This program can be used to scale together an Fobs (or Iobs) with a number of partial Fc's, vectorially combined. The partial structure factors might be e.g. from atoms and from back-transforming a solvent mask, to do a bulk solvent correction, or for scaling together contributions from a mixed structure.
The function minimised is
Sum{ w * [ Aobs - Acalc ]**2 } where Aobs = k0*Fobs or k0*Fobs**2 Acalc = scaled Fcalc(total) or scaled Fcalc(total)**2 complex Fcalc = k1 * exp (-B1 s**2) * complex FC1 + k1 * exp (-B2 s**2) * complex FC2 + . . . . . for up to 4 partial Fc's (defined by the number assigned by LABIN) w = weight = 1.0/(C1 * sigma(Fobs) + C2) (see WEIGHT command)
The output file contains Fobs and the total Fcalc, PhiCalc,
scaled either to Fobs or to one of the Fc's (see FIX command). Thus if
FC1 is the contribution from atoms, scaling to FC1 will put the output
on an absolute scale (FIX FC 1, this is the default)
KEYWORDED INPUT
The only compulsory command is LABIN. Keywords:
ANALYSIS, CYCLE, DEBUG, END, FILTER, FIX, INTENSITIES, LABIN, LABOUT, OUTPUT, PRINT, REFINE, RESOLUTION, RUN, SCALE, SFALL, TITLE, WEIGHT
FP, SIGFP native F, sigma(F) FC1, PHIC1 1st Fcalc, PhiCalc FC2, PHIC2 2nd Fcalc, PhiCalc FC3, PHIC3 3rd Fcalc, PhiCalc FC4, PHIC4 4th Fcalc, PhiCalc
The number of Fc's used for scaling is defined by which ones are assigned here. ALL required columns MUST be assigned, INCLUDING FP and SIGFP.
FP, SIGFP native F, sigma(F) FC, PHIC combined Fcalc, PhiCalc [W] weight (only if OUTPUT W or OUTPUT WDEL) [WDEL] w*(Fobs-Fcalc) (only if OUTPUT WDEL) [FCn,PHICn] partial Fcalcs, Phicalcs with n=1 to 4, from the input file (only if OUTPUT FPART)If LABOUT is omitted then the labels from LABIN are used (with W and WDEL being the defaults for the weights). See also LABIN and OUTPUT.
- NOHKL
- no output file
- NONE
- no o/p file and no final statistics
- FOFC
- output contains FP, SIGFP, FC, PHIC (default)
- WEIGHT
- output also contains WEIGHT
- WDEL
- output also contains WEIGHT WDEL (= w(Fo-Fc))
- FPART
- output also contains scaled component FCn, PHIn
The program can optionally output a file containing the refined scales and B-factors, to logical name RSTATSBKR: this is controlled by the keyword BKR:
- BKR
- write output file RSTATSBKR (NORST=0) [default no output]
It is a good idea to start the parameters at
approximately the right values. B for FP (=Fobs) will ignored
[Default scales = 1.0, B's = 0.0]
e.g. SCALE FP 0.8 SCALE FC2 1.2 80
- FOBS
- keep FOBS constant, apply scale to all Fc's (KEEP=0).
- FC <n>
- keep <n>-th Fcalc constant (KEEP = n). This is the default with <n> = 1.
With the default value (= FIX FC 1) the output FP and FC are on the scale of FC1.
REFINE FIX FC 1 or REFINE FIX BFC 1 FIX BFC 2 or REFINE TIE FC 1 2 FIX FC 3are all valid. Multiple <refine_subkey>s are allowed as long as they follow the correct syntax; alternatively, REFINE may be used multiple times (so not all the parameters need to be specified on a single line).
Set resolution limits. By default, all data are used and the limits are taken from the HKLIN file
This command can be used to allow refinement of some parameters, then a return to reading new commands to change which parameters are refined, e.g.
REFINE FIX FC 2 RUN REFINE FREE FC 2 OUTPUT FOFC END
- IFILT
- filter level = 10**(-IFILT) [default = 8]
- NEIGMX
- maximum number of eigenvalues to use [default = 0 = all]
icoefl hklin u1afofcfslv hklout u1afofcfslv_sc <<eof-icoefl title Scale Fc(atom) and Fc(solvent) to Fobs labin FP=FP SIGFP=SIGFP FC1=FC PHIC1=PHIC FC2=FCSOLV PHIC2=PHICSOLV labout FC=FCALL PHIC=PHICALL scale FP 1.4 scale FC1 1.0 15 scale FC2 0.3 80 cycles 20 eof-icoefl
MTZ version: Phil Evans, MRC LMB, Cambridge