Molecular Replacement
Ronan Keegan
Molecular Replacement and the
Phase problem
• Molecular replacement is the process of solving the phase
problem for an unknown structure by placing the atomic
model for a related, known structure in the unit cell of the
unknown structure in such a way as to best reproduce the
observed structure factors.
• The known model, once placed, may be used to calculate
phases which, in combination with the observed structure
factors for the unknown structure, allow the model to be
rebuilt and refined.
• The calculation involves a 6 dimensional search over all
possible orientations and translations of the known model
in the unit cell of the unknown structure.
• This calculation is generally too time consuming to perform
in full, so it is usually split into two parts:
– A 3 dimensional search over all possible orientations to determine
the orientation of the model.
– A 3 dimensional search over all possible translations to determine
the position of the orientated model.
Solving structures using
Molecular Replacement
• In 2006, over 67% of structures deposited in
the PDB where solved by MR
Before you start
• Number of molecules to search for
– Matthews_coef: Given molecular weight of the target
and the dimensions of the target cell it will return a set
of probabilities for the number of molecules in the
asymmetric unit
• Check for twinning
– Ctruncate or Sfcheck can be used to examine the data
to assess the likelihood that the data is twinned
• Is there NCS? – helps with refinement – use self
rotation function
Finding a search model
• Use online resources such as the EBI or OCA services for
doing FASTA sequence matching searches
– http://www.ebi.ac.uk/Tools/fasta33/
– http://oca.weizmann.ac.il/oca-bin/ocamain
• Use locally held related PDB models if available
• Use secondary structure matching based on best scoring
models from sequence based search
– http://www.ebi.ac.uk/msd-srv/ssm/
• Look for domains components (SCOP) or multimeric forms
of search models (PISA)
What makes a good search
model
• General rule-of-thumb is that the sequence identity of
homologue to the target must be > 30 % for the process to
work
• Where sequence identity is low it is important to get as
good a sequence alignment as possible
– Use multiple alignment pulling in many related sequences
rather than pair-wise
– Profile fitting alignment e.g. Blast
Preparing your search model
• Signal-to-noise problem – anything that is in your model
that is not likely to be in your target structure needs to be
removed as it will only contribute to the background noise
– Prune back side-chains that aren’t aligned
– Cut out flexible loops
– Cut out waters
• Various programs in CCP4 to help do this
– Chainsaw – prunes side chains based on a given alignment
– Molrep – creates its own alignment and prunes side chains
accordingly
– Cutting loops: look at B-factors, if above an acceptable threshold
cut out those residues using PDBCur
– Removing waters and other small molecules: use PDBset
CCP4 and MR
• Molecular Replacement
programs:
– Molrep
– Phaser
– Amore
• Automated MR:
– Balbes
– MrBUMP
• Helper Applications:
– Matthes_coef, Chainsaw, Pdbcur,
Pdbset, Coordformat, Superpose,
PISA
Molrep
• Molrep is program for automated
molecular replacement in the case
where a homologous structure has
already been identified.
• The program will attempt to find the
number of molecules expected in the
asymmetric unit as entered by the user.
• A PDB file for the best solution is
output.
• Additional options
– Self rotation function
– Search for model in a map
– Alignment only
• Can perform individual steps in more
difficult cases
Molrep Output
• Look at the output log file
– Examine RF scores
– Examine TF scores
– Look at the Contrast score
• Check to see if the number
of molecules asked for have
been found
• Output PDB file will contain
the best positioned model
Cross Rotation Function
Euler angles (CCP4)
polar angles
List of top
RF peaks
More details here
R factor
R factor
Score
Translation Function
polar angles
List of top
solutions:
contrast of solution
fractional
translation
Phaser
• Based on the use of
maximum likelihood
methods to find a solution
for the phases
• Likelihood measures the agreement of the model with the
data by using probabilities
• Allows use of Ensembles as search models in MR,
improving the chances of finding the correct position for
the template
Running Phaser
• CCP4i GUI
• Highly automated – input SF’s in
MTZ format, template search
model(s) and sequence information
• Can perform search over all possible
alternative space groups
• Specify each component of
Ensemble
• Specify composition of asymmetric
unit – number of molecules and
searches for hetrogeneous
complexes
Phaser Output
• Z-scores for rotation and
translations functions
– Translation score > 5 = good
solution
– < 5 and > 3 = potential to be a
good solution
– > 3 = poor solution
• LLG Scores – log likelihood gain
score – higher value = greater
confidence in result
• Top positioned PDB file output
along with corresponding MTZ
including phase columns
Balbes
• Automated MR from model selection
through to initial refinement
• Balbes has its own cut-down version of the
PDB to use for searching for potential
search models for a target.
Accessing Balbes
• Balbes can be accessed via
the YSBL web service page
at York University.
• Create an account and
upload your MTZ file and
sequence information
• Also possible to run Balbes
locally via CCP4i. Requires
installation of the Balbes
data base on the local
machine
http://www.ysbl.york.ac.uk/YSBLPrograms/index.jsp
Web Server Results
• Summary of processing for each
spacegroup
• Final best result highlighted
• For each spacegroup log file and
all output files are made available
for download (5 days)
• If user opted to use ARP/wARP
server a link to the ARP/wARP
results is provided
Balbes Output
• Spacegroup specific
output:
– Download files
– Main summary file
showing results of
MR and refinement
for each template
model that was
used.
– Q value scoring
MrBUMP
• An automation framework for doing Molecular Replacement
• Brute force approach that puts a particular emphasis on generating a
variety of search models
• It can use both Molrep and Phaser for MR
• Uses a variety of helper applications to find and prepare
search models (FASTA , Clustalw, Chainsaw)
• To source search models it access various online
databases such as the PDB and the SCOP database.
• In favourable cases it will give a one-button solution
• In complicated cases it will suggest likely search models
for further manual investigation (lead generation)
The MrBUMP Pipeline
Target MTZ
&
Sequence
Process Target Details
Template Model Search
N templates
Model Preparation
N x M models
Check scores
and exit or
select the next
model
Molecular Replacement
Refinement
Phase Improvement
Search for and preparation of
templates
Domain 1
• Search step automatically performs
FASTA search, multiple alignment of
results and searches for domains and
multimers based on the hits from the
FASTA search
Domain 2
• Best templates are then prepared using
several methods:
– Molrep model alignment and side chain
pruning
– Chainsaw side chain pruning
– Polyalanine model
Ensemble
• An Ensemble model is also created for
Phaser
MrBUMP CCP4i interface
• Select mode of
operation
• Input files and column
label details
• Template search options
• Search model
preparation options
• Molecular Replacement
and Refinement options
• Additional options
Summarised results..
Best search model so
far and file location for
this model
List of sorted
results so far
MrBUMP Output
• Log file gives summary of models tried and
results of MR
– May get several putative solutions
– Ease of subsequent model re-building, model
completion may depend on choice of solution
– Worth checking “poor” solutions
• Top solution available from ccp4i
Summary
• Molecular replacement is the process of retrieving the
phase information for a target structure using a related,
known structure
• CCP4 provides several programs and helper tools perform
MR
• If you are having difficulty try them all!
• If no search model is obvious use Balbes or MrBUMP to do
the work for you
• Before experiment use MrBUMP or Balbes “Model Search”
modes to check to see if good models are available
Tutorial Material
• Tutorial Document:
– CCP4-Workshop\Tutorial\MR\APS-MrBUMP-tutorial-2010.pdf
• Tutorial Data:
– $CCP4\examples\mr_tutorial\2006
• MR and MrBUMP
– Step-by-step guide to doing MR using Molrep and Phaser
– Automated using MrBUMP
• BALBES:
–
–
–
–
YSBL web server:
http://www.ysbl.york.ac.uk/YSBLPrograms/index.jsp
Create an account
Upload MTZ and sequence
Acknowledgements
•
•
•
•
Alexi Vagin, Garib Murshudov and Fei Long
Randy Read, Gabor Bunkoczi, Airlie McCoy
Martyn Winn, Norman Stein
CCP4 group