[Bioperl-l] Re: Sim4

hilmar.lapp@pharma.Novartis.com hilmar.lapp@pharma.Novartis.com
Fri, 28 Jul 2000 19:07:08 +0100

It is also using the outdated Bio::SeqFeature::Homol. I'll change that to
the newer Bio::SeqFeature::FeaturePair.

     Fine. This was my idea, too.

     There's a little semantic problem left with the notion of Exon for
     Sim4 alignments and the notion of feature and hit in FeaturePair. For
     Sim4, you put the shorter sequence first, so that some people might be
     referring to this as the query. In the FeaturePair object, query will
     however refer to the sequence having the Exons, that is, the second
     (genomic) sequence, because you obviously expect
          $exon_start = $exon->start()
     return the right result. $exon->hstart() would then return the
     corresponding coordinate on the 'other' sequence.

     I'll check that Sim4/Result.pm correctly initializes the Exon object,
     correctly meaning in the above sense.

     Then there's still ExonSet which is missing, but I haven't been able
     to find any special functionality this is supposed to provide. Do you
     remember the rationale behind this module? Could it be replaced by a
     2-dimensional array of Exon objects? Or, easier, and maybe more
     elegant, to handle, an array of Bio::Seq objects (without actual
     sequences) each having an array of Exon objects as Feature.

     Another possibility, which I tend to favor: process the Sim4 result in
     a similar way to multiple sequence files, that is, the user is
     requested to call repeatedly something like parse_next_alignment(),
     with each call returning an array of Exon objects.

     If there are no objections I'll go for that.