[Biopython] SQL Alchemy based BioSQL

[Kyle Ellrott]

> Yes - Leighton Pritchard has also done a DBSeqFeature object
> (with lazy loading of the qualifiers too). I guess your code will
> be similar. This is something I think could well be worth merging
> into BioSQL (and doesn't depend on SQLAlchemy at all).

The version in my branch uses the sqlalchemy query compositions
methods I wrote when porting DBSeqRecord from the original
_retrieve_features function.  So his code would probably be shorter
step for now.

> Is this idea just a shortcut for explicitly loading the GenBank
> file into a BioSQL database (which hopefully will include an
> SQLite backend option soon), and then iterating over its
> records? e.g.

Yes, it would also have a copy of the biodb-sqlite schema stored as a
string in the module, so it could build an in-RAM database on demand.
Make the setup and loading automatic.  It would appear to be just like
a regular file parser.  That way, if we start writing crazy feature
filters methods based on SQL queries, they can be easily reapplied to
file based usage.  And we wouldn't have to write a feature filter for
the database objects and another for file based objects.

Kyle