The following examples are included to illustrate the various ways in which REFMAC can be run:
Completing the data to include all possible hkls. Should do this after data reduction, and certainly before using REFMAC. This is now done with the uniqueify script. It is best done using CCP4i.
#! /bin/sh # set -e # # Case (1) # # Complete dataset and add free-R column. # Keep systematic absences with -s switch (though you probably wouldn't # want to do this). # uniqueify -s $CCP4_SCR/toxd_nan.mtz $CCP4_SCR/toxd-unique.mtz # # Case (2) # # Complete dataset and add free-R column. # Increase the fraction of reflections tagged with each free-R # indicator above the default 0.05 (sensible for toxd which has # small dataset). # uniqueify -p 0.1 $CCP4_SCR/toxd_nan.mtz $CCP4_SCR/toxd-unique2.mtz # # Case (3) # # First add free-R column to incomplete dataset. # (Silly thing to do - this is just to create a dataset with an existing # free-R column for illustrating use of uniqueify with -f switch.) # freerflag hklin $CCP4_SCR/toxd_nan.mtz hklout $CCP4_SCR/toxd_free.mtz << eof END eof # # Now complete with -f switch to indicate free-R column already present. # uniqueify -f FreeR_flag $CCP4_SCR/toxd_free.mtz $CCP4_SCR/toxd-unique3.mtz #
Restrained refinement with overall B-factor refinement. Method is sparse matrix method.
#!/bin/csh -f
#
# Example of refinement by refmac
#
set inmtz=toxd-unique.mtz
start:
set name = toxd
set last = 0
set cycles = 1
set count = 0
while ($count != $cycles)
echo '*******************************************************************'
echo $count
echo '*******************************************************************'
@ curr = $last + 1
#
# Refmac
#
refmac:
refmac5 \
HKLIN $inmtz \
HKLOUT $CCP4_SCR/${name}${last}.mtz \
XYZIN $CCP4_SCR/${name}${last}.pdb \
XYZOUT $CCP4_SCR/${name}${curr}.pdb \
<< eor
#
#####Do not add hydrogens
#
MAKE_restraints HYDRogens No
#
#####Do not check correctness of all monomers. Rely on users naming
#####One should be careful in using this option.
#
MAKE CHECk 0
#
####Input mtz labels.
#
LABIN FP=FTOXD3 SIGFP=SIGFTOXD3 FREE=FreeR_flag
#
####Output mtz labels
#
LABO FC=FC PHIC=PHIC FWT=2FOFCWT PHWT=PH2FOFCWT -
DELFWT=FOFCWT PHDELWT=PHFOFCWT
#
####Restrained refinement. Reflections between 20 1.5Å resolution will be used
#
REFI TYPE RESTrained RESOLUTION 20 1.50
#
####Use maximum likelihood residual
#
REFI RESI MLKF
#
####Refine isotropic B values.
#
REFI BREF ISOTropic
#
####Use 0.35 as weighting between X-ray and geometry
#
WEIGHT MATRIX 0.35
#
####Scaling parameters. Use BULK solvent based on Babinet's principle.
####NB: Unless otherwise SOLVENT NO given contribution of bulk solvent
####based on constant value will be used.
#
SCALe TYPE BULK
#
####Fix Babinet's bulk solvent B value to 200.0
#
SCALe LSSCale FIXBulk 200.0
#
####number of refinement cycles
#
NCYC 2
#
####Monitor only overall statistics
#
MONI FEW
end
eor
#
if ($status) exit
# make maps.
#
# Sigmaa style 2mfo-dfc map with restored data
#
fft:
fft \
hklin $CCP4_SCR/${name}${last}.mtz \
mapout $CCP4_SCR/2mfodfc_${name}${last}.mtz \
<<eof
title Sigmaa style 2mfo-Dfc map calculated with refmac coefficients
labi F1=2FOFCWT PHI=PH2FOFCWT
binmapout
end
eof
if ($status) exit
#
# Sigmaa style mfo-dfc map
#
fft \
hklin $CCP4_SCR/${name}${last}.mtz \
mapout $CCP4_SCR/mfodfc_${name}${last}.mtz \
<<eof
title Sigmaa style mfo-Dfc map calculated with refmac coefficients
labi F1=FOFCWT PHI=PHFOFCWT
binmapout
end
eof
if ($status) exit
#
@ last++
@ count++
end
#
Unrestrained refinement by maximum likelihood method.
#!/bin/csh -f
#
# Example of refinement by refmac
#
set CTEST=/y/programs/xtal/ccp4/examples
cp $CTEST/toxd/toxd.pdb $CCP4_SCR/toxd0.pdb
set inmtz=toxd-unique.mtz
start:
set name = toxd
set last = 0
set cycles = 1
set count = 0
while ($count != $cycles)
echo '*******************************************************************'
echo $count
echo '*******************************************************************'
@ curr = $last + 1
#
# Refmac step. Refine
#
refmac:
refmac5 \
HKLIN $inmtz \
HKLOUT $CCP4_SCR/${name}${last}.mtz \
XYZIN $CCP4_SCR/${name}${last}.pdb \
XYZOUT $CCP4_SCR/${name}${curr}.pdb \
<< eor
#
####Input mtz labels
#
LABIN FP=FTOXD3 SIGFP=SIGFTOXD3 FREE=FreeR_flag
#
####Output mtz labels for map calculation
#
LABO FC=FC PHIC=PHIC FWT=2FOFCWT PHWT=PH2FOFCWT -
DELFWT=FOFCWT PHDELWT=PHFOFCWT
#
####Unrestrained refinement. Use reflections between 20 1.5Å
#
REFI TYPE UNREstrained RESOLUTION 20 1.50
#
####Use maximum likelihood residual
#
REFI RESI MLKF
#
####Refine individual atomic isotropic B-values
#
REFI BREF ISOTropic
#
####Scaling parameters. Use Babinet's bulk solvent correction.
# Unless SOLVENT NO given, bulk solvent based on constant value
# will also be used in some cases.
#
#SCALe LSSCale FIXBulk 200.0 # should be added if scaling is not stable
#
SCAL TYPE BULK
#
####Number of refinement cycles
#
NCYC 2
#
####Monitor only overall statistics
#
MONI FEW
end
eor
#
if ($status) exit
# make maps.
#
# Sigmaa style 2mfo-dfc map with restored data
#
fft:
fft \
hklin $CCP4_SCR/${name}${last}.mtz \
mapout $CCP4_SCR/2mfodfc_${name}${last}.mtz \
<<eof
title Sigmaa style 2mfo-Dfc map calculated with refmac coefficients
labi F1=2FOFCWT PHI=PH2FOFCWT
binmapout
end
eof
#
if ($status) exit
# Sigmaa style mfo-dfc map
#
fft \
hklin $CCP4_SCR/${name}${last}.mtz \
mapout $CCP4_SCR/mfodfc_${name}${last}.mtz \
<<eof
title Sigmaa style mfo-Dfc map calculated with refmac coefficients
labi F1=FOFCWT PHI=PHFOFCWT
binmapout
end
eof
if ($status) exit
#
Idealization. Method of minimization is conjugate gradient method.
#!/bin/csh -f # # Example of refinement by refmac # # # Set parameters # set CTEST=/y/programs/xtal/ccp4/examples set crdin=$CTEST/toxd/toxd.pdb set crdout=toxd_1.pdb # # Refmac step. Idealise # refmac: refmac5 \ XYZIN $crdin \ XYZOUT $crdout \ << eor # # Do not include hydrogens # MAKE HYDR N # # Idealisation of the geometry # REFI TYPE IDEAlise # # Number of idealisation cycles # NCYC 2 # # In the first and last cycles information about outliers and # overall statistics will be given. In the intermediate cycles # only information about bond lengthes and bond angles will be # monitored # MONI MEDI end eor if ($status) exit #
Restrained refinement with partial contribution from hydrogens.
#!/bin/csh -f
#
start:
set name = kak1_
set last = 1
set cycles = 1
set count = 0
while ($count != $cycles)
echo '*******************************************************************'
echo $count
echo '*******************************************************************'
@ curr = $last + 1
#
# Run refmac - refinement adding in hydrogen contributions
#
refmac:
refmac5 \
HKLIN $SCRATCH/kak_freer_unobs_hyd.mtz \
XYZIN $SCRATCH/${name}${last}.pdb \
HKLOUT $SCRATCH/${name}${last}.mtz \
XYZOUT $SCRATCH/${name}${curr}.pdb << eop
#
# add hydrogens in their riding positins. It is default.
# Hydrogens will be added in their riding positions and
# they will be used for geometry gradient calculation and in
# structure factor calculation. They are excluded from X-ray graident and
# second derivative calculation.
#
MAKE HYDRogens All
#
# Do not write hydrogens to output pdb file. It is default
#
MAKE HOUT No
#
# Input mtz labels
#
LABI FP=F SIGFP=SIGF FREE=FreeR_flag
#
# Output mtz labels
#
LABO FC=FC_p PHIC=PHIC_p FWT=2FOFCWT_p DELFWT=FOFCWT_p
#
# Monitor all overall statistics and outliers at every cycle
# bond distance outliers will be written to log file if deviation
# from ideal values is more than 5.0 sigma. All other statistics
# are default
#
MONI MANY DIST 5.0
#
#Refinement parameters
#
# Restrained refinement
#
REFI TYPE REST
#
# Maximum likelihood residual. Reflections between 20 and 1.2 A
# will be used for the refinement
REFI RESI MLKF RESO 20.0 1.2
#
# Refine individual isotropic B values
#
REFI BREF ISOT
#
# Weight between X-ray and geometry. Usually at this resolution
# this value should be set to very high value (for example 10 or 20)
#
WEIGHT MATRIX 0.5
#
# Scaling parameters. Use Babinet's bulk solvent scaling.
# Solvent based on constant value in the solvent region is used
# by default.
#SCALE LSSCale FIXBulk 200.0 # should be added if scaling is not stable
#
SCAL TYPE BULK
#
# Number of refinement cycles
#
NCYC 5
#
# Number of resolution bins
#
BINS 20
END
eop
if ($status) exit
#
@ last++
@ count++
end
#
Restrained refinement with maximum likelihood method etc. 3Å data requires fixing of protein and Babinet's BULK B value. In these cases it is better to use TLS refinement before indivdual atomic refinement. See Example 9.
#!/bin/csh -f # set verbose set name = piitrig- set last = 0 set cycles = 1 set count = 0 while ($count != $cycles) @ curr = $last + 1 # refmac: refmac5 \ HKLIN pii_154free.mtz \ HKLOUT $SCRATCH/$name$last.mtz \ XYZIN $SCRATCH/$name$last.pdb \ XYZOUT $SCRATCH/$name$curr.pdb \ << 'END-sfrk' # # Do not include hydrogens even if they are present in the input pdb file # MAKE HYDRogens No # # Labels of input mtz file # LABIN FP=FP SIGFP=SIGFP FREE=FreeR_flag # # Labels of output mtz file # LABO FC=FC PHIC=PHIC FWT=2FOFCWT DELFWT=FOFCWT # # Restrained refinement with reflections between 20 and 3.0 A # REFI TYPE RESTrained RESOLUTION 20 3.0 # # Maximum likelihood residual # REFI RESI MLKF # # Isotropic B value refinement # REFI BREF ISOTropic # # Tight restrained refinement # WEIGHT MATRIX 0.1 # # Damp down shifts at each cycle by 0.5 (positional) 0.5 (thermal) # DAMP 0.5 0.5 !Scaling parameters SCAL TYPE BULK LSSC FIXBulk BBULk 200 ! Fixes Babinet's bulk solvent B SCAL LSSC ANISO !Scales aniso B # # Number of the refinement cycles # NCYC 4 # # It is good idea check geometry in the first and last cycles of the # refinement # MONItor MEDIum 'END-sfrk' if ($status) exit # @ last++ @ count++ end #
Example of rigid body refinement in refmac. Ordinary case with several domains.
#!/bin/csh -f
#
###################################################################
#####################################################################
set name = cytc_refmac_
set inmtz=/y/ccp4/cytc/pseudo_alc_cprime1_sc_feph_sq_sf2_sfx_sq_a+p-unique_pt2p.mtz
set last = 0
set cycles = 1
set count = 0
while ($count != $cycles)
@ curr = $last + 1
refmac5 \
HKLIN $inmtz \
HKLOUT ${name}${curr}.mtz \
XYZOUT ${name}${curr}.pdb \
XYZIN ${name}${last}.pdb \
<< eop
#
# Input mtz labels
LABI FP=Fnata SIGFP=SIGFnata FREE=FreeR_flag
#
# Refinement parameters. Maximum likelihood refinement.
# Reflections between 15 and 2.0Å will be used
#
REFI TYPE RIGID RESI MLKF RESO 15 2.0
#
# Scaling parameters. For rigid body sometimes bulk solvent based
# on constant value could be switched off
#SOLVent NO
#
# Fixing Babinet's bulk solvent parameters sometimes helps to
# stabilise scaling
#
SCALe TYPE BULK LSSC ANIS FIXBulk BBULk 200.0
#
# Rigid body parameters
#
# Number of cycles
#
RIGIdbody NCYCle 10
#
# Domain definition. Each group is one rigid body. It may consist of
# several unconnected pieces of chain
#
RIGIdbody GROUp 1 FROM 2 A TO 32 A
RIGIdbody GROUp 2 FROM 38 A TO 55 A
RIGIdbody GROUp 3 FROM 76 A TO 99 A
RIGIdbody GROUp 4 FROM 101 A TO 126 A
RIGIDbody GROUp 5 FROM 56 A TO 75 A
#
# First cycle will give R, free R over resolution. Last cycle also
# will give rotation matrices and angles and translations
#
MONI MEDIum
END
eop
if ($status) exit
#
@ last++
@ count++
end
#
Same problem but now using experimental phases.
#!/bin/csh -f
set inmtz=/y/ccp4/cytc/pseudo_alc_cprime1_sc_feph_sq_sf2_sfx_sq_a+p-unique_pt2p.mtz
set name = cytc_refmac_
set last = 0
set cycles = 1
set count = 0
while ($count != $cycles)
@ curr = $last + 1
refmac5 \
HKLIN $inmtz \
HKLOUT ${name}${curr}.mtz \
XYZOUT ${name}${curr}.pdb \
XYZIN ${name}${last}.pdb \
<< eop
#
# Input file labels. Experimental phases are present in the
# form of Hendrickson and Lattman coefficients.
#
LABI FP=Fnata SIGFP=SIGFnata FREE=FreeR_flag -
HLA=HLA_pt2 HLB=HLB_pt2 HLC=HLC_pt2 HLD=HLD_pt2
#
# Output mtz labels. They correspond to experimental phases combined with
# caclulated ones:
#
# FWT*exp(i*phwt) = 2.0 mcomb Fo*exp(i*phcomb) - D Fc *exp(i*phcalc)
#
#
LABO FC=FC_DP5 PHIC=PHIC_1 FWT=2FOFCWT DELFWT=FOFCWT
#
# If phases are from Density modification programs
# sometimes it is good idea to blur them. Usually density
# modification programs estimate reliability of phases wuth
# high optimism. Being a little bit sceptic would not harm
#
REFI PHASE BBLUrring 30.0 SCBLurring 0.9
#
# Use maximum likelihood residual between resoltution 15 and 2.0 Å
#
REFI TYPE RIGI RESI MLKF RESO 15 2.0 ! Maximum likelihood refinement
#
# Fixing bulk solvent could give better results. But I would run
# few cycles without fixing them.
SCALe LSSC FIXBulk BBULk 200.0
#
# Use working reflections for SigmaA (likelihood parameters).
# It could be useful at the early stages fo refinement.
#
SCAL MLSC WORK
#
# Fix SigmaA bulk solvent parameters
#
SCAL MLSC FIXBulk BBULk 100.0 SCBUlk -0.1
#
# Do not use bulk solvent based on constant value. You may remove
# it. Then at each cycle program will recalculate contribution
# from bulk solvent
#
SOLVent NO
#
# Rigid body parameters
#
# Number of rigid body refinement cycles
#
RIGIdbody NCYCle 10
#
# Rigid body groups.
#
RIGIdbody GROUp 1 FROM 2 A TO 32 A
RIGIdbody GROUp 2 FROM 38 A TO 55 A
RIGIdbody GROUp 3 FROM 76 A TO 99 A
RIGIdbody GROUp 4 FROM 101 A TO 126 A
RIGIDbody GROUp 5 FROM 56 A to 75 A
#
# Give only overall R and free R value at each cycle
#
MONI FEW
END
eop
if ($status) exit
#
@ last++
@ count++
end
Example of using experimental phase information . Very bad model ( RMS error 2A)
#!/bin/csh -f
#
#!/bin/csh -f
#set verbose
set name = tnc_test_
set last = 0
set cycles = 1
set count = 0
while ($count != $cycles)
@ curr = $last + 1
#goto refmac
#
refmac:
refmac5 \
HKLIN tnc_test.mtz \
HKLOUT ${name}${curr}.mtz \
XYZOUT ${name}${curr}.pdb \
XYZIN ${name}${last}.pdb \
<< eop
#
####Do not add hydrogens.
#
MAKE HYDRogens No
#
####Rely on users naming. Monomer names and atoms in monomers should
####correspond what is in the dictionary.
#
MAKE CHECK 0
#
# Input file lables. Phase are present in the form of Hendrickson and
# Lattman coefficients. They tell program that phased refinement
# should be used
#
LABI FP=Fnati SIGFP=SIGFnati FREE=FreeR_flag -
HLA=HLA_t HLB=HLBt HLC=HLC_t HLD=HLD_t
#
# ALternative to Hendrickson and Lattman coefficients is using
# centroid of phases and their figure of merit
#PHIB=PHIDM_t FOM=FOMDM_t
#
# Output mtz labels for map calculation.
#
LABO FC=FC_main PHIC=PHIC_main FWT=2FOFCWT DELFWT=FOFCWT
#
# Refinement parameters
#
# Restrained refinement. Reflections between 100 and 2.8A will be used
#
REFI TYPE RESTrained RESOlution 100 2.8
#
# Maximum likelihood refinement
#
REFI RESI MLKF
#
# Overall B value refinement
#
REFI BREF OVER ! Refine overall B-values
#
# Tight restraints
#
WEIG MATR 0.15
#
# Scale down shifts at every cycle by factor 0.5
#
DAMP 0.5 0.5
#
# Blur (scale down reliability of) phases
#
REFI PHASed BBLUrring 20.0 SCBLurring 0.7
#
# Babinet's bulk solvent scale parameters. But fix B value
#
SCALe TYPE BULK
SCALe LSSCale FIXBulk BBULk 200.0
#
# Anisotropic scaling. It is default
#
SCALe LSSCale ANISotropic
#
# Fix Babinet's bulkd solvent for sigmaA
SCAL MLSC FIXBulk BBULk 200.0 SCBUlk -0.05
#
# Number of refinement cycles 5
#
NCYC 5
#
# Give more detailed statistics at the first and last cycles.
#
MONI MEDI
#
# Number of resolution bins
#
BINS 20
END
eop
if ($status) exit
#
@ last++
@ count++
end
Example of refinement of individual anisotropic B values. Hydrogens must be included. They will be used for geoemtry gradient and second derivatives and structure factor calculations. But they do not contribute to X-ray gradients and second derivatives
#!/bin/csh -f
#set verbose
#
###########################################################################
# You have to edit labi, reso, name, data #
# #
# data - mtz file #
# name - name of coordinate file #
# reso - resolution of data or resoloution limits you want to refine #
# labi - input file labels #
# #
#######Following things should be optimised according to data used #
# #
# WEIG MATR <number> weighting x-ray and geometry. At low resolution#
# it should be small (sometimes 0.1) at high #
# resolution high (sometimes 6.0 or even higher) #
# SPHEricity restraints on sphericity of atoms. Larger #
# less spherical #
# #
# NB: with current version only CDIR (conjugate direction) method of #
# minimisation is active #
# #
###########################################################################
set name = 'p1lys_'
set last = 0
set cycles = 1
set count = 0
set data = 'p1lys_val-unique'
set reso = '100 0.92'
set spgr = '1'
set labi = 'FP=FO SIGFP=SIGFO FREE=FreeR_flag'
while ($count != $cycles)
@ curr = $last + 1
#goto refmac
#
#
# Run refmac to refine structure. It is refinement with
# anisotropic B values
###################################################################
refmac5 \
HKLIN ${data}_hydr.mtz \
HKLOUT ${name}${curr}.mtz \
XYZOUT ${name}${curr}.pdb \
XYZIN ${name}${last}.pdb \
<< eop
#
# Input mtz labels
#
LABI $labi
#
# Output mtz labels
#
LABO FC=FC_main PHIC=PHIC_main FWT=2FOFCWT DELFWT=FOFCWT
#
# Refinement parameters
#
# Restrained refinement. Resolution has been defined above
#
REFI TYPE REST RESO $reso
#
# Resildual is likelihood function
#
REFI RESI MLKF
#
# Individual anisotropic B value refinement
REFI BREF ANISotropic
#
# Weight X-ray and geometry. Sometimes this value for high resolution
# should be set to very high value. In one case WEIGht MATRix 50.0 gave
# good result
WEIG MATR 4.0
#
# Sphericity restraint
SPHERicity 5.0
SCAL TYPE BULK
#
# Change limits for B values
BLIM 1.0 150.0
#
# Number of refinement cycles
NCYC 5
#
# Monitor only overall statistics
#
MONI FEW
#
# Number of resolution bins
#
BINS 20
END
eop
if ($status) exit
#exit
#
@ last++
@ count++
end
Example of TLS refinement prior to individual isotropic B value and positional parameters.
#!/bin/csh -f
start:
set name = gere_test_
set last = 1
set cycles = 1
set count = 0
#
set title = ' GERE MAD (MAD data only + HL coeff.from mlphare)'
set resol = '6 2.05'
set cell = '109.019 61.749 71.743 90.000 97.082 90.000'
set symm = 'C2'
set xyzlim = '0 0.5 0 1.0 0 0.5'
set data = 'gere_new-rfree.mtz'
#
while ($count != $cycles)
echo '*******************************************************************'
echo $count
echo '*******************************************************************'
@ curr = $last + 1
#
refmac:
refmac5 \
#refmac \
HKLIN $data \
HKLOUT ${name}${curr}.mtz \
XYZIN ${name}${last}.pdb \
XYZOUT ${name}${curr}.pdb \
TLSIN tls.in \
<< eor
#
##### Makecif parameters
# Do not define links automatically
# Do not add hydrogens
MAKE LINK N
MAKE HYDR N
#
# Use NCS.
NONX NCHAIns 6 CHNID A B C D E F NSPANS 1 12 72 5
#
# Input mtz labels
#
LABIN FP=FP SIGFP=SIGFP FREE=FreeR_flag
#
# Output mtz labels
#
LABO FC=FC PHIC=PHIC FWT=2FOFCWT DELFWT=FOFCWT
#
# Parameters of bulk solvent based on constant value. They are default
#
SOLVENT VDWP 1.4 IONP 0.80 RSHR 0.8
#
# Weignting X-ray and geometry. Fairly tight restrants
#
WEIG MATR 0.2
#
# This keyword says program that use 20 cycles of tls before
# going to individual atomic reifnement. Parameters of rigid
# body groups are defined in TLSIN tls.in
REFI TLSC 20
#
# It is good idea to set all B values to some value prior to
# TLS refinement
#
BFACtor SET_to 40
#
# Use restrained refinement after TLS. Reflections between 20 and 2.05
# will be used
REFInement TYPE RESTrained RESOLUTION 20.0 2.05
#
# Maximum likelihood refinement
#
REFInement RESIdual MLKF
#
# After TLS use istotropic B values refinement. In very low resolution
# it could be set to Overall B value refinement
#
REFInement BREFinement ISOTropic
#
# Damp down shifts
3
DAMPing factor 0.5 0.5
#
# First and last cycles will print information about outliers
# In all other cycles only overall statistics will be monitorred
#
MONI MEDI
#
# Fix Babinet's bulk B value to 200.0. Usually it stabilises refinement
#
SCALE LSSCale FIXBulk BBULk 200
#
# Use Aanitoropic scaling. It is default
#
SCAL LSSC ANISOT
#
# Use Babinet's bulk solvent also.
SCAL TYPE BULK
#
# Number of individual atomic refinement 8
NCYC 8
eor
if ($status) exit
end