Added FALDO based position/location serialization.

[KimJBaran]

No description provided.

[KimJBaran]

A discussion about improving the implementation has been started for the corresponding BioRuby pull request: bioruby/bioruby#95

[KimJBaran]

Latest build failure seems to be due to a network connection problem in one of the environments -- not due to latest commit.

[peterjc]

Does this consider compound locations (joins in GenBank/EMBL terminology)?

[KimJBaran]

Are those covered by classes in the SeqFeature file? If so, then yes, otherwise, possibly not.

[peterjc]

Class CompoundLocation defined in SeqFeature.py does not have a new ._rdfize(...) method (like class FeatureLocation and most of the position classes).

[KimJBaran]

Right. I remember now. I did not implement CompoundLocation because it was more complex.

[peterjc]

Yep - property bags/lists required...

[KimJBaran]

Actually, I re-read the Turtle spec recently and the use of lists is well supported in that format (lots of syntactic sugar provided). If it is only about using an RDF list, then this could be implemented rather elegantly and quickly.

[peterjc]

This rebases cleanly onto the current master; I've also removed some unneeded imports in test_RDF.py and applied some PEP8 style changes (part of a general house keeping we're working on for the whole code base, see #406).

My work in progress is here (which I will likely rebase from time to time): https://github.com/peterjc/biopython/tree/faldo

[KimJBaran]

I have implemented a preliminary -- possibly incomplete -- RDFization of CompoundLocation objects.

@peterjc Please have a look at the use of rdfs:member. I recall there was a discussion about swapping a FALDO doman/range restriction, which might apply here, but I cannot find that discussion anymore.

If RDFized like I implemented it, it is currently not clear to me how I arrange the FeatureLocation objects in a list.

[peterjc]

See range/domain discussion at OBF/FALDO@43834f4 - fixed as of OBF/FALDO@4cd5bba which was merged (it should be range).

[peterjc]

Is seems an additional RDF tuple is required to link the generated FALDO location to the feature it belongs to. Perhaps the RDF method needs to return two values, the FALDO string as now, plus the URI of the location (or compound location) which the caller can then use?