Learn how easy it is to sync an existing GitHub or Google Code repo to a SourceForge project! See Demo

Close

#119 Problems with DNA/GC track display

closed
Lincoln Stein
Gbrowse (196)
7
2005-08-22
2005-08-10
Chris Fields
No

The DNA/GC track doesn't display the correct sequence
when viewed at 100 bp or less using the volvox tutorial
database. The sequence seems to have an offset
problem; when starting from 1-100, the sequence is
correct, but at the position above it is off by ~12
bases. This problem seems to be isolated to the DNA/GC
track; the sequence displayed in the EST track is
correctly plotted.

Attached is the file ctgA.png, which shows the problem.
The GBrowse version is 1.63, using Bioperl 1.51 (GMOD
version; problem was also seen under using Bioperl 1.5
from bioperl.org), ActiveState Perl 5.8.7. OS is WinXP
Pro, run on P4 processor (1.7 GHz), 256 MB. Unknown at
this point if it is a Gbrowse issue, Perl issue, or OS
issue.

Discussion

  • Chris Fields
    Chris Fields
    2005-08-10

    shows error on DNA/GC track display.

     
    Attachments
  • Chris Fields
    Chris Fields
    2005-08-11

    • priority: 5 --> 7
     
  • Chris Fields
    Chris Fields
    2005-08-11

    Logged In: YES
    user_id=1327166

    The bug persists in the included yeast chromosome I
    sequence. Attached is the PNG file from Gbrowse. The first
    10 bases displayed as 10000-10009 (GACAGTATCT) are bases
    9840-9849 in the real sequence. I notched up the priority
    on this one a bit; I'll check the GFF files to see if they
    plot correctly, but I think it is a problem localized to the
    DNA/GC track plotting.

     
  • Chris Fields
    Chris Fields
    2005-08-11

     
    Attachments
  • Scott Cain
    Scott Cain
    2005-08-11

    Logged In: YES
    user_id=616711

    Well, at least I can verify that this bug is real; last
    night, I created a screenshot almost identical to the one
    Chris created. I was hoping it was something as simple (and
    stupid) as the fasta file being in unix format instead of
    dos format (ie, just a CR instead of LF/CR), but that wasn't
    it. I'll work on it more today.

     
  • Scott Cain
    Scott Cain
    2005-08-12

    Logged In: YES
    user_id=616711

    I am punting to Lincoln, at least officially. I will
    probably still try to figure out what is going on, but if
    Lincoln looks, he'll probably track it down faster. The
    problem looks like it is in memory.pm, and is
    Windows-specific. This test script:

    #!perl -w
    use strict;

    use Bio::DB::GFF;

    my $db = Bio::DB::GFF->new( -adaptor => 'memory',
    -dir => 'C:/Program
    Files/Apache Group/Apache2/htdocs/gbrowse/databases/volvox');

    my $seg = $db->segment ( -name =>'ctgA', -class=>'Contig',
    -start => 1029, end => 1128 ) or die;

    my $dna = $seg->dna;

    print "$dna\n";

    returns the wrong string for those boundaries, but this only
    appears to happen on Windows and with the memory adaptor.

     
  • Scott Cain
    Scott Cain
    2005-08-12

    • assigned_to: scottcain --> lstein
     
  • Scott Cain
    Scott Cain
    2005-08-15

    Logged In: YES
    user_id=616711

    I've narrowed it down more: the problem is with DOS
    formatted files and Bio::DB::Fasta (I'm guessing it is a
    problem in the caloffset method, but I'm not sure). I
    indicated in a previous post that it didn't seem to matter
    whether it was DOS or Unix formatted file, but that is not
    true; if I use vim to `set ff=unix|dos` and then write the
    files and delete the indexes, I get either correct or
    incorrect behavior for unix and dos formatting respectively.
    I'm now going to take the following test script to the
    bioperl mailing list to see if anyone there knows the
    appopriate solution.

    Here's the test script:
    <code>
    #!perl -w
    use strict;

    use Bio::DB::Fasta;

    my $db = Bio::DB::Fasta->new('C:/Program Files/Apache
    Group/Apache2/htdocs/gbrowse/databases/volvox');

    my $dna = $db->seq('ctgA', 1029=>1150);

    print "$dna\n";
    </code>

     
  • Chris Fields
    Chris Fields
    2005-08-15

    Logged In: YES
    user_id=1327166

    I used Komodo 3.1 to switch the line endings of volvox.fa to
    Unix (\n) from Win32 (\r\n) and it seems to work now; that
    corroborates Scott's findings. Also, I tried switching to a
    mysql database (v. 4.1) and it works as well.

     
  • Logged In: NO

    This turns out to be a problem with the Bio::DB::Fasta module
    from bioperl when running under ActiveState perl. You can
    either reinstall bioperl from CVS, or simply replace
    \perl\site\lib\Bio\DB\Fasta.pm with the version found in gbrowse
    CVS (1.62 bugfixes branch) under
    extras\BioPerl\Bio\DB\Fasta.pm.

     
  • Lincoln Stein
    Lincoln Stein
    2005-08-22

    Logged In: YES
    user_id=224205

    Fixed in bioperl CVS. See the comment from "nobody" for
    explanation.

     
  • Lincoln Stein
    Lincoln Stein
    2005-08-22

    • status: open --> closed