[Bioperl-l] tests failing badly
Aaron J Mackey
Aaron J. Mackey" <amackey@virginia.edu
Tue, 10 Sep 2002 01:00:11 -0400 (EDT)
Turns out, to pass both SeqIO.t and LocationFactory.t you need this:
Index: Bio/Factory/FTLocationFactory.pm
===================================================================
RCS file:
/home/repository/bioperl/bioperl-live/Bio/Factory/FTLocationFactory.pm,v
retrieving revision 1.5
diff -u -b -B -r1.5 FTLocationFactory.pm
--- Bio/Factory/FTLocationFactory.pm 10 Sep 2002 02:24:36 -0000 1.5
+++ Bio/Factory/FTLocationFactory.pm 10 Sep 2002 04:48:54 -0000
@@ -196,6 +196,8 @@
# remove enclosing parentheses if any
$start =~ s/^\((.*)\)$/$1/;
$end =~ s/^\((.*)\)$/$1/ if $end;
+ $start =~ s/^\(//;
+ $end =~ s/\)$// if $end;
# Is this a simple (exact) or a fuzzy location? Simples have exact start
# and end, or is between two adjacent bases. Everything else is fuzzy.
Now all tests involving Locations pass cleanly; again, my apologies for
not understanding the full situation -- BioPerl has grown quite a bit for
one person to be able to "grok" all of it.
-Aaron
P.S. I also had to make this patch to t/data/test.genbank:
Index: t/data/test.genbank
===================================================================
RCS file: /home/repository/bioperl/bioperl-live/t/data/test.genbank,v
retrieving revision 1.1
diff -u -b -B -r1.1 test.genbank
--- t/data/test.genbank 30 Apr 2001 15:12:34 -0000 1.1
+++ t/data/test.genbank 10 Sep 2002 04:57:41 -0000
@@ -30,7 +30,7 @@
/clone="p427/428"
misc_feature 5.12
/note="Fuzzy location"
- misc_feature J00194:(100..202),1..245,300..422
+ misc_feature join(J00194:(100..202),1..245,300..422)
/note="Location partly in another entry"
BASE COUNT 118 a 46 c 67 g 79 t
ORIGIN
Presumably, the only way you can have ##..##,##..## is with a join()
around the whole thing, n'est-ce pas? Without this change, FTHelper
is very noisy on this line.
-Aaron
On Tue, 10 Sep 2002, Aaron J Mackey wrote:
>
> In fact, in reversing my changes to Bio::Factory::FTLocationFactory,
> you've messed up things. Here's what you have:
>
> $start =~ s/^\((.*)\)$/$1/;
> $end =~ s/^\((.*)\)$/$1/ if $end;
>
> This requires that start and end have *both* begin and end parentheses.
> With this code, I get lots of:
>
> Argument "202)" isn't numeric in numeric gt (>) at blib/lib/Bio/Location/Atomic.pm line 91, <GEN4> line 35.
> Argument "(100" isn't numeric in numeric gt (>) at blib/lib/Bio/Location/Atomic.pm line 91, <GEN4> line 35.
> Argument "202)" isn't numeric in numeric gt
> [...]
>
> With my patch:
>
> $start =~ s/^\(?(.*?)\)?/$1/;
> $end =~ s/\(?(.*?)\)?$/$1/ if $end;
>
> All tests pass cleanly. What's the issue here? Yes, when you mistakenly
> set a Location's start to "202)" it turns into 202, but what's wrong with
> having tests (and behavior) operate without spurious warnings? Is this a
> difference in your version of Perl and mine?
>
> -Aaron
>
> On Tue, 10 Sep 2002, Aaron J Mackey wrote:
>
> >
> > Remember that some of us quit working after 5 pm EDT; that means that if
> > you've found a problem with something I've committed, maybe you can just
> > ask me to fix it?
> >
> > In the development of Bio::Seq::EncodedSeq, I found that repeatedly when I
> > reset the seq() of an object, that PrimarySeq was "guessing" my alphabet,
> > and setting it to protein; I apologize greatly for not making sure
> > everything else tested out OK (but, in fact, all tests ran cleanly here).
> > But whatever you've changed seems to have introduced a myriad of warnings,
> > and now EncodedSeq seems to think it's a protein sometimes ...
> >
> > I'm now working on cleaning things up, because indeed I now have make test
> > issues ...
> >
> > mea culpa, mea culpa,
> >
> > -Aaron
> >
> > On Mon, 9 Sep 2002, Hilmar Lapp wrote:
> >
> > > I just did a cvs update ...
> > >
> > > Failed Test Stat Wstat Total Fail Failed List of Failed
> > > -------------------------------------------------------------------------------
> > > t/AlignStats.t 255 65280 15 11 73.33% 2-3 7-15
> > > t/EncodedSeq.t 31 ?? % ??
> > > t/LocationFactory.t 131 8 6.11% 13-15 21 23-25 31
> > > t/RepeatMasker.t 2 512 6 0 0.00% ??
> > > t/SeqAnal.t 255 65280 1 1 100.00% 1
> > > t/Tools.t 255 65280 8 4 50.00% 5-8
> > > Failed 6/105 test scripts, 94.29% okay. 18/4020 subtests failed,
> > > 99.55% okay.
> > >
> > > Many of which have to do with alphabet setting ... I'm going to
> > > investigate as I need to have this working.
> > >
> > > -hilmar
> > > --
> > > -------------------------------------------------------------
> > > Hilmar Lapp email: lapp at gnf.org
> > > GNF, San Diego, Ca. 92121 phone: +1-858-812-1757
> > > -------------------------------------------------------------
> > >
> > > _______________________________________________
> > > Bioperl-l mailing list
> > > Bioperl-l@bioperl.org
> > > http://bioperl.org/mailman/listinfo/bioperl-l
> > >
> >
> >
>
>
--
Aaron J Mackey
Pearson Laboratory
University of Virginia
(434) 924-2821
amackey@virginia.edu