Handling of modified nucleotides in ribosome structures
Dear community, I'm trying to work with a published ribosome structure (PDB 7K00, https://www.rcsb.org/structure/7k00 ) for example in phenix.combine_focused_maps or phenix.real_space_refine. Of course this structure contains a lot of modified residues and the monomer library in a fresh Phenix installation (version 1.21.2-5419) does not handle these out of the box. Since a lot of publications have used 7K00 as a starting point for their work in Phenix and since the current CCP4 monomer library plays nice with this structure I would not have expected this. Am I missing something in my Phenix setup to make it recognize such "common" modified residues in a ribosome or do I indeed need to generate parameters myself using AceDRG, eLBOW, Grade2 etc.? In the latter case, could I simply swap out the stock Phenix monomer library for the one from CCP4 as a drop-in replacement or would that be a bad idea? Thank you! Christian
Christian
While there is the possibility to point to the CCP4 mon_lib from Phenix, it
is generally better to generate the restraints because there are nuances. I
would use the command
phenix.pdb_interpretation model.cif
to find the missing restraints and then
phenix.elbow --chemical_component=<XYZ>
with the optional
--final_geometry=<XYZ>
until you have the restraints needed.
I'm happy to help more with restraints if needed.
Cheers
Nigel
---
Nigel W. Moriarty
Building 91, Molecular Biophysics and Integrated Bioimaging
Lawrence Berkeley National Laboratory
Berkeley, CA 94720-8235
Email : [email protected]
Web : CCI.LBL.gov
ORCID : orcid.org/0000-0001-8857-9464
On Tue, Dec 10, 2024 at 2:25 PM
Dear community,
I'm trying to work with a published ribosome structure (PDB 7K00, https://www.rcsb.org/structure/7k00 ) for example in phenix.combine_focused_maps or phenix.real_space_refine. Of course this structure contains a lot of modified residues and the monomer library in a fresh Phenix installation (version 1.21.2-5419) does not handle these out of the box.
Since a lot of publications have used 7K00 as a starting point for their work in Phenix and since the current CCP4 monomer library plays nice with this structure I would not have expected this. Am I missing something in my Phenix setup to make it recognize such "common" modified residues in a ribosome or do I indeed need to generate parameters myself using AceDRG, eLBOW, Grade2 etc.?
In the latter case, could I simply swap out the stock Phenix monomer library for the one from CCP4 as a drop-in replacement or would that be a bad idea?
Thank you! Christian _______________________________________________ phenixbb mailing list -- [email protected] To unsubscribe send an email to [email protected] Unsubscribe: phenixbb-leave@%(host_name)s
Dear Nigel, thank you for the quick reply. I got phenix.pdb_interpretation to run after also supplying the options "disable_uc_volume_vs_n_atoms_check=True" and "proceed_with_excessive_length_bonds=True". It would then later hang on a "Symmetry disulfide" step but at least I got the names of the missing residues. In case it is relevant, they are 4OC, G7M, MA6, UR3, IAS, 1MG, 2MA, 3TD, 6MZ and OMU. phenix.elbow with --final_geometry=... then produced the required cif files. I am however stumped how to proceed from here. When I try to run phenix.combine_focused_maps using any of the cif files as an input argument for the parameter "restraint_files" the same error remains (i.e. the number of unknown symbols remains unchanged at 282): " Fatal problems interpreting model file: Number of atoms with unknown nonbonded energy type symbols: 282 Please edit the model file to resolve the problems and/or supply a CIF file with matching restraint definitions, along with apply_cif_modification and apply_cif_link parameter definitions if necessary. " Based on what I saw in the documentation, I am not sure how to work with the "apply_cif_modification" parameter definition when it comes to modified/unusual nucleotides and amino acids. I also ran ReadySet on my 7K00 cif file to generate a combined .ligands.cif restraints file and feed that into phenix.combine_focused_maps, but the outcome was the same as before. An additional problem comes up when I try to supply these restraints, either as individual cif files or as the combined one from ReadySet, to phenix.pdb_interpretation. This fails with the following error: " Sorry: Error processing CIF file: "/home/csteinmetzger/cryo-EM/ecoli_ribosome/wt/ReadySet_14/7k00_add_models_calibrated.ligands.cif" (AssertionError: /usr/software/phenix-1.21.2-5419/modules/cctbx_project/mmtbx/monomer_library/server.py line 371) " The 7K00 model I'm using here has been stripped of all waters, metal ions, small molecule ligands, tRNAs and mRNA in ChimeraX. I was not sure whether that might have done anything to the file to make it problematic. So I downloaded the file freshly from the pdb and supplied it to phenix.combine_focused_maps as is. This results in the following error: " Sorry: The model 7k00.cif does not seem to have complete residues and only RNA or protein (17719 residues and 10141 CA or P atoms) " One thing that does seem to work (at least start without error) is phenix.real_space_refine using my stripped 7K00 model and the .ligands.cif restraints file from ReadySet. I hope we can get phenix.combine_focused_maps to work as well. Christian
Christian As Tom pointed out, you may need to use the nightly for phenix.combine_focused_maps. ReadySet! does not currently work on mmCIF files so ignore any output from there. "apply_cif_modification" is for changes to entities as when a polymer is formed so that is not your answer to restraints.
thank you for the quick reply. I got phenix.pdb_interpretation to run after also supplying the options "disable_uc_volume_vs_n_atoms_check=True" and "proceed_with_excessive_length_bonds=True". It would then later hang on a "Symmetry disulfide" step but at least I got the names of the missing residues. In case it is relevant, they are 4OC, G7M, MA6, UR3, IAS, 1MG, 2MA, 3TD, 6MZ and OMU. phenix.elbow with --final_geometry=... then produced the required cif files.
Great. Supply these to real_space_refine and you should be golden.
This has been a very insightful discussion! The nightly build suggested by Tom takes in my maps, model and restraints without giving any error. It is then sitting "Refining model 1 with rigid_body refinement" for a while before the terminal shows "Killed". For the moment I will look into some alternatives for creating composite maps, e.g. in ChimeraX. Christian
Hi Christian, Oh I'm sorry about the "killed". That means your computer ran out of memory running combine_focused_maps. It is possible that you might be able to run it if you first box your maps closer around your models, so that the maps are smaller. All the best, Tom T On Thu, Dec 12, 2024 at 1:59 PM Christian Steinmetzger < [email protected]> wrote:
This has been a very insightful discussion!
The nightly build suggested by Tom takes in my maps, model and restraints without giving any error. It is then sitting "Refining model 1 with rigid_body refinement" for a while before the terminal shows "Killed".
For the moment I will look into some alternatives for creating composite maps, e.g. in ChimeraX.
Christian _______________________________________________ phenixbb mailing list -- [email protected] To unsubscribe send an email to [email protected] Unsubscribe: phenixbb-leave@%(host_name)s
-- Thomas C Terwilliger Laboratory Fellow, Los Alamos National Laboratory Senior Scientist, New Mexico Consortium 100 Entrada Dr, Los Alamos, NM 87544 Email: [email protected] Tel: 505-431-0010
Hi Tom, I trimmed the maps down from it's original size of 576 x 576 x 576 to 297 x 322 x 355 using "phenix.map_box model.cif map1.mrc " and then submitted the resulting output model_box.cif, map1_box.ccp4 and map2_box.ccp4 to combine_focused_maps, but still got the same memory issue on the rigid-body refinement step. I monitored the job for a while and the python process had reached more than 90 GB memory use when I checked. My workstation has 128 GB RAM in total. I'm sending the full terminal output of the job below. Nigel, I also encountered another issue with using elbow for restraint generation. For 4OC and UR3, the command phenix.elbow --chemical_component="$RES" --final_geometry="$RES" doesn't produce a cif file because it encounters an unusually long P-O bond in the phosphate group, ~150 or 182 Å respectively. A pdb file is still saved and there atom OP3 is clearly in a very odd position. ------------------------------------------------------------------------------------------------ Here is the terminal output from combine_focused_maps: Starting job =============================================================================== Validating inputs ************************************************************ PHENIX combine_focused_maps Tue Dec 17 18:27:58 2024 ************************************************************ Working directory: /home/csteinmetzger/workdir/maps_composite PHENIX VERSION: dev-5533 Reading input data Removed 1 residues with missing CA/P atoms Model with 10208 residues read from /home/csteinmetzger/workdir/model_box.cif Map file read from /home/csteinmetzger/workdir/map1_box.ccp4 header_min: -0.215969517827 header_max: 0.890450239182 header_mean: 0.00580687774345 header_rms: 0.0463103465736 Information about FULL UNIT CELL: unit cell grid: (576, 576, 576) unit cell parameters: (436.32000732421875, 436.32000732421875, 436.32000732421875, 90.0, 90.0, 90.0) space group number: 1 Information about the PART OF MAP THAT IS PRESENT: map cell grid: (297, 322, 355) map cell parameters: (224.9775037765503, 243.91500409444174, 268.9125045140584, 90.0, 90.0, 90.0) map origin: (131, 125, 110) pixel size: (0.7575, 0.7575, 0.7575) Shift (grid units) to place origin at original position: (0, 0, 0) Wrapping (using unit_cell_translations to get map values) allowed: False Map file read from /home/csteinmetzger/workdir/map2_box.ccp4 header_min: -0.172103002667 header_max: 0.68276655674 header_mean: 0.0004433694412 header_rms: 0.0258353501558 Information about FULL UNIT CELL: unit cell grid: (576, 576, 576) unit cell parameters: (436.32000732421875, 436.32000732421875, 436.32000732421875, 90.0, 90.0, 90.0) space group number: 1 Information about the PART OF MAP THAT IS PRESENT: map cell grid: (297, 322, 355) map cell parameters: (224.9775037765503, 243.91500409444174, 268.9125045140584, 90.0, 90.0, 90.0) map origin: (131, 125, 110) pixel size: (0.7575, 0.7575, 0.7575) Shift (grid units) to place origin at original position: (0, 0, 0) Wrapping (using unit_cell_translations to get map values) allowed: False Input maps and matching models: Map 1: /home/csteinmetzger/workdir/map1_box.ccp4 Model 1: /home/csteinmetzger/workdir/model_box.cif Map 2: /home/csteinmetzger/workdir/map2_box.ccp4 Model 2: /home/csteinmetzger/workdir/model_box.cif Normalizing maps (mean=0 sd=1) before use Running rigid body refinement on each model and map Refining each chain of a molecule as a single unit Refining model 1 with rigid_body refinement Killed
eLBOW fails because the Chemical Component data is incomplete. You can get it to work by removing the final_geom option. Cheers Nigel --- Nigel W. Moriarty Building 91, Molecular Biophysics and Integrated Bioimaging Lawrence Berkeley National Laboratory Berkeley, CA 94720-8235 Email : [email protected] Web : CCI.LBL.gov ORCID : orcid.org/0000-0001-8857-9464 On Tue, Dec 17, 2024 at 1:24 PM Christian Steinmetzger < [email protected]> wrote:
Hi Tom,
I trimmed the maps down from it's original size of 576 x 576 x 576 to 297 x 322 x 355 using "phenix.map_box model.cif map1.mrc "
and then submitted the resulting output model_box.cif, map1_box.ccp4 and map2_box.ccp4 to combine_focused_maps, but still got the same memory issue on the rigid-body refinement step. I monitored the job for a while and the python process had reached more than 90 GB memory use when I checked. My workstation has 128 GB RAM in total. I'm sending the full terminal output of the job below.
Nigel, I also encountered another issue with using elbow for restraint generation. For 4OC and UR3, the command phenix.elbow --chemical_component="$RES" --final_geometry="$RES" doesn't produce a cif file because it encounters an unusually long P-O bond in the phosphate group, ~150 or 182 Å respectively. A pdb file is still saved and there atom OP3 is clearly in a very odd position.
------------------------------------------------------------------------------------------------ Here is the terminal output from combine_focused_maps:
Starting job
=============================================================================== Validating inputs
************************************************************ PHENIX combine_focused_maps Tue Dec 17 18:27:58 2024 ************************************************************
Working directory: /home/csteinmetzger/workdir/maps_composite
PHENIX VERSION: dev-5533
Reading input data
Removed 1 residues with missing CA/P atoms Model with 10208 residues read from /home/csteinmetzger/workdir/model_box.cif
Map file read from /home/csteinmetzger/workdir/map1_box.ccp4 header_min: -0.215969517827 header_max: 0.890450239182 header_mean: 0.00580687774345 header_rms: 0.0463103465736
Information about FULL UNIT CELL: unit cell grid: (576, 576, 576) unit cell parameters: (436.32000732421875, 436.32000732421875, 436.32000732421875, 90.0, 90.0, 90.0) space group number: 1
Information about the PART OF MAP THAT IS PRESENT: map cell grid: (297, 322, 355) map cell parameters: (224.9775037765503, 243.91500409444174, 268.9125045140584, 90.0, 90.0, 90.0) map origin: (131, 125, 110) pixel size: (0.7575, 0.7575, 0.7575) Shift (grid units) to place origin at original position: (0, 0, 0) Wrapping (using unit_cell_translations to get map values) allowed: False
Map file read from /home/csteinmetzger/workdir/map2_box.ccp4 header_min: -0.172103002667 header_max: 0.68276655674 header_mean: 0.0004433694412 header_rms: 0.0258353501558
Information about FULL UNIT CELL: unit cell grid: (576, 576, 576) unit cell parameters: (436.32000732421875, 436.32000732421875, 436.32000732421875, 90.0, 90.0, 90.0) space group number: 1
Information about the PART OF MAP THAT IS PRESENT: map cell grid: (297, 322, 355) map cell parameters: (224.9775037765503, 243.91500409444174, 268.9125045140584, 90.0, 90.0, 90.0) map origin: (131, 125, 110) pixel size: (0.7575, 0.7575, 0.7575) Shift (grid units) to place origin at original position: (0, 0, 0) Wrapping (using unit_cell_translations to get map values) allowed: False
Input maps and matching models:
Map 1: /home/csteinmetzger/workdir/map1_box.ccp4 Model 1: /home/csteinmetzger/workdir/model_box.cif
Map 2: /home/csteinmetzger/workdir/map2_box.ccp4 Model 2: /home/csteinmetzger/workdir/model_box.cif
Normalizing maps (mean=0 sd=1) before use
Running rigid body refinement on each model and map Refining each chain of a molecule as a single unit Refining model 1 with rigid_body refinement Killed _______________________________________________ phenixbb mailing list -- [email protected] To unsubscribe send an email to [email protected] Unsubscribe: phenixbb-leave@%(host_name)s
Hi Christian, Thanks for letting me know about this over-use of memory in phenix CombineFocusedMap that crashed your computer!. I found the problem and fixed it today, so starting with tomorrow's Phenix nightly build which will appear here when ready: https://phenix-online.org/download/phenix/nightly/?version=dev-5550 the CombineFocusedMap tool should work with large models and large maps, even if rigid-body refinement is done on each chain with the keyword rigid_body_refinement_single_unit=False (as is default). Let me know if this does not do it! All the best, Tom T On Tue, Dec 17, 2024 at 2:24 PM Christian Steinmetzger < [email protected]> wrote:
Hi Tom,
I trimmed the maps down from it's original size of 576 x 576 x 576 to 297 x 322 x 355 using "phenix.map_box model.cif map1.mrc "
and then submitted the resulting output model_box.cif, map1_box.ccp4 and map2_box.ccp4 to combine_focused_maps, but still got the same memory issue on the rigid-body refinement step. I monitored the job for a while and the python process had reached more than 90 GB memory use when I checked. My workstation has 128 GB RAM in total. I'm sending the full terminal output of the job below.
Nigel, I also encountered another issue with using elbow for restraint generation. For 4OC and UR3, the command phenix.elbow --chemical_component="$RES" --final_geometry="$RES" doesn't produce a cif file because it encounters an unusually long P-O bond in the phosphate group, ~150 or 182 Å respectively. A pdb file is still saved and there atom OP3 is clearly in a very odd position.
------------------------------------------------------------------------------------------------ Here is the terminal output from combine_focused_maps:
Starting job
=============================================================================== Validating inputs
************************************************************ PHENIX combine_focused_maps Tue Dec 17 18:27:58 2024 ************************************************************
Working directory: /home/csteinmetzger/workdir/maps_composite
PHENIX VERSION: dev-5533
Reading input data
Removed 1 residues with missing CA/P atoms Model with 10208 residues read from /home/csteinmetzger/workdir/model_box.cif
Map file read from /home/csteinmetzger/workdir/map1_box.ccp4 header_min: -0.215969517827 header_max: 0.890450239182 header_mean: 0.00580687774345 header_rms: 0.0463103465736
Information about FULL UNIT CELL: unit cell grid: (576, 576, 576) unit cell parameters: (436.32000732421875, 436.32000732421875, 436.32000732421875, 90.0, 90.0, 90.0) space group number: 1
Information about the PART OF MAP THAT IS PRESENT: map cell grid: (297, 322, 355) map cell parameters: (224.9775037765503, 243.91500409444174, 268.9125045140584, 90.0, 90.0, 90.0) map origin: (131, 125, 110) pixel size: (0.7575, 0.7575, 0.7575) Shift (grid units) to place origin at original position: (0, 0, 0) Wrapping (using unit_cell_translations to get map values) allowed: False
Map file read from /home/csteinmetzger/workdir/map2_box.ccp4 header_min: -0.172103002667 header_max: 0.68276655674 header_mean: 0.0004433694412 header_rms: 0.0258353501558
Information about FULL UNIT CELL: unit cell grid: (576, 576, 576) unit cell parameters: (436.32000732421875, 436.32000732421875, 436.32000732421875, 90.0, 90.0, 90.0) space group number: 1
Information about the PART OF MAP THAT IS PRESENT: map cell grid: (297, 322, 355) map cell parameters: (224.9775037765503, 243.91500409444174, 268.9125045140584, 90.0, 90.0, 90.0) map origin: (131, 125, 110) pixel size: (0.7575, 0.7575, 0.7575) Shift (grid units) to place origin at original position: (0, 0, 0) Wrapping (using unit_cell_translations to get map values) allowed: False
Input maps and matching models:
Map 1: /home/csteinmetzger/workdir/map1_box.ccp4 Model 1: /home/csteinmetzger/workdir/model_box.cif
Map 2: /home/csteinmetzger/workdir/map2_box.ccp4 Model 2: /home/csteinmetzger/workdir/model_box.cif
Normalizing maps (mean=0 sd=1) before use
Running rigid body refinement on each model and map Refining each chain of a molecule as a single unit Refining model 1 with rigid_body refinement Killed _______________________________________________ phenixbb mailing list -- [email protected] To unsubscribe send an email to [email protected] Unsubscribe: phenixbb-leave@%(host_name)s
-- Thomas C Terwilliger Laboratory Fellow, Los Alamos National Laboratory Senior Scientist, New Mexico Consortium 100 Entrada Dr, Los Alamos, NM 87544 Email: [email protected] Tel: 505-431-0010
Hi Tom, thank you for the help, I really appreciate it! Best, Christian
participants (4)
-
Christian Steinmetzger -
christian.steinmetzger＠imbim.uu.se
-
Nigel Moriarty -
Tom Terwilliger