[Biojava-l] Added Genbank writeSequence

[Matthew Pocock]

Thanks Keith, great work. I will have a look at the CompoundLocation 
funnies.

Keith James wrote:

> I've added a GenbankFileFormer class, analagous to EmblFileFormer, to
> enable Genbank writing. Both EMBL and Genbank writing are as complete
> as the parsing (i.e. limited header information), but see the bug
> below. With this fixed, we should be able to convert EMBL/Genbank
> without loss.
> 
> Unbounded fuzzy locations are not being written correctly for
> locations with unbounded min (but are fine for unbounded max). This
> seems to be due to a bug in CompoundLocation, which I can't
> locate. The symptom is that I can't cast (Fuzzy) Locations obtained
> from a CompoundLocation to FuzzyLocation where hasBoundedMin is false,
> while it is fine where hasBoundedMax is false. The odd thing is, I had
> no problem with the cast immediately before the object was added to a
> CompoundLocation.
> 
> The code for generating an EMBL/Genbank location string from a
> Feature's location is in the static method formatLocationBlock in
> org.biojava.bio.seq.io.SeqFormatTools.
> 
> I've taken the liberty of adding a static final Map FORMATS to the
> SequenceFormat interface where SequenceFormat implementations can
> register the format(s) accepted by their writeSequence method. This
> became necessary for EmblLikeFormat which needs to write EMBL and
> Swissprot (and possibly others later). The interface now has
> getFormats() and getDefaultFormat() methods.
> 
> Finally, features are not retrievable from a FeatureHolder in a
> defined order. Ordering by location is required for Genbank/EMBL
> writing, so I have added a Comparator to the Feature interface which
> compares by the int returned by getMin() of their top-level
> Location. As the Location interface has an analagous Comparator nested
> in it, this seemed a consistent place to put it.
> 
> The build seems okay, but if the changes have caused any problems, let
> me know.
> 
> cheers,