Tracker: Bugs

5 GBrowse on BioSQL and sequence versions - ID: 2168597
Last Update: Settings changed ( scottcain )

This was reported on the mailing list, and Lincoln suggested filing a bug.

Tested on Linux and MacOS, using GBrowse 1.69 and the then current SVN
stable branch.

Using BioSQL with GenBank files loaded using either BioPerl or Biopython.

GenBank files with revision/version equal to one work fine, e.g.

ftp://ftp.ncbi.nih.gov/genomes/Bacteria/Nanoarchaeum_equitans/NC_005213.gbk

ftp://ftp.ncbi.nih.gov/genomes/Bacteria/Pirellula_sp/NC_005027.gbk
ftp://ftp.ncbi.nih.gov/genomes/Bacteria/Pasteurella_multocida/NC_002663.gbk

ftp://ftp.ncbi.nih.gov/genomes/Bacteria/Rhodococcus_RHA1/NC_008268.gbk
ftp://ftp.ncbi.nih.gov/genomes/Bacteria/Rhodococcus_RHA1/NC_008269.gbk
ftp://ftp.ncbi.nih.gov/genomes/Bacteria/Rhodococcus_RHA1/NC_008270.gbk
ftp://ftp.ncbi.nih.gov/genomes/Bacteria/Rhodococcus_RHA1/NC_008271.gbk
ftp://ftp.ncbi.nih.gov/genomes/Bacteria/Escherichia_coli_K_12_substr__DH10B
/NC_010473.gbk

GenBank files with revision/version not equal to one work fail for the
details page, e.g.

ftp://ftp.ncbi.nih.gov/genomes/Bacteria/Escherichia_coli_K12_substr__MG1655
/NC_000913.gbk
ftp://ftp.ncbi.nih.gov/genomes/Bacteria/Bacillus_subtilis/NC_000964.gbk

The main page works, but the details browser fails. Here the URL appears
fine - containing the feature ID and other redundant information (like the
parent sequence and feature type). In BioSQL, given the feature ID is a
database record key, this is all that need be required to determine the
feature AND its parent sequence.

Also, I have not tested GBrowse with more than one revision of the same
accession in a BioSQL database (e.g. NC_000913.1 and NC_000913.2).


Peter ( maubp ) - 2008-10-15 14:15

5

Open

None

Lincoln Stein

Gbrowse

v1.0

Public


Comments ( 2 )




Date: 2008-12-19 11:14
Sender: maubp

Due to different time zones, I'm not sure Lincoln's comment was before or
after our exchange on the mailing list. It also seemed like a good idea to
put these notes on the bug tracker.

Today (19 Dec 2008) I updated a working Linux installation of GBrowse
1.69+ under apache to the latest stable code using:
$ sudo perl gbrowse_netinstall.pl --gbrowse_dev
--build_param_str="--SELINUX=1"

For sequences where the version number is one, everything seems to work
fine as before. However, as described in my email (17 Dec 2008) the
details page problem remains for sequences with a non-one version number in
the bioentry table.

See also:
https://sourceforge.net/mailarchive/message.php?msg_name=320fb6e00812170600ie1715d8la9ec99678c669816%40mail.gmail.com

e.g. viewing some of E Coli,
http://ppsrv/cgi-bin/gbrowse/biosql/?name=NC_000913:5528..8959
Clicking on a feature takes me to:
http://ppsrv/cgi-bin/gbrowse_details/biosql?feature_id=1213256;ref=NC_000913;start=6529;end=7959
This shows an error 500 message, "Internal Server Error".

$ sudo tail /etc/httpd/logs/error_log
[Fri Dec 19 11:00:46 2008] [error] [client 143.234.97.120] Can't call
method "seq" on an undefined value at /var/www/cgi-bin/gbrowse_details line
72., referer: http://XXX/cgi-bin/gbrowse/biosql/?name=NC_000913:5528..8959
[Fri Dec 19 11:00:46 2008] [error] [client 143.234.97.120] Premature end
of script headers: gbrowse_details, referer:
http://XXX/cgi-bin/gbrowse/biosql/?name=NC_000913:5528..8959

The relevant bit of code (in case my copy doesn't match yours) is:

my @features;
if ($f_id) {
@features =
$CONFIG->_feature_get($db,$name,$class,$start,$end,1,0,$f_id);

# HACK! This is for the BioSQL adaptor, whose sequences aren't
automatically
# attached to feature objects when they are retrieved from the database
if ($db->isa('Bio::DB::Das::BioSQL')) {
for my $f (@features) {
my $ref = $f->seq_id || param('ref');
# Need to fetch out the ENTIRE bioentry (chromosome) in order to attach
it to feature.
# This is horribly inefficient for performance and storage.
my @args = (-name=>$ref);
my ($segment) = $db->segment(@args);
$f->attach_seq($segment->seq); #Line 72, error here
}
}

}


Date: 2008-12-18 21:17
Sender: lsteinProject Admin

I have reworked how the details page works. Could you try the current
stable branch (or, if you are brave, the HEAD branch) and let me know if
the problem is still there.


Log in to comment.

Attached File

No Files Currently Attached

Changes ( 5 )

Field Old Value Date By
assigned_to scottcain 2009-01-08 13:51 scottcain
status_id Pending 2008-12-19 11:14 maubp
close_date 2008-12-18 21:17 2008-12-19 11:14 maubp
status_id Open 2008-12-18 21:17 lstein
close_date - 2008-12-18 21:17 lstein