weights optimization, add hydrogens
I am refining a 2.0 A structure in Phenix using mostly the graphical interface. The maps are very good and the R factors are low (0.208/0.173). But the geometry is not so great with bond deviations of 0.208 and Angle deviations of 1.85. What is the best thing to do to try to improve the geometry. Should I try to optimize the weights? The other question is about adding hydrogens to the model. I did add them in a previous refinement cycle. After rebuilding the model in coot, most of the model still has the hydrogens in the input file for the next refinement cycle, but some residues and waters don't have hydrogens. When I tried to add hydrogens again to the model, I got an error message about atoms being to close. Does Phenix know where to add hydrogens if the input model has already hydroegns on many residues? Thanks Ursula -- Ursula Schulze-Gahmen, PhD. QB3, Tjian Lab MCB, 16 Barker Hall #3204 University of California Berkeley Berkeley, CA 94720-3204 Phone: (510) 642 8258 [email protected]
On Feb 23, 2010, at 10:19 AM, Ursula Schulze-Gahmen wrote:
The other question is about adding hydrogens to the model. I did add them in a previous refinement cycle. After rebuilding the model in coot, most of the model still has the hydrogens in the input file for the next refinement cycle, but some residues and waters don't have hydrogens. When I tried to add hydrogens again to the model, I got an error message about atoms being to close. Does Phenix know where to add hydrogens if the input model has already hydroegns on many residues?
In theory it should, but the underlying problem is almost certainly incompatible names for hydrogen atoms, which will appear invisible to Reduce (the program that adds them). This usually results from switching between programs that have different conventions for hydrogen naming (e.g. Coot and Phenix), depending on which version of the PDB format they use. (Phenix uses PDB v3.2, which is very new.) Would you mind sending me (not the list!) the PDB file so I can double-check this? Unfortunately I don't have any suggestions for how to fix this other than wiping out the old hydrogen atoms (I like PyMOL for this: "remove elem h"). Since this keeps coming up, we'll add an option to the GUI to erase old hydrogen atoms before running the automatic addition. thanks, Nat ------------------- Nathaniel Echols Lawrence Berkeley Lab 510-486-5136 [email protected]
Dear Ursula, I usually remove the hydrogens and add them again in the next refinement cycle. Something strange happens If you simply try to update the hydrogens. I use phenix.pdbtools to remove the hydrogens. Also Coot cannot handle the hydrogens properly in real space refinement. At least in my version the Hydroges from the refined residue got to everywhers in any density. Best regards Christian Am Dienstag 23 Februar 2010 19:33:03 schrieb Nathaniel Echols:
On Feb 23, 2010, at 10:19 AM, Ursula Schulze-Gahmen wrote:
The other question is about adding hydrogens to the model. I did add them in a previous refinement cycle. After rebuilding the model in coot, most of the model still has the hydrogens in the input file for the next refinement cycle, but some residues and waters don't have hydrogens. When I tried to add hydrogens again to the model, I got an error message about atoms being to close. Does Phenix know where to add hydrogens if the input model has already hydroegns on many residues?
In theory it should, but the underlying problem is almost certainly incompatible names for hydrogen atoms, which will appear invisible to Reduce (the program that adds them). This usually results from switching between programs that have different conventions for hydrogen naming (e.g. Coot and Phenix), depending on which version of the PDB format they use. (Phenix uses PDB v3.2, which is very new.)
Would you mind sending me (not the list!) the PDB file so I can double-check this? Unfortunately I don't have any suggestions for how to fix this other than wiping out the old hydrogen atoms (I like PyMOL for this: "remove elem h"). Since this keeps coming up, we'll add an option to the GUI to erase old hydrogen atoms before running the automatic addition.
thanks, Nat
------------------- Nathaniel Echols Lawrence Berkeley Lab 510-486-5136 [email protected]
_______________________________________________ phenixbb mailing list [email protected] http://phenix-online.org/mailman/listinfo/phenixbb
-- Christian Roth Institut für Bioanalytische Chemie Biotechnologisch-Biomedizinisches Zentrum Fakultät für Chemie und Mineralogie Universität Leipzig Deutscher Platz 5 04103 Leipzig Telefon: +49 (0)341 97 31316 Fax: +49 (0)341 97 31319
phenix.reduce does generate "old-style" hydrogens if the proper options are used, ie. phenix.reduce in.pdb -BUILD -OLDpdb > out.pdb The hydrogens are compatible with coot v0.5 and v0.6 for hydrogens attached to standard protein residues. I agree that stripping the hydrogens and regeneration after rebuilding/mutation and prior to refinement is the safest path for now until better support is implemented. Cheers, Carsten
-----Original Message----- From: [email protected] [mailto:phenixbb- [email protected]] On Behalf Of Christian Roth Sent: Tuesday, February 23, 2010 5:26 PM To: [email protected] Subject: Re: [phenixbb] weights optimization, add hydrogens
Dear Ursula,
I usually remove the hydrogens and add them again in the next refinement cycle. Something strange happens If you simply try to update the hydrogens. I use phenix.pdbtools to remove the hydrogens. Also Coot cannot handle the hydrogens properly in real space refinement. At least in my version the Hydroges from the refined residue got to everywhers in any density.
Best regards
Christian
On Feb 23, 2010, at 10:19 AM, Ursula Schulze-Gahmen wrote:
The other question is about adding hydrogens to the model. I did add them in a previous refinement cycle. After rebuilding the model in coot, most of the model still has the hydrogens in the input file for the next refinement cycle, but some residues and waters don't have hydrogens. When I tried to add hydrogens again to the model, I got an error message about atoms being to close. Does Phenix know where to add hydrogens if
input model has already hydroegns on many residues?
In theory it should, but the underlying problem is almost certainly incompatible names for hydrogen atoms, which will appear invisible to Reduce (the program that adds them). This usually results from switching between programs that have different conventions for hydrogen naming (e.g. Coot and Phenix), depending on which version of the PDB format they use. (Phenix uses PDB v3.2, which is very new.)
Would you mind sending me (not the list!) the PDB file so I can double-check this? Unfortunately I don't have any suggestions for how to fix this other than wiping out the old hydrogen atoms (I like PyMOL for this: "remove elem h"). Since this keeps coming up, we'll add an
Am Dienstag 23 Februar 2010 19:33:03 schrieb Nathaniel Echols: the option
to the GUI to erase old hydrogen atoms before running the automatic addition.
thanks, Nat
------------------- Nathaniel Echols Lawrence Berkeley Lab 510-486-5136 [email protected]
_______________________________________________ phenixbb mailing list [email protected] http://phenix-online.org/mailman/listinfo/phenixbb
-- Christian Roth Institut für Bioanalytische Chemie Biotechnologisch-Biomedizinisches Zentrum Fakultät für Chemie und Mineralogie Universität Leipzig Deutscher Platz 5 04103 Leipzig Telefon: +49 (0)341 97 31316 Fax: +49 (0)341 97 31319 _______________________________________________ phenixbb mailing list [email protected] http://phenix-online.org/mailman/listinfo/phenixbb
Hi Ursula,
I am refining a 2.0 A structure in Phenix using mostly the graphical interface. The maps are very good and the R factors are low (0.208/0.173). But the geometry is not so great with bond deviations of 0.208 and Angle deviations of 1.85. What is the best thing to do to try to improve the geometry. Should I try to optimize the weights?
the R-factors seem good indeed. Yes, Although the angle deviations seem ok to me, I'm concerned about the bond deviations... Yes, you can try two things: - increase the restraints strength by increasing the corresponding weight: from the command line: "wc=VALUE", where VALUE is a number somewhat larger than 1.0; from the GUI: Refinement settings -> All parameters -> Refinement target weights... -> Stereochemistry weight scale - Run automatic weight optimization (note, this may take a while to run): from the command line: use keyword: "optimize_wxc=true"; from the GUI: Refinement settings -> All parameters -> Refinement target weights... -> Optimize X-ray/stereochemistry weight Alternatively, you can send me your PDB and data file and I will tell you what exactly is not right and how to fix it. Pavel.
Dear All, I´m quite new to crystallography and I´m refining a structure with ASN-NAG-NAG-BMA attached and already got help from Blaine who told me how to do it, but I´m making a small mistake somewhere and can´t figure out what I´m doing wrong. Must be just one paranthesis or so, I guess. In command line just for testing I put: phenix.refine refine_data.mtz refine_1-coot-1FrenumberedGlu.pdb cif_link.params BMA.cif strategy=individual_sites+individual_adp main.number_of_macro_cycles=2 --overwrite where I put in cif_link.params as suggested in the mailing list http://www.phenix-online.org/pipermail/phenixbb/2009-November/002837.html with and without paranthesis at the beginning } apply_cif_link { data_link = NAG-ASN residue_selection_1 = chain A and resname NAG and resid 500 residue_selection_2 = chain A and resname ASN and resid 297 } apply_cif_link { data_link = BETA1-4 residue_selection_1 = chain A and resname NAG and resid 500 residue_selection_2 = chain A and resname NAG and resid 501 } apply_cif_link { data_link = BETA1-4 residue_selection_1 = chain A and resname NAG and resid 501 residue_selection_2 = chain A and resname BMA and resid 502 } apply_cif_link { data_link = ALPHA1-3 residue_selection_1 = chain A and resname BMA and resid 502 residue_selection_2 = chain A and resname MAN and resid 503 and suggested by Blaine with and without paranthesis at the beginning refinement.pdb_interpretation.apply_cif_link { data_link = NAG-ASN residue_selection_1 = chain A and resname NAG and resid 500 residue_selection_2 = chain A and resname ASN and resid 297 } refinement.pdb_interpretation.apply_cif_link { data_link = BETA1-4 residue_selection_1 = chain A and resname NAG and resid 500 residue_selection_2 = chain A and resname NAG and resid 501 } refinement.pdb_interpretation.apply_cif_link { data_link = BETA1-4 residue_selection_1 = chain A and resname NAG and resid 501 residue_selection_2 = chain A and resname BMA and resid 502 } refinement.pdb_interpretation.apply_cif_link { data_link = ALPHA1-3 residue_selection_1 = chain A and resname BMA and resid 502 residue_selection_2 = chain A and resname MAN and resid 503 I see in the .geo file that the residues are not bonded, and also in the .eff file that these lines are not added there, which should be done I guess, I put the info also direct into the .def file but I get the same problem? By the way it shouldn´t matter which name I give to this file as long as the extension is correct? Thank you very much and all the best, Georg. -----Ursprüngliche Nachricht----- Von: [email protected] [mailto:[email protected]] Im Auftrag von Ursula Schulze-Gahmen Gesendet: Dienstag, 23. Februar 2010 19:20 An: [email protected] Betreff: [phenixbb] weights optimization, add hydrogens I am refining a 2.0 A structure in Phenix using mostly the graphical interface. The maps are very good and the R factors are low (0.208/0.173). But the geometry is not so great with bond deviations of 0.208 and Angle deviations of 1.85. What is the best thing to do to try to improve the geometry. Should I try to optimize the weights? The other question is about adding hydrogens to the model. I did add them in a previous refinement cycle. After rebuilding the model in coot, most of the model still has the hydrogens in the input file for the next refinement cycle, but some residues and waters don't have hydrogens. When I tried to add hydrogens again to the model, I got an error message about atoms being to close. Does Phenix know where to add hydrogens if the input model has already hydroegns on many residues? Thanks Ursula -- Ursula Schulze-Gahmen, PhD. QB3, Tjian Lab MCB, 16 Barker Hall #3204 University of California Berkeley Berkeley, CA 94720-3204 Phone: (510) 642 8258 [email protected] _______________________________________________ phenixbb mailing list [email protected] http://phenix-online.org/mailman/listinfo/phenixbb
participants (6)
-
Christian Roth -
Georg Mlynek -
Nathaniel Echols -
Pavel Afonine -
Schubert, Carsten [PRDUS] -
Ursula Schulze-Gahmen