The resolution in the reconstruction is determined by the number of
annulae used in the Fourier Bessel synthesis. The data is sampled
as a set of rings. The spacing of the rings is determined by the particle
size which is set here and the sampling (NSAMP) used. Typically NSAMP is
2. Specifically
for a particle of 122 pixels diameter
with nsamp =2
means that each ring will be spaced at 1/244 pixels
if each pixel is 5A in size this will be 1/1220 A
so for a reconstruction to 50 A resolution one needs to use
1220A/50A = 24 annulae
for 20 A need 61annulae
The limit which can be reached depends on the amount and distribution of data available and can be seen from the Inputs
- Inputs are indicated with >'s
- The input of this program is dependent upon whether one is
beginning a new set of particles (new ENMAT file) or
is adding particles to a pre-existing set.
- > Title - will be a header for the set (80a)
- > Open new ENMAT file? (I)
- 1 yes
- 0 or
no - there must be a preexisting EMICOMAT.DAT file.
- ---- if this run will input the first particle of a set
(ie if the previous input was =1) then next input is
- >DIAM (F)
The diameter of the particle in pixels. Default=64 pixels
- >DIAM (F)
- >LMAX, NSAMP, ICHECK (3I) LMAX
- LMAX - number of annulae sampled from the data NSAMP - the data are sampled at either 1/NSAMP*DIAM intervals. If NSAMP .gt.2, it will be set to 1
- DIAM is the diameter of the particle in pixels and is provided below.
- ICHECK - an output control parameter. If ICHECK >0, detailed and volumous output concerning the run is provided.
- > number of particles input in this run (i)
- --------- then for each particle --
- > fft file name for (A)
an emicomat output file is generated with the same name as the particle input fft file but the extension .mat this new file contains the values of the coefficients of the normal equations
- > is the default phase origin OK? (i)
- 0 or
yes - 1 no
- > new phase origin ORIGX,ORIGY
- 1 no
- 0 or
Orientation of the particle. There are two possible orientations of the particle which correspond to the flipped and normal orientations in SIMPLEX. The better of the two can be chosen from the error matrix. The difference between the flipped and unflipped orientations is acheived by adding 180 degrees. The values of THETA, PHI and OMEGA will be returned to the icosahedral asymmetric unit if they are outside it a message is written to inform the user. The default value for ANG_PER_PIXEL is taken from that for previous images in the set. If this is the first image the default value is 8.036 A/pix.
-
or 0 - no scaling, the raw data is used - 1 - scaling with a scaling function, a legendre polynominal whose coefficients must be provided.
- > min and max radius for scaling (2i)
- only a portion of the complete transform may be used for the scaling although the scaling function will be applied to the entire set of data sampled from the transform. Usually the values close to the origin and the weak values at the edge are excluded from the calculation of the scaling function.
- > file of coefficients for the fit (a)
- file which contains the parameters of the envelope
to which the transform will be matched. This is a
legendre polynomial with NTERMS.
The format is:
- title (a)
- NTERMS (i)
- INDEX, POWER,COEFF (2i,f) NTERMS lines which provide the list the list of coefficients for the envelope function. POWER>0 indicates a power of the resolution while POWER<0 indicates a power of cos(resolution).
- FMIN (f) cutoff for amplitudes
- REVRAD (f) this provides a way of correcting for phase reversal. For resolution greater than REVRAD the sign of the tranform value is reversed.an example of such a file is provided in SINDBIS_SMOOTH.DAT subroutines CORRF, FFITC, and SET_CORRF show the way the coefficients are derived.
- 2 - scaling of the average amplitudes to a provided profile
FAV will be set to the values in the provided profile this is not a smoothed function as above but simply a look up table of multipliers as a function of resolution
- > min and max radius for scaling (2i)
- only a portion of the complete transform may be used for the scaling although the scaling function will be applied to the entire set of data sampled from the transform. Usually the values close to the origin and the weak values at the edge are excluded from the scaling function
- > file name for profile (a)
the format for this file is:
- title (a)
- NVALUES (I) the number of sample points in the profile
- spacing of values in file (f) in 1/A
- exponential factor to be applied to profile in A profile will be profile*exp(FACT(A)*RESLN(1/A)) (f)
- amplitudes (f) a list of NVALUES linesan example of such a file is HARRISON_FIT.DAT
- 3 - scaling of the average amplitudes to the transform of a sum of spheres
- > NSPHERES, NSPHERE(DEN_SPHERE,R_SPHERE)
i,(f,f)
- - the number of spheres in the fit
- - the density of each sphere and
- - the radius of each sphere
so a fit to a transform of a 700A sphere of density 1.0 with a 300 sphere of density 3.0 at its center would be expressed 2,1.0,700.0,2.0,300.0
- > Exponential factor
- exponential factor to be applied to profile in A profile will be profile*exp(FACT(A)*RESLN(1/A)) (f)
- -4 The transform or the reconstruction will be modified by a Wiener filter
- > min, max radius for scaling with ctf
- > filter the transform(1) or the reconstruction (0)
- > df,Cs,KV,framp,decay,N_to_s (6f)
- df - defocus value in micron
- Cs - in mm
- KV - accel voltage
- ( which is used to calculate lambda - electron wavelength in nm)
- framp - fraction amplitude contrast (ca 0.1)
- decay - decay constant factor (A^2)
- N_to_s - Noise/Signal ratio. 0 = real CTF corr
- 1 - scaling with a scaling function, a legendre polynominal whose coefficients must be provided.
Sample runs
com file for complete reconstruction
corresponding log file
annotated com file
Update History
- Current version - 26jul95
- -sdf 26jul95
- version which includes the ctf correction made the current version
Author: Stephen Fuller