anisotropic refinement and hydrogens in coot.
Two hopefully quick questions; 1) The phenix.refine documentation has a section 'Refinement at "higher than medium" resolution - getting anisotropic'. I have a 1.4A structure that clearly falls into this category, I've added riding hydrogens and now I'd like to try anisotropic ADPs. Is there a way, to list all residues (preferably in phenix selection format) that have "relatively small isotropic B-factors ~5-20A**2 of so."? 2) The output.pdb file from phenix.refine is I think the pdb v3.0 format. Is there an option or apps to convert to v2.3, within phenix rather using remediator (duke)? Coot still uses v2.3 files, which is fine until hydrogens are added to a pdb, because the nomenclature has changed. RSR in coot causes these 3.0 named hydrogens to fly off the residue into surrounding density. Thanks in advance, Mark
Hi Mark,
1) The phenix.refine documentation has a section 'Refinement at "higher than medium" resolution - getting anisotropic'. I have a 1.4A structure that clearly falls into this category, I've added riding hydrogens and now I'd like to try anisotropic ADPs. Is there a way, to list all residues (preferably in phenix selection format) that have "relatively small isotropic B-factors ~5-20A**2 of so."?
Yes, at 1.6A resolution you should refine anisotropic ADP. You can manually select atoms which you want to refine as anisotropic and isotropic: phenix.refine model.pdb data.mtz params.par where "params.par" file contains these lines (edited copy-paste from all parameters file): refinement.refine { adp { individual { isotropic = (chain A and resseq 1-20) or (chain B and resseq 30-89) anisotropic= (chain A and resseq 234:426) or (chain B and resseq 765:900) } } } Currently phenix.refine does not allow to do selection on atomic properties, like B-factor values. So, you need to figure out yourself. As a start, I would just do this and see if it results in any problems (I think it will work just fine): refinement.refine { adp { individual { isotropic = element H or element D or water anisotropic= not (element H or element D or water) } } } which tells to refine all atoms as anisotropic except water and hydrogens (deuteriums if any).
2) The output.pdb file from phenix.refine is I think the pdb v3.0 format. Is there an option or apps to convert to v2.3, within phenix rather using remediator (duke)? Coot still uses v2.3 files, which is fine until hydrogens are added to a pdb, because the nomenclature has changed. RSR in coot causes these 3.0 named hydrogens to fly off the residue into surrounding density.
phenix.refine can handle both the PDB V2 and V3 hydrogen atom names. Since Coot does not with the PDB V3 conventions, you can do everything with PDB V2 hydrogen names. If you use phenix.reduce to add the hydrogens, use the -OLDpdb option. Pavel.
Hi Pavel, After refinement with TLS how should we switch to refinement with aniso individual ADPs. i know that both cannot be done simultaneously at present. Following up on Mark's question: we also have a set of structures of resolution 1.5 to 1.24 Å resolution that are good candidates for anisotropic refinement, but selecting the subset of residues to refine is not easy (for us). We can exclude loops and terminii with high isotropic ADPs, but even for the remaining well ordered parts of the protein we are left with surface residues with well ordered main chain but less well ordered side chains (e.g, Lys and Glu), which need to be refined isotropically. Also, we would like to exclude residues with alternative conformations from anisotropic refinment. At present its a fair amount of work to build up a residue selection to satisfy these requirements. -- Mark Mayer Ph.D.
Hi Mark,
After refinement with TLS how should we switch to refinement with aniso individual ADPs. i know that both cannot be done simultaneously at present.
Just turn off the TLS refinement option. Since input PDB file will contain atoms with ANISOU records they will be refined as anisotropic automatically. Alternatively, you can specify which atoms to refine as isotropic or anisotropic.
Following up on Mark's question: we also have a set of structures of resolution 1.5 to 1.24 Å resolution that are good candidates for anisotropic refinement, but selecting the subset of residues to refine is not easy (for us).
We can exclude loops and terminii with high isotropic ADPs, but even for the remaining well ordered parts of the protein we are left with surface residues with well ordered main chain but less well ordered side chains (e.g, Lys and Glu), which need to be refined isotropically. Also, we would like to exclude residues with alternative conformations from anisotropic refinment.
At present its a fair amount of work to build up a residue selection to satisfy these requirements.
The goal is to make refinement completely automated where you hit Enter and get refined structure. What you describe above is one of the items in to-do list for further automation. Also, I do not think you need to refine as isotropic the atoms with large B-factors. Yes, ac should probably be refined as isotropic. This is something to systematically study (as part of the overall planned automation). For the moment, I would just remove from anisotropic refinement some very obvious parts that should be refined as isotropic and I'm almost sure it will be enough in most of the cases. Pavel.
Hi I had been think of asking a friend to write a short python script that could list all residues or even all atoms (below a user defined B factor threshold) in a phenix selection format. Then I could run this and just cut and paste the residues with ADPs below 5, or 10 or 20, and run phenix.refine with anisotropic for these differing selections. The trick s getting someone to spend the hour or two I need. Any takers? :-D Mark On Aug 19, 2008, at 6:05 PM, Pavel Afonine wrote:
Hi Mark,
After refinement with TLS how should we switch to refinement with aniso individual ADPs. i know that both cannot be done simultaneously at present.
Just turn off the TLS refinement option. Since input PDB file will contain atoms with ANISOU records they will be refined as anisotropic automatically. Alternatively, you can specify which atoms to refine as isotropic or anisotropic.
Following up on Mark's question: we also have a set of structures of resolution 1.5 to 1.24 Å resolution that are good candidates for anisotropic refinement, but selecting the subset of residues to refine is not easy (for us).
We can exclude loops and terminii with high isotropic ADPs, but even for the remaining well ordered parts of the protein we are left with surface residues with well ordered main chain but less well ordered side chains (e.g, Lys and Glu), which need to be refined isotropically. Also, we would like to exclude residues with alternative conformations from anisotropic refinment.
At present its a fair amount of work to build up a residue selection to satisfy these requirements.
The goal is to make refinement completely automated where you hit Enter and get refined structure. What you describe above is one of the items in to-do list for further automation.
Also, I do not think you need to refine as isotropic the atoms with large B-factors. Yes, ac should probably be refined as isotropic. This is something to systematically study (as part of the overall planned automation).
For the moment, I would just remove from anisotropic refinement some very obvious parts that should be refined as isotropic and I'm almost sure it will be enough in most of the cases.
Pavel.
_______________________________________________ phenixbb mailing list [email protected] http://www.phenix-online.org/mailman/listinfo/phenixbb
Hi Mark, attached is a simple Python script that could be a good start for someone interested to make its output nicer. Although, it could be used as is (with some minor reformatting of the output). Pavel. On 8/19/2008 7:06 PM, Mark Collins wrote:
Hi I had been think of asking a friend to write a short python script that could list all residues or even all atoms (below a user defined B factor threshold) in a phenix selection format. Then I could run this and just cut and paste the residues with ADPs below 5, or 10 or 20, and run phenix.refine with anisotropic for these differing selections. The trick s getting someone to spend the hour or two I need. Any takers? :-D Mark
On Aug 19, 2008, at 6:05 PM, Pavel Afonine wrote:
Hi Mark,
After refinement with TLS how should we switch to refinement with aniso individual ADPs. i know that both cannot be done simultaneously at present.
Just turn off the TLS refinement option. Since input PDB file will contain atoms with ANISOU records they will be refined as anisotropic automatically. Alternatively, you can specify which atoms to refine as isotropic or anisotropic.
Following up on Mark's question: we also have a set of structures of resolution 1.5 to 1.24 Å resolution that are good candidates for anisotropic refinement, but selecting the subset of residues to refine is not easy (for us).
We can exclude loops and terminii with high isotropic ADPs, but even for the remaining well ordered parts of the protein we are left with surface residues with well ordered main chain but less well ordered side chains (e.g, Lys and Glu), which need to be refined isotropically. Also, we would like to exclude residues with alternative conformations from anisotropic refinment.
At present its a fair amount of work to build up a residue selection to satisfy these requirements.
The goal is to make refinement completely automated where you hit Enter and get refined structure. What you describe above is one of the items in to-do list for further automation.
Also, I do not think you need to refine as isotropic the atoms with large B-factors. Yes, ac should probably be refined as isotropic. This is something to systematically study (as part of the overall planned automation).
For the moment, I would just remove from anisotropic refinement some very obvious parts that should be refined as isotropic and I'm almost sure it will be enough in most of the cases.
Pavel.
_______________________________________________ phenixbb mailing list [email protected] http://www.phenix-online.org/mailman/listinfo/phenixbb
_______________________________________________ phenixbb mailing list [email protected] http://www.phenix-online.org/mailman/listinfo/phenixbb
from mmtbx import utils from cStringIO import StringIO def run(pdb_file, b_max, b_min): processed_pdb_files_srv = utils.process_pdb_file_srv(log = StringIO()) processed_pdb_file, pdb_inp = \ processed_pdb_files_srv.process_pdb_files(pdb_file_names = [pdb_file]) hierarchy = processed_pdb_file.all_chain_proxies.pdb_hierarchy # assert len(hierarchy.models()) == 1 for model in hierarchy.models(): for chain in model.chains(): for residue_group in chain.residue_groups(): for atom_group in residue_group.atom_groups(): for atom in atom_group.atoms(): if(atom.b < b_max and atom.b > b_min): chain_id = chain.id.strip() resseq = residue_group.resseq.strip() atom_name = atom.name.strip() if(len(chain_id) > 0): print "(chain %s and resseq %s and name %s)" % ( chain_id, resseq, atom_name) else: print "(resseq %s and name %s)" % (resseq, atom_name) if (__name__ == "__main__"): run(pdb_file = "model.pdb", b_max = 30., b_min = 10.)
Hello again, When selecting groups within a pdb file for anisotropic refinement, i.e. portions of the structure will be anisotropic and the rest will be isotropic. Would it be better to a) select any atoms below a user defined Bfactor limit b) select only residues with all atoms below a user defined Bfactor limit c) select any residue with a residue average below a user defined Bfactor limit My concerns with these options are (a) may give only some atoms within a particular residue, is that ok? (b) could limit the anisotropic selction beacuse a residue with all but one atom under the limit would be excluded. Does (c) really solve both these problems? Or are there other options I'm missing? Thanks, Mark
Hi Mark, just to add to my previous email: you can easily change the script to be able to exclude the alternative conformations or hydrogen atoms or water (which you most likely want to do). Pavel. On 8/19/2008 7:06 PM, Mark Collins wrote:
Hi I had been think of asking a friend to write a short python script that could list all residues or even all atoms (below a user defined B factor threshold) in a phenix selection format. Then I could run this and just cut and paste the residues with ADPs below 5, or 10 or 20, and run phenix.refine with anisotropic for these differing selections. The trick s getting someone to spend the hour or two I need. Any takers? :-D Mark
On Aug 19, 2008, at 6:05 PM, Pavel Afonine wrote:
Hi Mark,
After refinement with TLS how should we switch to refinement with aniso individual ADPs. i know that both cannot be done simultaneously at present.
Just turn off the TLS refinement option. Since input PDB file will contain atoms with ANISOU records they will be refined as anisotropic automatically. Alternatively, you can specify which atoms to refine as isotropic or anisotropic.
Following up on Mark's question: we also have a set of structures of resolution 1.5 to 1.24 Å resolution that are good candidates for anisotropic refinement, but selecting the subset of residues to refine is not easy (for us).
We can exclude loops and terminii with high isotropic ADPs, but even for the remaining well ordered parts of the protein we are left with surface residues with well ordered main chain but less well ordered side chains (e.g, Lys and Glu), which need to be refined isotropically. Also, we would like to exclude residues with alternative conformations from anisotropic refinment.
At present its a fair amount of work to build up a residue selection to satisfy these requirements.
The goal is to make refinement completely automated where you hit Enter and get refined structure. What you describe above is one of the items in to-do list for further automation.
Also, I do not think you need to refine as isotropic the atoms with large B-factors. Yes, ac should probably be refined as isotropic. This is something to systematically study (as part of the overall planned automation).
For the moment, I would just remove from anisotropic refinement some very obvious parts that should be refined as isotropic and I'm almost sure it will be enough in most of the cases.
Pavel.
_______________________________________________ phenixbb mailing list [email protected] http://www.phenix-online.org/mailman/listinfo/phenixbb
_______________________________________________ phenixbb mailing list [email protected] http://www.phenix-online.org/mailman/listinfo/phenixbb
participants (4)
-
Dr. Mark Mayer -
Mark Collins -
Pavel Afonine -
Pavel Afonine