by J. William Bell

The atom's iconic image--with electrons whizzing around the nucleus in tidy ellipses--is elegant. But it's also wrong.

For most of the last century, scientists have taken a more complicated view based on the advent of quantum mechanics. The electron still spends its life in orbit around a nucleus. It's a peculiar life, however. The electron doesn't run circuits around a track; it bobs and weaves through a misshapen volume of space that confines the particle without further defining its location. There's no precise spot or series of spots that will house an electron, only a statistical likelihood that the electron will be in a spot at a given moment.

Scientists have given up the clockwork conception of an atom's state as described by compass-drawn transits, leaving them with a game of probabilities.

But probabilities are not without power. Modern chemists can use quantum-mechanical data to determine the reactivity of atoms, molecules, and large biomolecular systems. Atoms and more complex structures all have what are known as wave functions, from which critical features cascade. With the wave function as their guide, researchers can determine a structure's charge information, its highest occupied and lowest unoccupied orbitals, and the orbital electron density. Every chemical reaction is controlled by these characteristics.

Though quantum-mechanical studies of single molecules and small systems are widespread and influential, these studies are more seldom conducted on larger systems. A team from Pennsylvania State University is making them less rare. They are currently soldiering through about 5,000 protein and nucleic acid structures, calculating the systems' fundamental quantum-mechanical features.

Using NCSA's SGI Origin2000 supercomputer, the team has already completed about 2,000 structures, each composed of about 2,000 atoms. They currently plan to move their work to NCSA's IBM p690 system as the Origin2000 is retired.

"Computational chemistry traditionally hasn't been a reuse discipline," says Kenneth Merz, who leads the team and is a chemistry professor at Penn State. "Researchers do their calculations, write their papers, and much of what they've done is repeatable but unrecoverable. [With the constant advance of computing power], it's easy to do some of the basic calculations, but why do them over and over?"

Motivated by that simple question, Merz's team is building what they call a quantum bioinformatics database to make the data available to anyone with Web access and an interest in conducting higher-order theoretical work. Among other applications, this work can include identifying molecules that are promising therapeutic drug candidates. Beyond that, they can be used to identify what parts of those molecules are most important to the formation of the molecules and the curative interactions that they take part in.

"The data that we are populating the QBD [quantum bioinformatics database] with can be used down the line. Calculations that we can't even think of now will have something to start with."

go to page 2

Access Online | Posted 1-13-2004

Electrostatic potential map developed using classical atomic charges for the FK binding protein bound to an immunosupressant known as FK506.

Electrostatic potential map developed using divide-and-conquer method for the FK binding protein bound to an immunosupressant known as FK506. Electrostatic potentials are available in the QBD as atomic charges or as stored surfaces.