[Bioperl-l] "richer" tree nodes

[Jason Stajich]

On Mon, 15 Jul 2002, Aaron J Mackey wrote:

>
> Well, of course the ancestral sequence object is next on my hit list; and
> I've found that turning things into a TreeFactory isn't so hard at all
> (instead of new Bio::Tree:Tree, it's just $self->treetype->new()).  Unless
> anyone objects, I'll continue down line #2 ... that way Tree::Node::Node
> stays "clean".

this is fine in my mind - I think that if we wanted to define some
'generic' additionally annotations/features/info that we want to attatch
that are required to be represented by strings then I have no problem.
But attatching whole sequences objects will call for new type of factories
- hence the whole factory design in the first place as I imagine won't be
pulling this info from a generic newick file....

>
> slightly off topic, but does anyone have any plans for implementing a few
> more tree methods like:
>
> $n1 = $tree->find_node('node_id');
> $n2 = $tree->find_node('node2id');
> $n3 = $tree->lca($n1, $n2); # last-common-ancestor
> unless ($tree->monophyletic($n1, $n2, $n3) {
>   # should never happen
>   die "Huh?";
> }
>
> # more useful example:
> my @nodes = map { $tree->find_node($_); } qw(eco hae stm ype vch pae);
> unless ($tree->monophyletic(@nodes)) {
>     warn "bacterial sequences not monophyletic - lateral gene transfer!?!";
> }
>
> These aren't calculations so much as topological investigations ...

oh yes - I really want these in there, I just haven't had a chance to
program them in.  I'd like to work on the monophyletic one if you haven't
already...



>
> -Aaron
>
> On Mon, 15 Jul 2002, Ewan Birney wrote:
>
> >
> > In this sort of case, I reckon 3 is both the simplest and the sanest - if
> > you wanted to attach truely something funky (an ancestral sequence object,
> > pubmed abstracts, kitchen sink, whatever) I would have suggested the
> > factory solution but reasonably-common statistics or (even) a pretty
> > generic tag-value system is "ok" in my book.
> >
> >
> > This is in the keep-it-simple-until-it-starts-to-look-ugly design mode
> >
> > (KISUISTLU principle... now that is catchy! you heard it here first
> > folks...)
> >
> >
> >
>
>

-- 
Jason Stajich
Duke University
jason at cgt.mc.duke.edu