[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
[ccp4bb]: SCALA probs under Linux
*** For details on how to be removed from this list visit the ***
*** CCP4 home page http://www.ccp4.ac.uk ***
Dear all,
We are having a bit of a problem with SCALA on an AMD Athlon box running
Mandrake 8.1. A data set processed with MOSFLM produces strange output
in SCALA. The "%poss" and "Cm%poss" columns in the statistics are
screwed up. They are either 0.0 or overflow the output field (stars).
All the other statistics are OK. This data is then impossible to
continue with in TRUNCATE. The Wilson plot contains no data.
MTZDUMP on the data after SCALA shows overflow in the most negative
values of the IMEAN and I+ columns:
1 ASC 0 33 0 100.00 12.5 12.5 29.21 2.30 H H
2 NONE 0 37 0 100.00 14.1 14.1 29.21 2.30 H K
3 NONE 0 43 0 100.00 16.4 16.4 29.21 2.30 H L
4 NONE*******670752.4 0 100.00 21930.99 21991.65 29.21 2.30 J
IMEAN
5 NONE 2.0 47655.4 0 100.00 1526.74 1526.74 29.21 2.30 Q
SIGIMEAN
6 NONE*******804244.9 0 100.00 21585.43 21658.87 29.21 2.30 K
I(+)
7 NONE 0.0 64406.7 0 100.00 2091.73 2091.73 29.21 2.30 M
SIGI(+)
8 NONE-8288.7569135.2 0 100.00 21844.94 21924.49 29.21 2.30 K
I(-)
9 NONE 0.0 87471.5 0 100.00 2178.36 2178.36 29.21 2.30 M
SIGI(-)
10 NONE 0.0260102.1 0 100.00 2410.57 2410.57 29.21 2.30 Q
SCATTER
11 NONE 0.0183920.0 0 100.00 1113.97 1113.97 29.21 2.30 Q
SCATIMEAN
Running the same script on an Octane with the same CCP4 version gives no
problems, i.e. normal SCALA and TRUNCATE output. However the output of
MTZDUMP is essentially identical. Clearly we have some large negative
values, larger than MTZDUMP can display in f7.1, but I don't understand
how this affects the completeness statistics rather than the R-factors...
We tried to recompile CCP4 with gcc-3 using Kevin's instructions, but
this didn't help (the Mandrake 8.1 compiler was gcc-2.96-0.62) Has
anyone else experienced similar problems? I have a feeling it's
something to do with MNFs but I'm not sure...
Derek
--
Derek Logan tel: +46 46 222 1443
Molecular Biophysics fax: +46 46 222 4692
Lund University
Box 124, Lund, Sweden