Skip to content

Latest commit

 

History

History
125 lines (89 loc) · 4.09 KB

README.md

File metadata and controls

125 lines (89 loc) · 4.09 KB
Raw

Objective and steps

a) Select structure-candidates.

  • Select 25 P1 structures from PDB and prepare them for quantum refinement.

b) Run cheap HF energy and gradients calculation using clustering.

  • pick up results from "a)" and use it as input to run cheap HF calculation using clustering.

  • write only gradient-based LBFGS (the line-search version)

c) Final HF based refinement.

  • Analyse results of "b)" and pick one or two most favorable (showing most improvement) to run HF based refinements.

Details for each step a-b-c) follow below.

Structure preparation

1) Get structures from PDB.

  • This is done manually using RCSB site. Selection criteria are: P1, not too many atoms, resolution ~3-4 Å, bad validation metrics. The criteria are rather arbitrary: we try to find low-resolution models that we believe can be improved by quantum refinement.

  • File were downloaded using phenix.fetch_pdb PDB_CODE --mtz

  • Some models have unknown to Phenix ligands. Corresponding CIF files were created using phenix.ready_set file_name.pdb This also adds hydrogens with are needed for the run_finalise.py script. The output is file_name.updated.pdb which needs to be renamed? We can use phenix.reduce which has it's own problems.

  • Some data file were missing R-free flags. They were added at subsequent (re-refinement).

  • .pdb, .mtz and .cif files are stored in 00 folder.

2) Phenix-Refine

  • Refine structures from "1)" using phenix.refine. List of commands per each structure is in 01_run_phenix_refine file.

  • Refinement results are stored in 01 folder. MTZ files from this folder are to be used in all subsequent refinements.

3) Finalize

  • Run structure from "2)" through 02_run_update_pdb.py using command phenix.ready_set is used to add hydrogens to amino acids and ligands run_finalise.py is used to complete structures, remove altlocs and resset occupancies all completed structrues by 02_run_update_pdb.py are in 02

XXX This needs a careful and clear description what's being done and how. Adding H and missing non-H atoms? Remove altlocs? Reset occupancies? XXX

All PDBs in folder 01 have no errors in running 02_run_update_pdb.py are listed in folder 02.
9 out of 25 structures in folder 01 pass 02_run_update_pdb.py successfully.

Current issues to be resolved ASAP:

  • Errors that have been postponed or manually resolved:

    • 1u0d - multiple models
    • 2jee - has a terminal amino acid with just a N

  • Known errors:

    • 3kyi - nonstandard amino acid has strange charge
    • 4k2r - ANP does not have partial charges
    • 4rnf - bug in cctbx geometry restraints
    • 3oe9 - ITD
    • 3uds - ADP
    • 4ctd - C8E

  • Unknown errors:

    • 2x10 - tardy errors, maximum charge error
    • 3nak - strange nonstandard amino acid

  • No errors:

    • 1il5 - ✅
    • 1va7 - ✅
    • 1y1l - ✅
    • 1ok9 - ✅ ACT, GOL added to GeoStd
    • 2oeq - ✅
    • 2oy0 - ✅
    • 2ghj - ✅
    • 3tz9 - ✅ AQU added to GeoStd
    • 3dtj - ✅
    • 3uj4 - ✅
    • 4xa1 - ✅
    • 4drw - ✅
    • 4fsx - ✅
    • 4rnf - ✅
    • 4p7h - ✅
    • 5diz - ✅ Has S-S bond, better check them.
    • 5d12 - ✅ G97 added to GeoStd

  • Hi/Lo pairs

    • 3dtj - works in my hands

  • Waters

    • Do we make a policy about the inclusion of water below a certain resolution?

  • Program completeness

    • need complete list of D-peptides
    • need complete list of nonstandard peptides

Testing

Testing is very simple. Run tests/run_tests.py and check for Sorry's.

  • Tests to add:
    • Carbohydrates are correctly charged - taken from Chemical Components but polymerisation is not checked
    • Test that a ligand has partial charges

Structure refinement

XXX more results here, in folders 04, 05, etc.