comprehensive validation"
available from the GUI. This includes validation of model, data and
model-to-data fit. It is data-specific: there is one for crystallography
(X-ray or neutron) and one for Cryo-EM. For best results, you need the
As others suggested, you can use Polder map:
- how-to video tutorial can be found here:
- background is described here:
On Tue, Jul 24, 2018 at
It's important to remember that free-R reflections are not only used to
calculate Rfree, but also are used in calculation of m and D scales in
2mFo-DFc and mFo-DFc maps, as well as in likelihood-based refinement
targets. The fact is that you need to have a sufficient amount of free-R
This should work in latest nightly builds:
phenix.pdbtools model.pdb clear_seg_id=true
If it doesn't please report a bug (on appropriate Phenix lists).
On Thu, Jul 5, 2018 at 4:33 AM, Eugene Osipov wrote:
> Hello everyone,
> is there any simple way in CCP4 to clear segid
More re disulfide bonds:
On Tue, Jul 3, 2018 at 11:26 PM, 张士军 <21620150150...@stu.xmu.edu.cn> wrote:
> Hi all
> I got a structure which has COA in it, and the SH in the tail of COA
> is very close to the SH side chain of
> If I am not wrong, I remember that someone proposed to standardize
> B-factors of protein atoms as “BS = B - Bave”, where Bave is the average
> B-factor of the protein.
This will make some of BS negative (if B
Just in case you find it helpful, you can get 100% complete set of
reflections (Fcalc) in specified resolution range using
phenix.fmodel model.pdb high_res=2.5 low_res=15
or if you leave out low_res it will go all the way up to theoretical limit
of low resolution.
If you have/use cctbx then I
This is discussed, for example, here:
Also, here I calculated the distribution of map values (scaled in r.m.s.)
for four groups of atoms: main-chain atoms, side-chain oxygen atoms of ASP
and GLU (negatively charged OD1, OD2, OE1, OE2), side chain atoms of
> I am new to building into an EM map, and I wonder if I could get a
> recommendation for some standard routine for programs to use. I tried "find
> helixes and strands" tool in phenix, it found some secondary structure
> elements, however, it seems that there is more that could be done
It's amusing how a seemingly innocent ad for a new tool can ignite a rather
prickly thread.. I see two keys to this.
Firstly, for those who are not familiar with the issue the add could be
better structured by providing a clearer statement of what the problem is
or why it is important (with
Perhaps this can be automated:
Software doesn't get tired or bored, and thus potentially can try more and
produce more plausible interretations. Then one can hire a number of people
of various expertise to choose "best" result according to
R-factor value is almost useless unless you know the resolution (which you
did not tell us): 26% is ok for 3A resolution and is nonsense for 1A, for
example. The Rfree-Rwork gap is obviously large, suggesting sub-optimal
refinement strategy. Try optimizing weights, let program update
See Table 1 and corresponding discussion here:
Hope that hints you the answer. If not get back to me with questions.
All the best,
On Mon, Nov 6, 2017 at 7:18 PM, Eze Chivi wrote:
> Hi, my PDB file
If by "it was truncated at 5A for clarity" you really mean you truncated
all low-resolution data from 5A and lower then I am not surprised you see
funny densities all over or don't see density where it is expected. Why?
Consult a textbook for the answer.
All the best,
On Fri, Nov 3, 2017
core functionality described in that paper is implemented in cctbx. Re the
stand-alone program -- I'm cc'ing to the author.
On Thu, Oct 26, 2017 at 8:39 AM, Hattne, Johan
> Dear all;
> Would anybody know where I can find efresol (as detailed in
A round of refinement with simulated annealing followed by minimization
should address your concern.
On Wed, Oct 11, 2017 at 4:48 PM, Karsten Dreifus
> Dear all,
> I have a 120 aa protein. Matthews coefficient indicates 3 mol in asu.
> Molrep (template with
And I should add this works just great! (given we are on the same page
I used this for Cryo-EM model challenge; Nigel added this functionality at
that time to make this possible.
On Wed, Oct 11, 2017 at 2:17 AM, Nigel Moriarty wrote:
> Since you
Or using cctbx:
from scitbx.array_family import flex
xyz_1 = iotbx.pdb.input(file_name="file_1.pdb").atoms().extract_xyz()
xyz_2 = iotbx.pdb.input(file_name="file_2.pdb").atoms().extract_xyz()
print flex.mean(flex.sqrt((xyz_1 - xyz_2).dot()))
also keep in mind that the total model structure factor used in refinement
and anywhere where model-to-data agreement needs to be evaluated (such as
maps or R factors) is:
Fmodel = ktotal * (Fcalc_atoms + F_bulk_solvent + F_something_else)
where ktotal ~ scale * exp(-h*Uoverall*h_transpose)
your suggestion makes perfect sense to me, and it's trivial to add an
option to do what you want. This will be available in next Phenix nightly
build (clearly not tomorrow given today's power outage).
Command line: use "write_map_coefficients_only=True" (by default is is
> I know space is cheap these days, but is there a reason for Refmac to
> generate all those extra columns in the output mtz file? Refmac (as well
> as phenix.refine and buster-tnt) output mtz file is almost always used for
> only one purpose - look at the map in coot. You only need 4 columns
phenix.refine may not use reflection-outliers (Read, R. J. (1999). Acta
Cryst. D55, 1759–1764.). Typically this is just a few reflections. If you
have a good reason to disable this, then use
P.S.: There is Phenix mailing list for Phenix-related
it's not a problem if you use mmCIF or PDB with two-letter chain ID (both
supported in Phenix).
On Mon, Jul 24, 2017 at 5:09 PM, Lijun Liu wrote:
> Hi: this must be an old problem but I would like to know if there are
> other ideas to make things easier.
thanks for sharing the data (off-list). I did some detective work and yes,
Clemens is correct: what you see is the effect of bulk-solvent. After
adjusting the solvent contribution (mask) in regions occupied by altlocs A
and B the positive density mostly disappears.
On Fri, Jul 21,
I agree with Dominika, I can't see major problems with this entry (I also
did some quick refinement and briefly looked at maps). Reported R factors
match re-calculated values using data from PDB, which is good.
Smaller issues I see are:
- no solvent (water) in the model, while density suggests
try without "metal restraints" and see if that helps. As others suggested,
make sure 2+ is present in rightmost column of PDB file. The side may be
partially occupied, so refining occupancy of Mg2+ is not a bad idea.
On Wed, Jun 14, 2017 at 2:44 PM, Mohammad Rahman
Yes, that is what I have been doing. Build one subunit and assemble into
> tetramer before realspace refinement (with "ncs" constraints). I used
> tetramer for refinement because I want the distances between the inter
> subunits interaction partners to be considered. The problem is, whenever I
Just use P1 and "ncs" constraints. What's the problem? Or just keep entire
map and have only symmetry independent copy to work with until finishes,
then make the whole molecule. For real-space refinement it's totally
irrelevant whether you have whole molecule or 1/Nth of it. So.. it isn't
2. How do I convert cryoEM map file to MTZ file?
While technically you can do it, normally there is absolutely no need to do
it. In cryo-EM the map is your data, not reflection data (structure
factors!). So no need to 'massage' your data (the map) by converting it
into "Fobs" and storing
I'm working on a crystal structure with resolution of 2.2A. At the final
> step, I use different strategies to refine the structure, they are:
> no4: strategy=individual_sites+individual_adp+tls / set_b_iso=20
> results: Rwork/free=0.2052/0.2658 b-factor=11.4/136.8/48(min/max/average)
I am sure others
> are certain to propose a cooler name for that very same type of map
> some day ;-) .
a free tip: how about DDM (Decarboxylation Detector Map)?
All the best,
Also, see figure 4 here:
that illustrates the difference.
On Fri, Apr 28, 2017 at 10:33 AM, Bernhard Rupp
> Dear Fellows of the Bond,
> when validating a QM refined homology model with Molprobity, I
Note, Molprobity has an option to use longer (neutron) X-H distances.
On Fri, Apr 28, 2017 at 11:46 AM, Tristan Croll wrote:
> I believe the reason for the discrepancy here is that MolProbity by
> default places the hydrogens according to the centroid of the electron
I suspect most (if not all) refinement software now have a nice way to deal
with NCS in refinement locally. Technically, this means you can use NCS
restraints at any resolution and software should be able to be careful and
not wipe out local differences between NCS copies. In phenix.refine this is
Minimizing a red bar may be tricky.. Have you tried to make it less red
(blue or may be green)? Otherwise Rw/Rf~20/25 is just fine at 2.2A
resolution. What exactly your worry is about?
On Mon, Apr 24, 2017 at 12:48 AM, Vipul Panchal
> Hi all,
> I am solving
Except that Procheck is now ages behind the standard, with Molprobity being
the standard. I'm not sure even if Coot uses the latest libraries. Those I
quoted in example below come from latest Molprobity (Phenix that is) and
are the latest.
On Fri, Mar 24, 2017 at 2:55 PM, Edward A. Berry
phi and psi? for
> example :
> A 2 ASN:56.93:-60.58:141.19:Favored:General alpha helix
> A 3 ASN:48.44:-119.25:125.15:Favored:General alpha helix
> On Fri, Mar 24, 2017 at 1:09 PM, Pavel Afonine <pafon...@gmail.com> wrote:
>> Trivial using command
(1) best practices in refining against lower resolution data (~4 angstrom)
> to achieve the best model,
obtain a model that fits data best under requirement that it has zero
geometry violations (Ramachandran, Cbeta deviations, rotamers, CABLAM,
Note, a geometry outlier
Normally, these days at least, a model that is result of TLS refinement
contains total B factor in ANISOU records and its TLS component in TLS
records (REMARK3), with Btotal = Btls+Bresidual.
If TLS matrices are available, it's trivial to calculate Btls from TLS
matrices in REMARK3 and subtract
Is there a straight-forward way to estimate the amount of missing electron
> density that a particular protein structure is missing based on the
> difference between Fo and Fc?
Any refinement or map calculation software that uses likelihood-based
approach does this routinely. In
> economical in its acknowledgment of "prior art" - a notion that surely
> has to be recognised as existing outside the confines of standard,
> neatly packaged, immediately quotable Acta D publications :-) .
> With best wishes,
> On Th
In addition to excellent Kay's reply..
Also make sure to check refined B factors. Note, if the ligand is not there
then that volume is likely filled with bulk-solvent. Now low occupancy in
combination with very large B factors may approximate bulk-solvent quite
well. The Polder map along with
1 - 100 of 562 matches
Mail list logo