Cryo_fit1 FAQ

Contents

cc values have kept decreased (or have not increased that much) during step 8

  • Case 1 <A user input pdb file has reasonable structural geometry> For this case, try below possible solutions. If cryo_fit still can't find better (higher) cc, then the initial correlation between user input pdb file and cryo-EM map is already high enough. Just run phenix.real_space_refine only and deposit. Doo Nam recommends to argue/claim in a paper like "Our fit is so high, even cryo_fit did not find higher level of fit than our initial fit". This bold recommendation is based on his limited observations so far. Therefore, if you have a better cc model (from manual fit?) than cryo_fitted model, we would appreciate if you can email us (jprajapati@lanl.gov, dgirodat@lanl.gov, doonam.kim@gmail.com).
  • Case 2 <A user input pdb file has unreasonable structural geometry> For this case, although the initial fit between user's atomic model and map looks good, it is a fictitious fitting without consideration of ideal atomistic model geometry. Run cryo_fit to find decent fit to the cryo-EM map that restores/maintains reasonable structural geometry based on molecular dynamics forcefield (amber03.ff).
  1. Provide a higher resolution map (like better than 4.5 Angstrom) tends to find higher cc since there is a more chance of finding fittable regions.
  2. Cryo_fit calculates the gradient of cc. If cryo_fit is provided a giant cryo-em map with a tiny atomic model, then there is a large empty space (not filled). Therefore, the constraint forces for the model are very small. Consequently, these small forces are not helpful.

2-1. Re-run cryo_fit with an atomic model that fits the majority of the map. If you fit multiple atomic models into a symmetric map or do sequential fitting into a non-symmetric map, watch Tom Goddard's lecture (2013)

2-2. Re-run cryo_fit with only a relevant map region. See 'How to extract a relevant map region?' in this FAQ.

  1. If the initial model is not properly aligned to a map, see 'How to improve initial cc?' in this FAQ.
  2. A user may enforce a stronger initial map weight (e.g. emweight_multiply_by) manually. However, since cryo_fit automatically increases emweight_multiply_by if it doesn't find higher cc anyway, manual increase in map weight may not necessarily find higher cc.

How can I provide gaussian filtered maps?

How can I use double digit/character chain ID pdb file?

  • python <User Phenix>/modules/cryo_fit/files_for_steps/9_after_cryo_fit/solve_duplicate_atoms_issue/add_all_prefix.py

How long does it take to run cryo_fit?

../images/cryo_fit_FAQ_how_long1.png ../images/cryo_fit_FAQ_how_long2.png

How to draw a figure with cc values?

  • If a user is using ssh linked linux, run 'set DISPLAY' to avoid the message of "Unable to access the X Display, is $DISPLAY set properly?".

How to enlarge map box size?

  1. cryo_fit which reports a map size between step 6 and step 7 with codes of
  • from iotbx import ccp4_map
  • ccp4_map = ccp4_map.map_reader(user_input_map)
  • target_map_data = ccp4_map.map_data()
  1. EMAN2's e2iminfo
  • For example, e2iminfo.py <user>.mrc
  • (note) As of EMAN2.2, the file extension should be mrc, not ccp4
  1. VMD which can visualize current map size as in this FAQ.
  • For example, relion_image_handler --i <user>.mrc --new_box 370 --o <user_box_size_370>.mrc

How to extract a relevant map region?

How to generate and record movie?

How to improve initial cc?

How to maximize cc?

  • Note that these options (phenix.dock_in_map or UCSF Chimera's 'fit in map') do rigid-body fitting only. Therefore, these are useful as global fitting before cryo_fit1. However, when overall orientation is already well fitted, often these are not that needed before cryo_fit1.
../images/cryo_fit2_example_1.png

), enforcing stronger emweight_multiply_by tends to fit better (obviously).

  • For example, phenix.cryo_fit <user.pdb> <user.map> emweight_multiply_by=50
  • At this high emweight_multiply_by, the secondary structures may be broken (slightly or seriously) right after cryo_fit running.
  • However, phenix.model_idealization which is followed by phenix.real_space_refine perfectly restored ideal secondary structures and fitted very well.

I can't run phenix.superpose_pdbs with cryo_fitted pdb file

  • This applies to all pdb input files (not only cryo_fitted files).

I see "Fatal error: A charge group moved too far between two domain decomposition steps. This usually means that your system is not well equilibrated" at my 8_cryo_fit step.

I see "Fatal error: Incomplete ring in HIS50" at my 8_cryo_fit step

  1. Open pdb file
  2. [menu]
  3. Select -> Residue -> HIS
  4. Tools -> Structure Editing -> Rotamers -> OK
  5. (select the most probable rotamer each)
  6. File -> Save PDB

I see "Fatal error: Number of grid cells is zero. Probably the system and box collapsed." at my 8_cryo_fit step.

  • Most likely, this means that initial cc is too low for MD simulation.
  • When Doo Nam ran phenix.real_space_refine first, then run real_space_refined atomic model in cryo_fit, it was solved.
  • Alternatively, improve initial cc by fitting initial atomic model into a map (see "How to improve initial cc?" in this FAQ)
  • Less likely, but still a possible case is when the map weight is too high, lowering emweight_multiply_by may help.

I see "pdb file cleaning is not done" at my step 1 (Make gro and topology file by regular gromacs)

"I edited out lipids, HEM and other hetero atoms and I verified that they are all gone. However, still my pdb file is not clean enough for gromacs_cryo_fit".

I see "Software inconsistency error" at my 8_cryo_fit step

  1. When he tried to run cryo_fit on Ubuntu 16.0. However, the same data set (map and model) didn't cause any error on his macOS 10.13.6.
  2. When he tried to fit a very lowly correlated initial atomistic model to a map (therefore, he observed like step 3600 correlation coefficient: -0.000820). Therefore, he ran phenix.dock_in_map first then, provided placed_model.pdb as the input model to cryo_fit. Then, the error disappeared.

I see "state.cpt not found, step_8 may be full of stepxb_nx.pdb" at my 8_cryo_fit step

  • Most likely, this means that a cryo_fit input pdb file is not yet stable enough for sensitive Gromacs MD simulation (step 8) even after cryo_fit's ample minimization step (e.g. step 4).
  • When Doo Nam ran phenix.real_space_refine first, then run real_space_refined atomic model in cryo_fit, it was solved more than 2 cases.
  • If initial cc is too low, improve initial cc by fitting initial atomic model into a map,
  • either by following "How to improve initial cc?" in this FAQ
  • or adding/copying more atoms.
  • For example, when Doo Nam tried to fit a monomer into a trimer map, this error occured. However, simply adding dimer atomic models into an input pdb file and running cryo_fit again removed the error.
  • Less likely, but still a possible case is when the map weight is too high, lowering emweight_multiply_by may help.

I see "step 0 correlation coefficient: nan" at my 8_cryo_fit step.

  • Therefore, please improve initial cc by fitting initial atomic model into a map (see "How to improve initial cc?" in this FAQ)

I see "step 0 correlation coefficient: nan" and "Range checking error: Explanation: During neighbor searching, we assign each particle to a grid based on its coordinates. If your system contains collisions or parameter errors that give particles very high velocities you might end up with some coordinates being +-Infinity or NaN (not-a-number). Obviously, we cannot put these on a grid, so this is usually where we detect those errors. Make sure your system is properly energy-minimized and that the potential energy seems reasonable before trying again." at my 8_cryo_fit step.

  1. Doo Nam observed this error when starting molecule is not stabilized structurally. There are two possible solutions.
  • Run phenix.real_space_refine before cryo_fit. Then, provide real_space_refined molecule into cryo_fit. Then, most problems were solved.
  • When Doo Nam skipped minimization, this error occured. Since most users minimize the starting structure by default (cryo_fit default), most users do not need care about this. Besides, as of 9/18/2019, the default value of number_of_steps_for_minimization=20000 (which is a lot).
  1. It means that the map dimensions need to be larger. Therefore, check map size and solve a problem with VMD/EMAN2/relion like followings.
  • Like other MD simulations, gromacs need enough map box size to cover the atomic model to run (ziggle and wiggle). Refer Waters seems to be out of the box

  • For example, stuck-out red oxygen atoms outside the right edge of the box are the problem.

  • ../images/cryo_fit_FAQ_sticking_out.png
  • In order to run any MD simulation (including cryo_fit), a box should be large enough like

  • ../images/cryo_fit_FAQ_not_stick_out.png
  • Make map box size larger (see "How to enlarge map box size?" in this FAQ), and run cryo_fit again. A user can check map box size by VMD. Alternatively, remove sticking out atoms if these are unnecessary, then run cryo_fit again.

  • For protein modeling, Doo Nam would use cryo_fit2 which is not limited by box size requirement.

I see "The domain decomposition grid has shifted too much" at my 8_cryo_fit step

  • For example, phenix.cryo_fit <input.pdb> <input.ccp4> nproc=2

I see "The initial cell size (xxx) is smaller than the cell size limit (xxx), change options -dd, -rdd or -rcon, see the log file for details" at my 8_cryo_fit step

  1. When he provided a pdb file that has unspecified/deleted region of a molecule. Even when vmd assisted map box/cell dimension is larger than initial molecule space, this error appeared.
  2. When he provided a pdb file that has small number of residues (e.g. 5~16 amino acids). Even when vmd assisted map box/cell dimension is larger than initial molecule space, this error appeared. This error appeared regardless of the existence of CRYST1 and SCALE header.

I see "Too many LINCS warnings" at my 8_cryo_fit step

  1. Run phenix.real_space_refine before cryo_fit. Then, provide real_space_refined molecule into cryo_fit. Then, most problems were solved.
  2. When Doo Nam skipped minimization, this error occured. Since most users minimize the starting structure by default (cryo_fit default), most users do not need care about this. Besides, as of 9/18/2019, the default value of number_of_steps_for_minimization=20000 (which is a lot).
  3. If starting molecule has non-standard chemical geometry that is too extreme that are not stabilized neither by phenix.real_space_refine nor by cryo_fit minimization, then a user needs to fix those non-standard geometry manually.

I see "User's provided atomic model had 0.0 cc" in my cryo_fit.overall_log.

My model is shifted after phenix.model_idealization.

When cryo_fit is useful?

What to do after cryo_fit?