Not exactly the same setup, but maybe close enough. I used Intel's 11.1
compilers on RHEL4.8 to build refmac (5.5.0102 and 5.5.0105) and it will
not run the refmac_tls script in the CCP4 example area.
It turns out that the LAPACK libraries (3.0) that comes with RHEL4
and the Intel (11.1) compiler does not play nice (10.1 worked). Building
a newer version of netlib LAPACK or using Intel's mkl_lapack
makes refmac happy.
This may not have anything to do with the problems you see but
nevertheless, it's worth considering...
Thomas
On 10/23/2009 06:04 AM, Ian Tickle wrote:
All – we’re having a problem with Refmac (version 5.5.0102) in CCP4
6.1.2 that I compiled from source using ifort v11.0 on Centos 4.6. When
I refine a structure with a HIS in alternate conformations (all atoms
except N, C & O doubled up) it completely destroys the sidechains of
both copies. Same thing happens with any other residue type (e.g. SER).
All single-conformer residues, including some ligands are fine.
Has anyone else noticed this? Is this a known problem with this
version/compiler/OS?
Cheers
-- Ian
--
Thomas Eriksson | thomas.eriks...@slac.stanford.edu
SSRL/Structural Molecular Biology | Tel (650)926-4530
2575 Sand Hill Rd. MS99 | Cell (650)714-6004
Menlo Park, CA 94025 | Fax (650)926-3292
