SRS5.0.3 - GCG format problems

Martin Hilbers mph at dl.ac.uk
Mon Mar 17 06:09:31 EST 1997


Hi folks,

I had a few problems when indexing GCG formatted embl and
genbank databases with srs5.03. This all happened without
modifications of the software and embl/genbank icarus files.
(Well, for genbank I had to change the file definitions to
GCG - bot otherwise it was all "out of the box")
   

srsbuild crashed on em_fun:scchrix - yeast chromosome IX, a
450 Kb sequence. This happened with DEC/OSF1 3.2 and SGI/Irix 5.3
There was no problem with using the same sequence in the
original EMBL file, nor with the same sequence in a GCG formatted
GenBank database.


With the whole of EMBL indexing crashed after about 860000 entries 
with the message:

SRSDB:embl.is:42: error: insufficient memory - error during malloc,
could not allocate "sequence"

It could hardly be "insufficient memory" - I monitored memory
usage by srsbuild, and it was using about 115 Mb when it crashed,
and the  system I used (DEC Alpha/ OSF1 3.2) has 190 Mb + 660 Mb
swapspace. It occurred during the indexing of em_fun, but well
before it got to the scchrix entry.

In another try, without the section in which the above problem 
occurred (em_fun) it ended after about 890000 entries with:
 
31419 Memory fault - core dumped

So there seems to be some kind of memory problem with gcg formatted
embl.


With Genbank (GCG formatted) I got a bit further, but after
about 1048800 entries, it suddenly created the section "11"
subindices (I was using chunks of 100000), and proceeded 
creating sections 12,13,14 etc, without reading any data.
I managed to stop it when it was at section 35 !


Any suggestions ?


Cheers,

Martin
-- 
-----------------------------------------------------------------
| Martin Hilbers  http://www.dci.clrc.ac.uk/People/m.p.hilbers/ |
| SEQNET                |     E-mail: m.p.hilbers at dl.ac.uk      |
| Daresbury Laboratory  |     Tel:    +44-1925-603492           |
| Daresbury, Warrington |     Fax:    +44-1925-603100           |
| Cheshire WA4 4AD      | SEQNET is the UK national EMBNet node |
| United Kingdom        | http://www.dl.ac.uk/SEQNET/home.html  |
-----------------------------------------------------------------




More information about the Bio-soft mailing list