Some time ago, our GCG program profilesearch started crashing middle of
searches with an error message:
Sequence information overflows string buffer!
After some head scratching and a mail message to gcg.com, things were
The problem was there only when the whole SWISS-PROT database was
searched. Profilesearch does something very stupid. It buffers all of the
descriptions as it searches the database and it happens that descriptions
in SWISS-PROT are longer than in PIR.
To get going again, one has to modify the critical parameter and
!* The names and description of each sequence is saved in an
!* array called bigstring. This is the major item using memory.
!* If your memory is limited change this so that only the sequence
!* entry name is saved, and decrease the size of MaxStrBuff.
Parameter (MAXSTRBUFF=4 000 000)
But to what value one should give to MAXSTRBUFF? The default value
crashed the program at 36 k sequences. I have it now at its maximmum 7
000 000 which should be enough until SWISS-PROT reaches 60 k entries. (It
is currently 43470+483). Luckily, a note from GCG said that they "will
change the way the program handles the documentation for individual
sequences and try to remove the limit of 60,000".
______ _/ _/____________________________________________________
_/ _/ _/ Heikki Lehvaslaiho <Heikki.Lehvaslaiho at CSC.FI>
_/_/_/_/_/ X.400: /G=Heikki/S=Lehvaslaiho/O=CSC/A=fumail/C=fi/
_/ _/ _/ Center for Scientific Computing
_/ _/ _/ Tietotie 6, P.O. Box 405, FIN-02101 Espoo FINLAND
_/ _/ _/ Phone: +358 0 457 2076 FAX: +358 0 457 2302
_/ URL: http://www.csc.fi/molbio/