Deciphering molecular structures requires indirect approaches to tease out information about the positions of the atoms in a molecule. Two of the most commonly used techniques are X-ray crystallography and NMR (nuclear magnetic resonance) spectroscopy. In both cases, the researcher, with the help of numerous computer programs, pieces together a model of the molecule that best fits the experimental data.
In X-ray crystallography, researchers shine X-rays on crystals that contain trillions of copies of a molecule. The crystals diffract the rays in patterns that, when analyzed mathematically, reveal the positions of the atoms within the molecules.
NMR spectroscopy uses molecules in solution rather than in a crystal, and relies on the innate property of atoms to orient themselves in a magnetic field. In this situation, atoms produce distinct "signatures" that give clues to their environments within the molecule. About 15 percent of the structures in the Protein Data Bank (PDB) were solved using NMR.
These procedures result in a plethora of information — how the protein was extracted and purified, how it was prepared (crystallized or concentrated in solution), as well as the raw data from the experiments themselves that reveal the atomic coordinates. While both techniques yield maps of atom locations within the molecule, there are many differences in the data leading to the final coordinate file.
The mmCIF (macromolecular crystallographic information file) format — the set of data dictionaries used in the PDB's relational database — was developed specifically for crystallographic results, but additional dictionaries are in the works for NMR. "NMR data are quite a bit more complex than X-ray data, much richer and more difficult to capture," says the University of Wisconsin's John Markley, who also heads the NMR database for the BioMagResBank (BMRB).
The PDB and the BMRB teams are working together to ensure that the important data are archived efficiently; they are also developing a common deposition tool so that the appropriate data go to the right database.
Back to Banking on Structures