New features in v9.09:
- The log likelihood values (LogP) in the parameter files
are now not negated anymore.
New features in v9.08:
- New scripts to run Frealign. Run INSTALL for instructions
on how to install scripts. After installation, type
frealign_help
for information on available run scripts.
- New flag FDUMP (replacing old flag FSTAT): If set to T,
Frealign will dump intermediate files from a 3D
reconstruction into a file for later merging with new
program merge_3d. This allows parallelization of 3D
reconstruction over many CPUs with speedup factors of
10 or more. See example script included with the
distribution.
New features in v9.07:
- New Card 7 parameter RCLAS: high-resolution limit used
for classification. It should typically be set to the
same resolution limit used also for the refinement, or a
bit lower.
New features in v9.06:
- Memory management: The old flag IBLOW has been replaced
by a new flag IMEM. For refinement in Mode 1, 2, 3 or 4
several instances of Frealign are usually run to refine
several parts of a stack of particles in parallel. For
the refinement, the non-mp version of Frealign should be
used (no _nm at the end of the executable). In previous
versions, setting IBLOW = 4 accelerated the processing
but required additional memory. To get the same speedup
and memory usage as before, set IMEM = 1 or IMEM = 3
(no difference if no reconstruction is calculated).
Reconstruction can also use parallelization (usually done
with Mode 0). The script running Frealign should include a
line like this:
setenv NCPUS 8
to set the number of parallel CPUs to be used. There is
now a choice of how Frealign does the parallelization. If
IMEM = 0 (or 1 but this makes no difference if no
refinement is done), it parallelizes the way it used to.
For small particles, it is recommended to use 4 or 8 CPUs
as additional CPUs will probably not make it run faster.
If IMEM = 2 (or 3, again no difference unless refinement
is done) Frealign uses a new parallelization scheme that
should run faster but requires more memory. For this
scheme, 8 or 16 CPUs are recommended provided there is
sufficient memory available (try it and monitor usage).
- Maximum likelihood multi-reference refinement
- Helical symmetry operator HP: allows the processing of
helical filaments with a seam, such as microtubules (the
additional 'P' stands for 'pseudo').
- Weighted correlation coefficient now uses SSNR table
from the previous iteration for more accuracy.
- Option for nearest-neighbor and trilinear interpolation
used in 3D reconstruction (INTERP = 0 or 1).
- IMPORTANT: Shifts are now measured in Angstroms (see below).
- Each particle is now evaluated in terms of a score. This
replaces the previous "phase residual". This means that a
high number now indicates a good fit with the reference
(previously, a high number indicated a bad fit). This
should be considered when setting thresholds to exclude
particles from a reconstruction.
- The user has to enter an approximate molecular mass of the
particle (parameter MW in CARD 2) to calculate the
optimal filter for the 3D reconstruction (option FFILT).
New features in v8.10:
- Implemented single particle Wiener filter (use old FCREF
flag to activate; FCREF now called FFILT).
- Implemented B-factor correction of the final output map.
The B-factor is determined from a line fit to the log of
the map power spectrum. Use new FBFACT to activate.
New features in v8.09:
- Added helical symmetry operator (H): If used, an
additional input line right after CARD 5 is required
with 5 numbers (see helical refinement example included
with the distribution):
ALPHA, RISE, #SUBUNITS, #STARTS, STIFFNESS
ALPHA The rotation angle in deg around the helical
axis between subunits.
Example for TMV test data: 22.03
RISE The shift in Angstrom along the helical axis
from one subunit to the next.
Example for TMV test data: 1.408
#SUBUNITS The number of unique subunits per helical
segment. This will depend on how the helical
filaments were segmented. If segments are
taken with a step size D, then
#SUBUNITS = #STARTS x D / RISE
Example for TMV test data: 49
#STARTS The number of starts in the helix.
Example for TMV test data: 1
STIFFNESS A parameter that determines how strongly the
PSI and THETA angles of each helical segment
are restrained to the average PSI and THETA
angles of the helical filament. Reasonable
values are between 1 (weak restraint) and 100
(very strong restraint).
Example for TMV test data: 20.0
The included program set_polarity can be used to check that
the PSI angles of each segment point all segments into a
consistent direction to ensure uniform polarity along a
filament. A tolerance for PSI deviations from the average
PSI angle can be provided. set_polarity reads a parameter
file and adjusts the PSI angle of segments that fall outside
this tolerance. Also, filament polarity can be flipped
depending on a phase residual (PRES) threshold.
New features in v8.08:
- Now requires Fortran 90 compiler (gfortran, pgf90)
- Dynamically allocated memory: Parameters for NN1 and
NNBIG are now set automatically
- New order of control flags in Card 1
- New flags: FSTAT (T or F) to switch on/off additional
statistics (requires more memory); IBLOW (1,2 or 4) to
specify size of padded reference structure (IBLOW=4
requires the most memory but results in the fastest
search & refinement
New features in v8.07:
- Removed FLIP option (Card 1)
- Added flag FPART for defocus refinement of individual
particles (Card 1)
- Added restraining function for defocus refinement (active
when FPART=T): Set expected defocus uncertainty with
parameter DFSIG (Card 7)
New features in v8.06:
- Memory saving (IFSC flag on CARD1)
- Multiprocessor support (requires Portlan Group compiler)
>>> Need to compiler with _mp Makefiles and set NCPUS
environmental variable to the number of cores to be used
- Output of the two maps containing half the data each
- New I/O subroutines
- IMAGIC image format implemented
- x,y shift parameter distrubution function used for restraints
- Optional use of FFTW
FREALIGN notes for V8.00 Niko 1.2.08
FREALIGN notes for V7.00 Niko 1.2.06
FREALIGN notes for V6.00 Niko 9.3.02
FREALIGN notes for V5.00 RH 2.9.01
FREALIGN notes for V4.06 RH 27.8.01
FREALIGN notes for V4.02 RH 9.8.01
FREALIGN notes for V4.00 RH 22.4.01
FREALIGN notes for V3.05 RH 25.2.01
FREALIGN notes for V3.04 RH 29.6.00
FREALIGN notes for V3.03 RH 10.5.00
FREALIGN notes for V3.02 RH 14.3.00
FREALIGN notes for V3.01 RH 27.8.99
FREALIGN notes for V3.00 RH 27.7.99
FREALIGN notes for V2.07 RH 4.7.99
FREALIGN notes for V2.06 RH 1.6.99
FREALIGN notes for V2.05 RH 21.2.99
FREALIGN notes for V2.03 RH 2.2.99
FREALIGN notes for V2.02 RH 24.1.99
FREALIGN notes for V2.01 RH 4.1.99
FREALIGN notes for V2.00 RH 24.12.98
================================================
FREALIGN carries out search and refinement of particle parameters, CTF
correction (allowing for astigmatism), and 3D reconstruction using
interpolation in Fourier space. The magnification of each dataset and the
defocus and astigmatism values of particles grouped by their film number can
also be refined. Reconstructions can be corrected for Ewald sphere curvature.
On output, there are also diagnostic data, such as Fourier Shell Correlation,
Fourier Shell Phase Residual, Q-factors in resolution zones, Average Particle
Phase Residual between particles and 3D reconstruction, variance of the
reconstruction, and point spread function indicating resolution in 3 dimensions.
FREALIGN uses MRC, SPIDER or IMAGIC image file formats. The MRC
mapformat is identical to that used by CCP4 crystallography programs. Current
array dimensions provide space for 256x256 pixel images (set parameter NN1 in
frealign_v6.f) and the program is limited at present to transforms of even
dimensions.
Some of the principles of FREALIGN are explained in:
1) N. Grigorieff (1998), J. Mol. Biol. 277, 1033-1046.
2) Stewart, A. & Grigorieff, N. (2004), Ultramicroscopy 102, 67-84.
3) Wolf, M., DeRosier, D. J. & Grigorieff, N. (2006), Ultramicroscopy 106, 376-382.
4) Sindelar, C. V. & Grigorieff, N. (2012), J. Struct. Biol. 180, 26-38.
5) Lyumkis, D. & Brilot, A. F., Theobald, D. L. & Grigorieff, N. (2013), J. Struct. Biol. 183, 377-388.
********************************************************************************
Input cards : cards 1 to 18, including either 10a or 10b are always needed.
=========== cards 5a and 18a are used rarely only if appropriate flag is set
Card 1 describes the overall program flow control logic cards
Cards 2 to 5 are global parameters governing all the data for the run
Cards 6 to 12 form a repeatable group, one set describing each dataset which
may be made up of any number of particle images from different films
Cards 13 and 14 are filenames for the 3D structure I/O (overwrites input)
Cards 15 to 18 are filenames for 3D diagnostic output files (18a is backup)
Card 1 CFORM,IFLAG,FMAG,FDEF,FASTIG,FPART,IEWALD,FBEAUT,FFILT,FBFACT,FMATCH,IFSC,FDUMP,IMEM,INTERP
Card 2 RO,RI,PSIZE,MW,WGH,XSTD,PBC,BOFF,DANG,ITMAX,IPMAX
Card 3 PMASK,DMASK - [0/1] parameters to include in refinement [1 1 1 1 1]
Card 4 IFIRST,ILAST - FIRST, LAST PARTICLES
Card 5 ASYM - symmetry required Cn,Dn,T,O,I,I1,I2,N or H (can be zero)
Card 5a only if N and N.ne.0, ((SYMOP(J,K,I), J=1,3), K=1,3), I=1,N
Card 5b only if H, ALPHA, RISE, #SUBUNITS, #STARTS, STIFFNESS
If H is followed by P ("HP"), Frealign will not reset the shift
parameters to be within one helical asymetric unit. This can be
useful when processing pseudo-helical filaments such as
microtubules with a seam.
Cards 6 to 12 describe each dataset, terminating with RELMAG=0 (or negative)
T Card 6 RELMAG,DSTEP,TARGET,THRESH,CS,AKV,TX,TY
| Relative magnification to apply (1.0)
| Densitometer step size in microns (7.0)
| Target phase residual for search/refine (15.0)
| Worst phase residual for inclusion (90.0)
| CS, KV (2.0, 120.0)
| Beam tilt in X, Y direction (0.0, 0.0)
| Card 7 RREC,RMAX1,RMAX2,DFSTD,RBFACT - map resoln, refine low/high,
| defocus uncertainty, B-factor
| Card 8 FINPAT1 - PARTICLE IMAGE STACK FILENAME
| Card 9 FINPAT2 - MATCHING PROJECTIONS STACK (for O/P if FMATCH=T)
| Card 10 - at least one of the following is required
| Card 10a FINPAR - INPUT PARAMETER FILE, required if mode key=0,1,2,3,4
| Card 10b NIN,ABSMAGPIN,IFILMIN,DFMID1IN,DFMID2IN,ANGASTIN,MORE if mode<0
| number of particles, magnification, film number, defocus
| parameters for this dataset. If MORE='1', then more cards
| of this type follow, until MORE=0 terminates the parameter
| data. The information on these cards is used to create a
| new parameter file.
| Card 11 FOUTPAR - OUTPUT PARAMETER FILE
| Card 12 FOUTSH - OUTPUT PARAMETER SHIFTS FILE, File name for file
| containing the parameter shifts for this refinement
|______ cycle (should become smaller if refinement converges).
Card 13 F3D - 3D MAP FILE FOR INPUT and OUTPUT, Input 3D reference
reconstruction, overwritten by output reconstruction!
Card 14 FWEIGH - 3D WEIGHTS FILE FOR INPUT and OUTPUT, Input 3D file
containing the sum of weights (as defined in the JMB
reference), will be overwritten with new file!
Card 15 MAP1 - 3D MAP FILE FOR OUTPUT, containing only odd-numbered
particles (if IFSC = 0, 1), even-numbered particles
(IFSC = 2) or all particles (IFSC = 3).
Card 16 MAP2 - 3D MAP FILE FOR OUTPUT, containing only even-numbered
particles (if IFSC = 0), or no output (if IFSC = 1,2,3).
Card 17 FPHA - 3D PHASE RESIDUAL FILE FOR OUTPUT, Output 3D file with
average phase differences for each voxel (in Fourier
space) between images and reference.
Card 18 FPOI - 3D POINT SPREAD FUNCTION FOR OUTPUT, Output 3D file
with point spread function indicating anisotropies
in resolution (in real space).
++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
Parameters as read in on above cards 1 to 7 and 10b (other cards are filenames)
================================================================================
CFORM M/S/I Input/Output image file format: M=MRC, S=Spider, I-Imagic
IFLAG 0/1/2/3 Mode key. -4:bootstrap parameter file, then IFLAG=4
-3:bootstrap parameter file, then IFLAG=3
0:Reconstruction only parameters as read in
1:Refinement & Reconstruction
2:Random Search & Refinement
3:Simple search & Refine
4:Search,Refine,Randomise & extend to RREC
-4 = create a parameter file from scratch, then IFLAG=4
-3 = create a parameter file from scratch, then IFLAG=3
0 = use previously determined particle parameters
and calculate a 3D reconstruction;
1 = carry out refinement and reconstruction starting with
previous roughly determined parameters
NOTE: If RELMAG in the termination line in card 6
is set to a negative value instead of 0.0 no 3D
reconstruction is calculated (refinement only). This
is useful when processing smaller parts of the data
stack on a computer cluster. A reconstruction is then
done with another run and CFORM=0 (see example scripts).
2 = carry out refinement with randomly assigned particle
parameters (to check if current particle parameters
correct; if correct then all other parameters should
give worse phase residual)
3 = carry out systematic parameter search for initial
assignment, with subsequent refinement.
4 = systematic search & refinement of particle orientation
parameters, with a randomised ITMAX loop to speed up
convergence (only tries randomisation until phase
residual goes below TARGET). Then, resolution is
subsequently extended step-by-step out to RREC - if
residual never goes below TARGET, earlier versions of
the program set the residual to 180 degrees (it could
then be used in subsequent refinement
as flag with TARGET/THRESH = negative (essentially
a flag for complete and permanent subsequent exclusion
of that particle image)). This option was most useful
for adding data to an existing structure analysis as
resolution is gradually improved, but it has now been
removed so that the residual is always the real value.
FMAG T/F Magnification refinement - one number is calculated for each
new film. The orientation parameters must already be
available in the input parameter file and should
preferably have been determined using a different 3D
reference model than that used for the magnification
refinement.
FDEF T/F Defocus refinement, DF1 and DF2 coupled/uncoupled if FASTIG=F/T
FASTIG T/F Astigmatism refinement - if F, then DF1 and DF2 are coupled
FPART T/F Defocus refinement for individual particles if FPART=T, otherwise
defocus change is constrained to be the same for all
particles in one image
IEWALD 0/1/2/-1/-2 Ewald correction:
0 = No correction
1 = Do correction, simple insertion method (see publication #3
above).
2 = Do correction, reference-based method (see publication #3
above).
-1 = 1 but with inverted handedness (see publication #3 above).
-2 = 2 but with inverted handedness (see publication #3 above).
FBEAUT T/F Apply extra real space symmetry averaging and masking to
beautify final map just prior to output.
FFILT T/F Apply single particle Wiener filter to final reconstruction
(see publication #4 above).
FBFACT T/F Determine and apply B-factor to final reconstruction
FMATCH T/F Write out matching projections after the refinement (for
diagnostic purposes): this will have spacegroup 0 in MRC header.
Card 9 (whether or not FMATCH is set) describes FINPAT2,
the matching projections image stack filename.
IFSC 0/1/2/3 Calculation of FSC tabel:
0 = Internally calculate two reconstructions with odd and even
numbered particles and generate FSC table at the end of the
run.
Options 1, 2 and 3 reduce memory usage:
1 = Only calculate one reconstruction using odd particles.
2 = Only calculate one reconstruction using even particles.
3 = Only calculate one reconstruction using all particles.
FDUMP T/F If set to T, dumps intermediate files from a 3D reconstruction
and then terminates run. This feature allows splitting up
3D reconstruction into several parts (using IFIRST,ILAST). The
dump files can be read in and merged for a full reconstruction
using the new program merge_3d (included with the distribution).
IMEM 0/1/2/3 Memory usage: 0 = least memory, 3 = most memory.
0 - no padding of reference during refinement, no multi-volume
parallelization during reconstruction (least memory usage)
1 - padding of reference during refinement, no multi-volume
parallelization during reconstruction
2 - no padding of reference during refinement, multi-volume
parallelization during reconstruction
3 - padding of reference during refinement, multi-volume
parallelization during reconstruction (most memory usage)
INTERP 0/1 Interpolation scheme used for 3D reconstruction:
0 = Nearest neighbor
1 = Trilinear (more time-consuming)
---------------------------------------
RO - Outer radius of reconstruction in Angstroms from centre of particle: 108.0
Enter the outer radius of the volume to be reconstructed. The program will
also apply a mask with a cosine edge to the particle image before
processing (done inside CTFAPPLY using HALFW=6 pixels for cosine bell).
RI - Inner radius of reconstruction in Angstroms from centre of particle: 0.0
Enter the inner radius of the volume to be reconstructed. This is useful
for reconstructions of viruses and other particles that might be hollow
or have a disordered core.
PSIZE - Required pixel size [Angstrom]: 3.00
Enter pixel size in Angstroms required for output map. The input
particle images are then reinterpolated by using the densitometer step
size and the magnification of the micrograph to calculate the relative
magnification. RELMAG, normally set to 1.0, can be used to make further
manual adjustments. A warning message is printed if the pixel size in
the header of the input 3D map used as the reference differs by more
than 1% from the required pixel size for the output map. The input 3D
map is not interpolated, but it is assumed the user knows what they are
doing. Similarly, a warning is printed if the input particle image
pixel size is less than 0.65 or more than 1.50 times PSIZE.
MW - Approximate molecular mass of the partcle, in kDa. This is used to
calculate the optimal filter for the 3D reconstruction (option FFILT).
WGH - % Amplitude contrast (-1...1): 0.07
Amplitude contrast for CTF correction. A negative value will invert the
sign of the CTF to flip image contrast. Can use WGH = -1.0 to set
CTF = 1 to fit images to a model without CTF correction.
XSTD - number of standard deviations above mean for masking of input low-pass
filtered 3D model - note this 3D masking does not use RI
- if positive, calculates mask with subroutine D3MASK, equiv to
solvent flattening with 5-pixel-cosine-bell smoothed mask
boundary. The mask is then used to multiply the input 3D map,
which is then used for all parameter refinement and subsequent
calculations.
- if negative, calculates mask with subroutine D2MASK resulting
in a sharp binary (0/1) mask boundary for which is used for
both parameter refinement and reconstruction, and to mask and
output the matching projections. Each matching particle image
is also always masked with a cosine bell edged function of
radius RI (see above RI).
PBC - Phase residual / pseudo-B-factor conversion Constant: 5.0
Automatic weighting is applied to each particle: a pseudo-temperature (B)
factor is applied to each particle according to its relative phase
residual against the reference. The weight is calculated as
W = exp (-DELTAP/PBC * R^2)
with DELTAP = relative phase residual (actual phase residual minus BOFF),
PBC = conversion constant (5.0 in the example),
and R^2 the squared resolution in Fourier units (R = 0.0 ... 0.5).
A large value for PBC (e.g. 100.0) gives equal weighting to each particle
BOFF - average phase residual: 60.0, approximate average phase residual of
all particles, used in calculating weights for contributions of different
particles to 3D map (see Grigorieff, 1998).
DANG - angular step size for the angular search used in modes IFLAG=3,4
There is a program default value calculated taking resolution into
account, but if this input value is non-zero, the program value is
overridden.
ITMAX - number of cycles of randomised search/refinement used in modes IFLAG=2,4
There is a program default value (10 cycles), but if this input value is
non-zero, the program value is overridden.
IPMAX - number of potential matches in a search that should be tested further in
a subsequent local refinement.
---------------------------------------
PMASK - 0/1 mask to exclude parameters from refinement. There are five numbers to
allow psi, theta, phi, deltaX, deltaY to be included or excluded from the
search or refinement. The use of 1 1 1 1 1 causes all 5 parameters to be
refined for each particle. The use of 0 0 0 1 1 would cause only the
position of each particle to be optimised.
DMASK - X,Y,Z and radius to discribe an area in the map that should be used for
classification ("focussed classification"). Not extensively tested - use
with caution.
---------------------------------------
IFIRST,ILAST - First and last particle to be included: 1,5000
Number of first and last particle to be included in the complete
reconstruction, using data from all datasets.
---------------------------------------
ASYM - symmetry required Cn,Dn,T,O,I,I1,I2 or N (can be zero)
n = rotational symmetry required in pointgroup Cn or Dn
N = number of symmetry matrices to read in.
T = tetrahedral pointgroup 23
O = octahedral pointgroup 432
I = icosahedral 532 symmetry in setting 1 (5fold is on X)
I1 = also in setting 1 (X) - as used by Imagic
I2 = in setting 2 (Y) - as used by Crowther et al
For any standard pointgroup symmetries and setting, the symmetry
requested is used to calculate the symmetry operators to be applied
to the reconstruction at the end of the calculation and to the search
angles in modes 3 and 4.
Alternatively the symmetry matrices can be read in explicitly to allow
for unusual symmetries or unusual orientations. If the number of
symmetry operators, N, is not 0 then the symmetry matrices must follow
immediately, e.g.
0.5 0.8660254037844 0.0
-0.8660254037844 0.5 0.0
0.0 0.0 1.0
(this matrix carries out a 60deg rotation about the z axis)
In version 2.01 and earlier, the possibility of applying Cn,Dn,T,O, or I
symmetry in standard settings is not available.
---------------------------------------
RELMAG,DSTEP,TARGET,THRESH - Relative magnification to apply to data set (0=END)
densitometer step size in microns,
target phase residual for parameter refinement,
and phase residual threshold: 1.00, 7.0, 15.0, 90.0
RELMAG: Relative magnification of data set. The magnification feature
allows for manual variations of magnification between data sets. The
micrograph magnification should be in the input parameter file or read
in on card 8b. Enter 0.0 or a negative value if no further data sets
are to be added. If RELMAG is a negative number (e.g. -100.0), there
is no map output, application of symmetry or statistics.
DSTEP: Densitometer step size. Each dataset defined by cards 5 to 10 must
have the same values for DSTEP as well as other parameters on this card.
TARGET: Target phase residual (for resolution between RMAX1 and RMAX2)
during parameter saerch and refinement, below which the search and/or
refinement is terminated. This is normally set low (e.g. THRESH=10.0)
This will give excellent determination of particle orientations.
THRESH: Phase residual cut-off. Any particles with a higher overall phase
residual will not be included in the reconstruction when IFLAG=0,1,2,3.
This variable is often used with IFLAG=0 in separate runs to calculate
maps using various values of THRESH to find the optimum value to
produce the best map as judged from the statistics.
CS,AKV - CS [mm], V [kV]: 2.60,200.0
Microscope parameters for this data set: Cs and voltage.
TX,TY - Beam tilt [mrad] in X, Y direction: 0.0, 0.0
---------------------------------------
RREC - Resol. of reconstruction in Angstroms, e.g. 10.0
Resolution to which the reconstruction is calculated.
If several datasets have different values, the data is individually
limited in the summation to the RREC resolution but symmetry is
applied, statistics output and the final map calculated to the
maximum resolution requested for any dataset.
RMAX1,RMAX2 - Resol. in refinement in Angstroms, low & high: 200.0,25.0
Resolution of the data included in the search/refinement. These
two parameters are very important. The successful alignment of
particles depends critically on the signal-to-noise ratio of the
cross-correlation or phase residual calculation, and exclusion of
weak data at high resolution or spurious, very strong artefacts at
low resolution can make a big difference. Success can be judged
by whether the X,Y coordinates of the particle centres are reasonable.
DFSIG - Defocus uncertainty in Angstroms, e.g. 200.0
This will restrain the change in defocus when refining defocus values
for individual particles.
RBFACT - B-factor to apply to particle image projections before orientation
determination or refinement. This allows inclusion of high resolution
data but with a reduced (or increased if negative) weight. WGH and
RBFAC can be manipulated in particle parameter refinement as if they
were low pass and high pass filters. WGH and the CTF are used to
correct the density in the final map, whereas RBFAC is not.
NOTE: This option should be used with great care as amplification of
noisy high-resolution terms can lead to over-fitting and artificially
high values in the FSC curve (se publication #2 above). FREALIGN uses an
automatic weighting scheme and RBFACT should normally be set to 0.0.
---------------------------------------
NIN - number of particle images for this film
ABSMAGPIN - real magnification for this film (e.g. 45,000 x)
IFILMIN - film number for this film
DFMID1IN,DFMID2IN,ANGASTIN - defocus and astigmatism for this film
MORE - use '1' for more cards describing more particle images in
this stack of images, '0' to terminate.
********************************************************************************
File formats for input/output files
==================================
All output files containing 3D maps have spacegroup 1 and 80 bytes of symmetry
information in the header (containing " X,Y,Z "), whereas the matching image
stack, has spacegroup 0 with no symmetry bytes. Note that the MRC format
also defines a mapmode depending on how many bytes of real or complex data is
written out per pixel.
Parameter FILE for input or output: Twelve entries per particle:
## PSI THETA PHI SHX SHY MAG FILM DF1 DF2 ANGAST PRESA
Particle number
Phi - first Eulerian angle, rotates model clockwise about Z (alpha)
Theta - rotates about new Y after rotation about Z (beta), to produce a
characteristic view of the particle
Psi - third rotation about new Z (gamma), rotates particle without
changing the viewpoint
this ZYZ rotation is standard Eulerian [a.k.a. alpha, beta, gamma]
Shift X, Shift Y (in Angstrom)
Magnification - this is real magnification (e.g. 45,000 x). If
magnification has been refined, then the output parameter file will
contain the appropriate magnification so that the new value of
RELMAG = 1.0
Film number - identifies which particles are from same film
defocus1 in Angstrom, defocus2 in Angstrom, angle for astigmatism
The two defocus values (positive for underfocus) give the defocus in two
perpendicular directions. They are the same if there is no astigmatism.
The angle is measured between the X-axis and the direction of the first
defocus measurement (a positive angle means a rotation of the defocus
direction to the left).
Phase residual: This is actually the inverse cosine of a weighted correlation
coefficient calculated during refinement (see publication #2 above).
Statistical table 1: Thirteen entries per radius bin in table comparing two halves of data
C NO. RESOL RING RAD FSPR FSC QFACT QRAN+SIG SSNR RFACT PFREE PNOISE NONZERO TOTVOX
C 2 398.1 0.0078 0.00 0.997 0.971 0.030 47.51 0.073 26.22 108.44 17 17
C 3 199.0 0.0156 0.03 0.997 0.559 0.040 19.98 0.080 80.41 75.08 41 41
C 4 132.7 0.0234 0.04 0.998 0.538 0.045 17.86 0.061 77.51 65.14 89 89
C 5 99.5 0.0313 0.66 0.982 0.398 0.057 13.37 0.184 57.01 109.59 129 129
C 6 79.6 0.0391 0.30 0.965 0.426 0.063 13.67 0.290 55.10 93.69 225 225
C 7 66.3 0.0469 3.07 0.908 0.166 0.065 4.93 0.385 104.20 82.42 253 253
C 8 56.9 0.0547 11.57 0.882 0.209 0.069 4.70 0.337 86.66 103.94 393 393
C------------------------------------------------------------------------------------------------
C 17 24.9 0.1250 54.79 0.328 0.122 0.107 1.56 0.671 85.50 80.16 1701 1701
C 18 23.4 0.1328 54.18 0.345 0.100 0.109 1.02 0.687 87.59 87.67 2021 2021
C 19 22.1 0.1406 73.02 0.169 0.083 0.110 0.73 0.649 77.97 87.75 2181 2181
C 20 21.0 0.1484 76.84 0.136 0.088 0.114 0.69 0.661 93.42 83.76 2473 2473
C 21 19.9 0.1563 56.51 0.311 0.132 0.117 1.44 0.712 85.63 86.15 2697 2697
+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
Statistical table 2: Eight entries comparing merged data with reference data
C NO. RESOL RING RAD FSPR FSC RFACT NONZERO TOTVOX
C 2 398.1 0.0078 0.14 0.998 0.094 17 17
C 3 199.0 0.0156 1.30 0.969 0.705 41 41
C 4 132.7 0.0234 4.55 0.854 0.575 89 89
C 5 99.5 0.0313 0.68 0.958 0.953 129 129
C 6 79.6 0.0391 1.60 0.937 0.667 225 225
C 7 66.3 0.0469 41.69 0.452 1.213 253 253
C 8 56.9 0.0547 5.74 0.793 1.103 393 393
C----------------------------------------------------------
C 17 24.9 0.1250 38.32 0.520 0.650 1701 1701
C 18 23.4 0.1328 50.69 0.391 0.649 2021 2021
C 19 22.1 0.1406 65.20 0.228 1.052 2181 2181
C 20 21.0 0.1484 58.91 0.306 0.954 2473 2473
C 21 19.9 0.1563 43.31 0.482 0.613 2697 2697
+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
Abbreviations:
FSPR: Fourier shell phase residual
FSC: Fourier shell correlation
APPR: Average particle phase residual (against reference)
Av. Q: Average Q-factor
Critical Q: Q-factor for pure noise (as defined in the JMB reference)
SOME IMPORTANT NOTES:
--------------------
1. Shifts are measured in Angstrom to make it easier to work with binned
data. This is an important change from earlier versions (<9.00) that
measured shifts in pixels. Older parameter files should be compatible
with later versions as Frealign attempts to recognize older parameter
files and converts pixel shifts into Angstroms.
2. In earlier (<5.00) versions of the program, the application of symmetry
in subroutine APPLYSYMC was carried out using an extra phase shift (PSHFT)
which moved the centre of the symmetry axes of the particles by half a
pixel to be at (NSAM/2 - 0.5) along all three axes (e.g. [63.5,63.5,63.5]
in a 128x128x128 map). This meant that the output map had the exact centre
at the midpoint of 8 pixels. While this had no effect on the internal
consistency of the program itself since each successive round of
refinement automatically located the (X,Y) coordinates of each particle
image correctly relative to the 3D model, it meant that it was hard to
interface the program to others. Accordingly, the centre for application
of the symmetry was moved to be exactly centred on the pixel at NSAM/2+1
from version 5.00 onwards. The program changes were : to set HALFP to
be zero in subroutine APPLYSYMC and to move the centre appropriately
inside subroutine BEAUTIFY.
3. The recommended initial parameters for more accurate orientation parameter
determination are:-
a high resolution (e.g. 10-15 Angstroms)
a reasonably but not excessively high value for ITMAX (e.g. 200)
a reasonably but not excessively high value for IPMAX (e.g. 10)
a high rather than a small value for DANG (i.e. 200.0 instead of 2.0)
RBFACT should be set to 0 but different values can be tried to
decrease the weight of the noisier data at higher resolution
(e.g. use 2000 or 3000) but to keep them in.
A range of values should always be tried since the optimal values may
depend on the structure and the quality of the pictures.