[Biojava-l] FramedFeature interface

[Thomas Down]

On Tue, Sep 04, 2001 at 08:14:02PM +0100, Simon Brocklehurst wrote:
> 
> In general, I think there are no right or wrongs with this kind of thing.  For
> example, in regard of *sequences* starting at zero or one, it can obviously be
> sensibly argued that index positions should start at zero, to mirror the way
> programming languages tend to work.
> 
> In regard of reading frames, I'm unclear about the point of referring to the
> frame directly by the *value* of an integer.  Rather, I would have it that the
> *name* of a public static final int constant be used e.g.
> FramedFeature.FORWARD_1 or FramedFeature.PLUS_1. That is, a name with a clear
> meaning.  In this case, the actual *value* of the constant itself is irrelevant.

I've been trying to stay out of this one, but just a quick
commnet now that I see this:  I think Frame and Strand are
actually orthogonal properties.  An Exon in a DNA context
clearly has both Frame and Strand.  However, an exon in
(unspliced, presumably) RNA context arguably only has Frame:
reverse-strand features on RNA don't generally make much
sense (unless you're talking viruses, maybe...).

On the other hand, at this point we're talking about
multiple inheritance of Feature interfaces, which might
make life rather difficult -- especially with regards
to the templating mechanisms :(.

What do people thinK?

   Thomas.