[EMBOSS] sort from dbiflat - problem solved.

Oliver S. Smart osmart at pugh.bip.bham.ac.uk
Tue Sep 9 12:24:24 UTC 2003


My colleague Tony Pemberton reported a problem with dbiflat to the list on
6 August. We thought it would be worth posting the solution. The exact
problem was produced when indexing EMBL release 75.0 using the dbiflat
command on a dual athlon running redhat 9.0. The sort command launched:

$ sort -o embl.acnum_sort2 -T . -k 1,1 embl.acnum_id2

never finished (>3 weeks) and produced no output. Interestingly the
command works fine under redhat 8.0, suse7.2 and 8.0 and irix. Also
indexing smaller databases works fine as the same sort command will work
on smaller files.  The solution is to set the LC_ALL environment variable
(mentioned in the sort man page) by:

# for tcsh, csh,...
$ setenv LC_ALL C
# OR for bash, ksh,...
$ export LC_ALL=C

The command then runs fine in around 80 seconds.  Consequently dbiflat
also runs without a problem. (Note that setting LC_COLLATE does not work),

Hope this may save someone else a big headache!

Oliver Smart (o.s.smart at bham.ac.uk) &
Tony Pemberton (A.J.Pemberton at bham.ac.uk)
The University of Birmingham



More information about the EMBOSS mailing list