Neutron structure determination

(1)

Neutron structure determination



 

An X-ray structure is always available and is used as a starting model for

neutron structure determination, therefore once the neutron data is available,

the main steps to get final refined model are:

-

 

model preparation (add H or D, or H/D; library files for ligands),

-

 

model refinement and completion (adding DOD, OD or O),

(2)

Model preparation



 

Model preparation (add H or D, or H/D; library files for ligands),

-

 

add H or D, or H/D,

-

 

library files for ligands (CIF file)



 

Solution in PHENIX:

-

 

add H to everything and create ligand CIF file (if any)

phenix.ready_set

model.pdb

-

 

add D to everything and create ligand CIF file (if any)

phenix.ready_set

model.pdb

perdeuterate=true

-

 

add H and exchangeable H/D:

phenix.ready_set

model.pdb

neutron_exchange_hydrogens=true

-

 

To see all options of phenix.ready_set:

phenix.ready_set --show-defaults

-

 

To remove all H and D atoms:

(3)

Model preparation

  phenix.pdbtools – set of tools for PDB file manipulations

  For any selected model part:

- shake coordinates, ADP, occupancies - rotation-translation shift of coordinates

- shift, scale, set ADP (add, multiply, assign a constant) - converting to isotropic / anisotropic

- removing selected part of a model   Easy to run:

% phenix.pdbtools model.pdb set_b_iso=30 selection="chain A” This will sell all B-factors to 30 for all atoms in chain A.

  Also:

- complete model statistics (geometry, B-factors) - geometry regularization

- output MTZ with complete calculated structure factors (that may include all scales and bulk solvent)

(4)

Model refinement



 

Designed to be very easy to use

(5)

Some basic examples of running phenix.refine from the command line



 

Refinement of individual coordinates, B-factors, and occupancies for some

atoms:

phenix.refine

model.pdb

data.hkl



 

Add water picking and Simulated Annealing to default run above:

phenix.refine

model.pdb

data.hkl

simulated_annealing=true

ordered_solvent=true



 

Refinement of individual coordinates and B-factors using neutron data:

phenix.refine

model.pdb

data.hkl

main.scattering_dictionary=neutron



 

To see all parameters (more than 300):

(6)

Running phenix.refine

%

phenix.refine

model.pdb

data.hkl

parameter_file

where parameter_file contains following lines:

refinement.main { high_resolution = 2.0 low_resolution = 15.0 simulated_annealing = True ordered_solvent = True number_of_macro_cycles = 5 } refinement.refine.adp { tls = chain A tls = chain B }

For typing enthusiasts, the equivalent command line run would be:

% phenix.refine model.pdb data.hkl xray_data.high_resolution=2 xray_data.low_resolution=15 simmulated_annealing=true

ordered_solvent=True adp.tls=“chain A” adp.tls=“chain B” main.number_of_macro_cycles=5

(7)

Typical way of phenix.refine run from the command line

1.  Get the file with all parameters:

%

phenix.refine

--show-defaults=all > parameter_file

2.  Edit the file parameter_file:

-  Remove all parameters that you are not planning to change (make sure to have all { } matched)

-  Change the rest of parameters 3.  Run phenix.refine as following:

%

phenix.refine

model.pdb

data.hkl

parameter_file

or (If model.pdb and data.hkl are included into parameter_file file)

%

phenix.refine

parameter_file

Useful tip: to compare the set of parameters in your parameter_file file against the set of all default parameters, type:

(8)



 

When running:

%

phenix.refine

model.pdb

data.hkl

each item in

model.pdb

is matched against the CCP4 Monomer Library to

extract the topology and parameters and to automatically build corresponding

restraints.



 

If

model.pdb

contains an item not available in CCP4 Monomer Library, e.g.

a novel ligand, use

ReadySet!

program to generate topology and parameter

definitions for refinement:

%

phenix.ready_set model.pdb

This will produce the file

LIG.cif

and updated PDB file

_{model.updated.pdb}

with all H atoms added which can be used for refinement:

%

phenix.refine

model.pdb

data.hkl

LIG.cif

(9)

Restraints and novel ligands: REEL

%

phenix.reel LIG.cif

(10)

Refinement runs required two steps: twinning



 

Two steps to perform twin refinement:

- run

phenix.xtriage

to get twin operator (twin law):

% phenix.xtriage

data.mtz

- run

phenix.refine

:

(11)

Individual neutron and joint X+N refinement

Running joint X/N refinement:

% phenix.refine

model.pdb

parameter_file

refinement.input.neutron_data { file_name = data.mtz labels = "F-obs-neutron,SIGF-obs-neutron" r_free_flags { file_name = data.mtz label = "R-free-flags-neutron" } } refinement.input.xray_data { file_name = data.mtz labels = "F-obs,SIGF-obs" r_free_flags { file_name = "data.mtz" label = "R-free-flags" } }

(12)

Important remark about free-R flags in joint X+N refinement

•  Free-R flags in both datasets (X-ray and neutron) must match, that is: what is test reflection in X-ray dataset, must be a test reflection in neutron dataset. Same for work reflections.

• phenix.refine automatically checks for this, and if there is a mismatch, phenix.refine stops and offers two options:

-  keep going with existing flags (but keep in mind that R_free will be unreliable); -  create new free-R flags that will be consistent for both datasets.

(13)

Model refinement - output



 

Input command

phenix.refine

model.pdb

data.mtz

[parameters]



 

Output files

model_refine_001.eff

summary of all input parameters

model_refine_001.geo

summary of all restraints used

model_refine_001.log

complete information about refinement

model_refine_001.pdb

refined structure

model_refine_001_map_coeffs.mtz

Fourier map coefficients

model_refine_002.def

parameters for the next run

If data file is not in MTZ format, or there are multiple data files at input

(example: one with Fobs and the other one with free-R flags), then

phenix.refine will combine them into one MTZ data file called:

model_data.mtz

and this file should be used in all subsequent runs.

(14)

Model validation

•  Inspect .log file created by phenix.refine.

•  From PHENIX GUI run “Comprehensive validation” •  From the command line type:

phenix.model_vs_data

model.pdb

data.mtz

If

_data.mtz

contains several data columns, then

phenix.model_vs_data

model.pdb

data.mtz

--f-obs-label=Fo

In case of neutron data:

phenix.model_vs_data model.pdb data.mtz --scattering-table=neutron

(15)

Reporting bugs, problems, asking questions

 Something didn’t work as expected?... program crashed?... missing feature?...

Bad: silently give up and run away looking for alternative software.

Good: report us a problem, ask a question, request a feature (explain why it’s good to have), ask for help (send data).

 Reporting a bug / problem:

Bad: “Hi! phenix.refine crashed and I don’t know why and what to do. Goodbye.” Good: “Hi! phenix.refine crashed. Here are:

1) PHENIX version;

2) The exact command I used;

3) Input and output files (at least logs).”

(16)