A number of people had valuable suggestions but it appears that, as
is
often the case, the simplest solution was the most relevant.
Based upon scaleit.doc, I had not been using the REFINE keyword in
scaleit. "Default for program is REFINE ANISOTROPIC"
Therefore, I had not included REFINE ANISOTROPIC in my script.
It turns out that refinement is _NOT_ the default.
Adding REFINE ANISOTROPIC to my script took care of my ill-centered
distribution of isomorphous differences and the FOMs from mlphare are
much more reasonable for the centric reflections.
Even adding REFINE ISOTROPIC takes care of most of the problem.
For debugging purposes, here's my script:
The only thing I've added is the REFINE ANISOTROPIC line.
scaleit hklin mtz/${all}.cad \
hklout mtz/${all}_${base}.scit
\
<< EOF >log/${all}_${base}.scit
title ${in1}, ${in2}, ${in3} scaled against ${base}
LABIN FP=FP_${base} SIGFP=SIGFP_${base} -
FPH1=FP_${in1} SIGFPH1=SIGFP_${in1} DPH1=DANO_${in1} SIGDPH1=SIGDANO_${in1}
-
FPH2=FP_${in2} SIGFPH2=SIGFP_${in2} DPH2=DANO_${in2} SIGDPH2=SIGDANO_${in2}
-
FPH3=FP_${in3} SIGFPH3=SIGFP_${in3} DPH3=DANO_${in3} SIGDPH3=SIGDANO_${in3}
resolution 10 3.0
REFINE ANISOTROPIC
end
EOF
Comparing log files with and without explicitly stating REFINE (AN)ISOTROPIC
clearly shows that it is not the default. The section showing
"SCALES (for
F)", the section listing anisotropic temperature factors, and the table
of isomophous differences all show major differences.
-- Dr. Linda Hannick, Structural Biology National Institute of Allergy and Infectious Diseases National Institutes of Health 12441 Parklawn Drive 301-496-3792 Voice Rockville, MD 20852 301-402-0284 FAX