Re: [EMBOSS] seqret segfault (refseq protein sequence, indexed with dbxflat)
Dear Peter, this fixes the problem on my laptop so I'll roll this out on the Ubuntu box as well. Thank you for your quick reply, and apologies for forgetting the attachment. (I've been comparing the two records but I would not have guessed that the difference in length of the coded_by value is the problem -- not without using the source anyway.) Best regards, Jan On Fri, Mar 01, 2013 at 07:03:44PM +, Peter Rice wrote: > Dear Jan, > ? > > I've run into a weird problem with seqret after downloading the complete > > protein refseq database and indexing that with dbxflat. The problem > > seems to be triggered by a rare condition, so far I've only encountered > > it with accession ZP_10312765: > > The problem is in ajax/core/ajfeatwrite.c in function featTagSwissWrapC (so > it will only affect Swissprot format feature output). > ? > Variable j is declared ajulong but should be ajlong (signed, so it can have a > value of -1 in this case). > ? > If you change the line to > ? > ??? ajlong j; > ? > and rebuild, that should fix it. > ? > I will check the other wrapping functions in case they have a similar problem > with other inputs. > ? > Again, many thanks for finding it. > ? > Peter Rice > EMBOSS Team -- +- Jan T. Kim ---+ | email: jtt...@gmail.com| | WWW: http://www.jtkim.dreamhosters.com/ | *-=< hierarchical systems are for files, not for humans >=-* ___ EMBOSS mailing list EMBOSS@lists.open-bio.org http://lists.open-bio.org/mailman/listinfo/emboss
Re: [EMBOSS] seqret segfault (refseq protein sequence, indexed with dbxflat)
Dear Jan, > I've run into a weird problem with seqret after downloading the complete > protein refseq database and indexing that with dbxflat. The problem > seems to be triggered by a rare condition, so far I've only encountered > it with accession ZP_10312765: The problem is in ajax/core/ajfeatwrite.c in function featTagSwissWrapC (so it will only affect Swissprot format feature output). Variable j is declared ajulong but should be ajlong (signed, so it can have a value of -1 in this case). If you change the line to ajlong j; and rebuild, that should fix it. I will check the other wrapping functions in case they have a similar problem with other inputs. Again, many thanks for finding it. Peter Rice EMBOSS Team ___ EMBOSS mailing list EMBOSS@lists.open-bio.org http://lists.open-bio.org/mailman/listinfo/emboss
Re: [EMBOSS] seqret segfault (refseq protein sequence, indexed with dbxflat)
Dear Jan, > I've run into a weird problem with seqret after downloading the complete > protein refseq database and indexing that with dbxflat. The problem > seems to be triggered by a rare condition, so far I've only encountered > it with accession ZP_10312765: > > % seqret -feature -outseq=stdout -osformat=swiss ptest:ZP_10312765 > > Monitoring the seqret process using top, I noticed that the process > grows to a size of 2g before segfaulting. > > Trying the same with ZP_10312766, the next record in the file, causes > no problem. Also, -osformat=fasta and -osformat=genbank work with > ZP_10312765, so the problem seems to be with outputting the swiss format. I can reproduce the problem with the latest EMBOSS. The problem seems to be trying to wrap a feature line to 41 bytes per record when the coded_by location is longer than the available width. I will work on a patch for the latest release and also for the 6.4 release you were using. Many thanks for finding this one. Peter Rice EMBOSS Team ___ EMBOSS mailing list EMBOSS@lists.open-bio.org http://lists.open-bio.org/mailman/listinfo/emboss