Protein Databank file format library.

Protein Data Bank file format is a most popular format for holding biological macromolecular data.

This is a very fast sequential parser:

• below 7s for the largest entry in PDB - 1HTQ which is over 70MB - as compared with

• 11s of RASMOL 2.7.5,

• or 2m15s of BioPython with Python 2.6 interpreter.

In its parallel incarnation it is most probably the fastest parser for PDB format.

It is aimed to not only deliver event-based interface, but also a high-level data structure for manipulating data in spirit of BioPython's PDB parser.

hPDB - Haskell library for processing atomic biomolecular structures in Protein Data Bank format - Michal Jan Gajda. BMC Research Notes 2013, 6:483.

hPDB

Haskell PDB file format parser.

Protein Data Bank file format is a most popular format for holding biomolecule data.

This is a very fast parser:

• below 7s for the largest entry in PDB - 1HTQ which is over 70MB
• as compared with 11s of RASMOL 2.7.5,
• or 2m15s of BioPython with Python 2.6 interpreter.

It is aimed to not only deliver event-based interface, but also a high-level data structure for manipulating data in spirit of BioPython's PDB parser.

Details on official releases are on Hackage

This package is also a part of Stackage - a stable subset of Hackage.

Projects for the future:

Please let me know if you would be willing to push the project further.

In particular one may considering these features:

• Implement basic spatial operations of RMS superposition (with SVD), affine transform on a substructure.
• Use lens to facilitate access to the data structures.
  • torsion angles within protein/RNA chain.
• Add Octree to the default data structure (with automatic update.)
• Migrate out of text-format, since it gives portability trouble, and slows things down when printing.
• Write a combinator library for generic fast parsing.
• Checking whether GHC 7.8 improved efficiency of fixed point arithmetic, since PDB coordinates have dynamic range of just ~2^20 bits, with smallest step of 0.001.
• Class-based wrappers showing Structure-Model-Chain-Residue-Atom interface with possible wrapping of Repa/Accelerate arrays for fast computation.

Please ask me any questions on Gitter.