To my knowledge, there are no existing structural biology libraries for Haskell; or if there are, they are effectively invisible - which amounts to the same thing.

BioHaskell is a relatively new addition to the Bio* projects family, presumably because interest in the language for bioinformatics is relatively recent. At present it contain libraries only for sequence-based data.

If you were to develop a library, it should presumably have the same functionality as the existing libraries for other languages: e.g. Bio::PDB for BioRuby, Bio::Structure for Bioperl or Bio.PDB (PDF) for BioPython.

However, given the existence of these already-excellent libraries, I would question the wisdom of developing yet another for a little-used language. If you use only Haskell, you need the functionality, you're happy to write it and you think it's worthwhile and useful to others then fine, go ahead. If you want to get your research done quickly, it's better to learn how to use what's already available.

There are few Protein Data Bank parsing modules that appeared on GitHub over the last year. There is for example long-developed hPDB on Hackage, which is mostly tuned for speed and convenient high-level interface.