Phenix.reflection_file_converter enhancement request
Hi, would it be possible to enhance the reflection file converter so that it can simultaneously write out IOBS and FOBS into the same mtz file? Right now these options seem to be mutually exclusive. The automatic conversion in phenix.refine handles that task automatically, so it should not be too hard to implement. Thanks Carsten
Hi Carsten, Nat's reflection file editor in the GUI is meant to give you full flexibility in composing a new mtz file. It is tricky to make everything people want to do work from the command line. -- Nat was working on command-line support for some of the reflection file editor functionality, but I'm not quite sure where that stands. Nat is out-of-town and may not be able to respond quickly in the next few days. If you have a specific recurring mtz conversion task you need to automate, a small custom Python script may be the best answer. If you send us example inputs and explain expected outputs we can try to help. Ralf
On Tue, Apr 13, 2010 at 12:07 PM, Ralf W. Grosse-Kunstleve wrote: Nat's reflection file editor in the GUI is meant to give you full
flexibility in composing a new mtz file. . . . however, the current version of the GUI can't make multiple copies of
a single data array, unless you fool it by making a copy of the input file
and add that. There is no reason why it can't be modified to allow this - I
just need to figure out how to keep the interface simple.
It is tricky to make everything people want to do work from the command line. -- Nat was working on command-line support for some
of the reflection file editor functionality, but I'm not quite
sure where that stands. Functional, but not fully tested. I'm pretty sure it will do what you need
here (as long as you don't mind output in MTZ format):
iotbx.reflection_file_editor data.sca dry_run=True
This will print out (among other info) the resulting parameter set, which
you can paste into a file and edit to do exactly what you want.
(Specifically, duplicate the block of parameters for the intensity array,
and choose "amplitudes" for "output_as" in the second copy.)
-Nat
Hi Nat,
I tried that and it chokes on a dtrek reflection file. It doesn't like the I(+) notation, when the output contain I+ and I- separated
Here is the header of the file, which is standard output from dtrek.
3 6 0 5
CRYSTAL_MOSAICITY=1.3945 0.0000 0.0000;
CRYSTAL_ORIENT_ANGLES=101.2180 6.9156 -21.4276;
CRYSTAL_SPACEGROUP=5;
CRYSTAL_UNIT_CELL=98.6333 98.0339 132.5167 90.0000 111.4590 90.0000;
SOURCE_WAVELENGTH= 1.0000 1.0000;
nH
nK
nL
fIntensity
fSigmaI
fIntensity+
fSigmaI+
fIntensity-
fSigmaI-
0 0 3 18.3045 5.10113 18.3045 5.10113 -1.00000 -1.00000
0 0 4 5710.08 188.795 5755.10 269.073 5666.41 264.967
0 0 5 8143.22 236.253 8604.30 351.377 7762.81 319.165
0 0 6 125.538 22.1897 113.212 28.2108 145.538 35.9342
0 0 7 7203.69 246.220 7335.97 353.565 7079.14 343.087
....
From: [email protected] [mailto:[email protected]] On Behalf Of Nathaniel Echols
Sent: Tuesday, April 13, 2010 3:40 PM
To: PHENIX user mailing list
Subject: Re: [phenixbb] Phenix.reflection_file_converter enhancement request
On Tue, Apr 13, 2010 at 12:07 PM, Ralf W. Grosse-Kunstleve
On Wed, Apr 14, 2010 at 9:46 AM, Schubert, Carsten [PRDUS] < [email protected]> wrote:
I tried that and it chokes on a dtrek reflection file. It doesn't like the I(+) notation, when the output contain I+ and I- separated Here is the header of the file, which is standard output from dtrek.
Sorry, I just discovered this myself yesterday. It should be fixed in last night's build (dev-373). -Nat
Hello when I use phenix.refine to refine my structure, the rmsd's of bonds and angles are pretty large. My data is 2.3 A resolution. REMARK REMARK ******************** REFINEMENT SUMMARY: QUICK FACTS ******************* REMARK Start: r_work = 0.2838 r_free = 0.2789 bonds = 0.029 angles = 2.002 REMARK Final: r_work = 0.2407 r_free = 0.2466 bonds = 0.029 angles = 2.002 REMARK ************************************************************************ REMARK I tried many options but none worked so far. Here are the options that I tried: phenix.refine mtz_file pdb_file strategy=tls tls_group_selections.params wxc_scale=0.3 wxu_scale=0.8 phenix.refine mtz_file pdb_file strategy=tls tls_group_selections.params fix_wxc=0.3 fix_wxu=0.8 phenix.refine mtz_file pdb_file strategy=tls tls_group_selections.params fix_wxc=0.3 fix_wxu=0.8 optimize_wxc=false optimize_wxu=false These gave the same refinement results. Any suggestions? Thanks Rongjin Guan
Hi Rongjin,
when I use phenix.refine to refinemy structure, the rmsd's ofbonds and angles are pretty large. My data is 2.3 A resolution. REMARK REMARK ******************** REFINEMENT SUMMARY: QUICK FACTS ******************* REMARK Start: r_work = 0.2838 r_free = 0.2789 bonds = 0.029 angles = 2.002 REMARK Final: r_work = 0.2407 r_free = 0.2466 bonds = 0.029 angles = 2.002 REMARK ************************************************************************ REMARK
really (just wondering)? What I would be really worry is the difference between Rfree and Rwork, which apparently is terribly small, indicating that there is something wrong with either free-R flags or refinement strategy.
I tried many options but none worked so far. Here are the optionsthat I tried: phenix.refine mtz_filepdb_file strategy=tls tls_group_selections.paramswxc_scale=0.3 wxu_scale=0.8 phenix.refine mtz_filepdb_file strategy=tls tls_group_selections.paramsfix_wxc=0.3 fix_wxu=0.8 phenix.refine mtz_filepdb_file strategy=tls tls_group_selections.paramsfix_wxc=0.3 fix_wxu=0.8 optimize_wxc=false optimize_wxu=false These gave the same refinement results. Any suggestions?
It's not surprising, because in all refinements above you did not refine the coordinates at all. The keyword "strategy=tls" means you refined B-factors only (more precisely, you did constrained anisotropic refinement, called TLS refinement). And therefore using wxc_scale or fix_wxc or optimize_wxc would not make any difference. Also, optimize_wxc is set to False by default, so you don't need to explicitly say it. I would run it like this: phenix.refine mtz_file pdb_file strategy=tls+individual_adp+individual_sites tls_group_selections.params optimize_wxc=true and hope that this fixes the "problem". Let me know if you have any questions! Pavel.
Hi Pavel Thanks for your suggestion. I have another question: I am refining another 3.3 A resolution structure. In coot I manually adjusted the model to have >95% residues in preferred region in ramanchandran plot, but after refinement in Phenix only <80% residues in the preferred region and a lot more in the outlier region. This was wen I used the default refinement parameters "phenix.refine mtz_file pdb_file". R and R_free are 0.22/0.27 in this case. Then when I used "phenix.refine mtz_file pdb_file wxc_scale=0.1 wxu_scale=0.3", I can get better results in terms of ramanchandran plot (say, ~92% in preferred region, and a little more in the outlier). R and R_free factors are 0.25/0.30. I guess at low resolution (like in my case, 3.3 A) I have to give more weights to geometry restrains. Did I do it in the right way, or did I missing anything? Thanks Rongjin Guan = = = = = = = = On 2010-04-13 16:42:21 You wrote = = = = = = = = Hi Rongjin, when I use phenix.refine to refine my structure, the rmsd's of bonds and angles are pretty large. My data is 2.3 A resolution. REMARK REMARK ******************** REFINEMENT SUMMARY: QUICK FACTS ******************* REMARK Start: r_work = 0.2838 r_free = 0.2789 bonds = 0.029 angles = 2.002 REMARK Final: r_work = 0.2407 r_free = 0.2466 bonds = 0.029 angles = 2.002 REMARK ************************************************************************ REMARK really (just wondering)? What I would be really worry is the difference between Rfree and Rwork, which apparently is terribly small, indicating that there is something wrong with either free-R flags or refinement strategy. I tried many options but none worked so far. Here are the options that I tried: phenix.refine mtz_file pdb_file strategy=tls tls_group_selections.params wxc_scale=0.3 wxu_scale=0.8 phenix.refine mtz_file pdb_file strategy=tls tls_group_selections.params fix_wxc=0.3 fix_wxu=0.8 phenix.refine mtz_file pdb_file strategy=tls tls_group_selections.params fix_wxc=0.3 fix_wxu=0.8 optimize_wxc=false optimize_wxu=false These gave the same refinement results. Any suggestions? It's not surprising, because in all refinements above you did not refine the coordinates at all. The keyword "strategy=tls" means you refined B-factors only (more precisely, you did constrained anisotropic refinement, called TLS refinement). And therefore using wxc_scale or fix_wxc or optimize_wxc would not make any difference. Also, optimize_wxc is set to False by default, so you don't need to explicitly say it. I would run it like this: phenix.refine mtz_file pdb_file strategy=tls+individual_adp+individual_sites tls_group_selections.params optimize_wxc=true and hope that this fixes the "problem". Let me know if you have any questions! Pavel.
Hi Rongjin,
I am refining another 3.3 A resolution structure. In coot I manually adjusted the model to have >95% residues inpreferred region in ramanchandran plot, but after refinement in Phenix only <80% residues in the preferred region and a lot more in the outlier region. This was wen I used the default refinement parameters "phenix.refine mtz_file pdb_file". R and R_free are 0.22/0.27 in this case.
you may want to consider: - using secondary structure restraints; - tightening restraints (using wxc_scale=...); - using H atoms in refinement; - doing some quick geometry regularization before refinement; - fixing manually and excluding from refinement certain parts of your model. You need to experiment and find empirically which combination of the above options works in your case. I would appreciate if you share your experience with this and tell us what worked for your in the end. Pavel.
Thank you very much for your suggestions. I will try all these options and let you know the results. Rongjin = = = = = = = = On 2010-04-14 12:03:10 You wrote = = = = = = = = Hi Rongjin, I am refining another 3.3 A resolution structure. In coot I manually adjusted the model to have >95% residues in preferred region in ramanchandran plot, but after refinement in Phenix only <80% residues in the preferred region and a lot more in the outlier region. This was wen I used the default refinement parameters "phenix.refine mtz_file pdb_file". R and R_free are 0.22/0.27 in this case. you may want to consider: - using secondary structure restraints; - tightening restraints (using wxc_scale=...); - using H atoms in refinement; - doing some quick geometry regularization before refinement; - fixing manually and excluding from refinement certain parts of your model. You need to experiment and find empirically which combination of the above options works in your case. I would appreciate if you share your experience with this and tell us what worked for your in the end. Pavel.
Hi Jason,
Just a thought, but running the default "phenix.refine" command is going to refine individual atomic coordinates and ADPs, unless you specify otherwise.
where ADPs can be refined as individual isotropic or individual anisotropic. Plus, it will refined occupancies in some cases: http://www.phenix-online.org/documentation/refinement.htm#anch20
At 3.3A are you sure you have enough data to model all of these parameters?
Most of the time at 3.3A you can refine individual ADPs and coordinates. Just use tighter restraints. Of course it's case dependent and in each case you need to see what would work best.
Even with heavy restraints, i'm not sure individual coordinate refinement is feasible.
Why? I've see such a case only once, when refining against 60% complete neutron data at 3A resolution where the model contained all H/D explicitly present (not riding) -:)
Perhaps doing a rigid-body refinement of multiple groups with grouped ADPs would yield better results. You could even model in TLS groups.
It is always good to use TLS, at 1.7A or at 6A resolution (slides 30-36): http://www.phenix-online.org/presentations/neutron_japan_2009/phenix_japan_p... Pavel.
participants (6)
-
guanrj＠foxmail.com -
Jason C Porta -
Nathaniel Echols -
Pavel Afonine -
Ralf W. Grosse-Kunstleve -
Schubert, Carsten [PRDUS]