NMR and Supercomputing

For the last 15 years, James and other researchers who wanted to understand how molecules move have been able to glean some understanding using NMR, which bombards atoms with radio waves when they're in magnetic fields, creating a spectrum that acts like a fingerprint. These spectra, explains James, make it possible to decipher molecular structures in solution, and they can potentially give insight into dynamic aspects of those structures. "We were able to talk to a certain extent about frequencies and amplitudes of motion," James says, "but we have not been capable of saying anything very meaningful. For example, I may know that this particular bond moves 60 degrees and that it fluctuates back and forth in one nanosecond. But, in fact, by my saying that, I still didn't know exactly what the motion was."

Nowadays, James is developing computational techniques that incorporate experimental structure and dynamics results from NMR, which essentially supplies details about distances between hydrogen atoms and torsion angles between bonds, as well as previously available information such as length and angles between bonds. "If we use that available information, plus experimental information, we should be able to conduct an intelligent search of the millions of possible conformations to find out which ones are consistent with experimental data. So what we'll end up with is not just a single structure, but a family of structures that details the molecule's movement in solution."

Using Pittsburgh's C90, James and his colleagues have performed their calculations with software packages they developed. Known as MARDIGRAS and CORMA, they determine distances between atoms based on NMR data. And to determine how the atoms move and also their energy levels, the researchers use the AMBER software package developed by colleague Peter Kollman's group at the university.

go back to the main page