- American Physical Society Sites
- Meetings & Events
- Policy & Advocacy
- Careers In Physics
- About APS
- Become a Member
Researchers from Los Alamos National Laboratory and Stanford University have developed a promising new approach for computer modeling of the structure and behavior of molecules using standard workstations. The technique has been successful in simulating a single polyalanine peptide --a short protein component comprised of about 160 atoms--and additional experiments are underway to confirm that the same technique works for more complex proteins containing thousands of atoms.
Developed by LANL's Niels Gronbech-Jensen and Stanford's Sebastian Doniach, the technique employs calculated shortcuts so that the simulations can range from the level of individual atoms to the unfolding of the entire protein as it interacts with its environment--a range of detail in a desktop simulation that is currently beyond the reach of even the most powerful supercomputers.
"Because of the vastly different time and spatial scales involved, no one previously has successfully simulated this range of time scales using this kind of detailed model," said Gronbech-Jensen. "We also showed that the final shape the peptide moves toward depends on the original positions of the individual atoms. That's why multiple simulations are important; these molecular systems are inherently chaotic."
As a protein responds to its surroundings, it can reshape itself, and its final shape determines how it interacts with other molecules and thus achieves its biological function. The physical positions of individual atoms with respect to their neighbors are critical for the final conformation of the overall molecule. However, their relativistic positions may change trillions of times per second through chemical interactions.
Tracking these details slows a simulation to a snail's pace, placing a model of the molecule's overall behavior beyond the reach of all but the most powerful supercomputers, like the CM-5 at Los Alamos, which is ideal for problems involving many interactions because it breaks up a calculation into many different parts that it works on simultaneously before compiling the results into a single product. "But such tools are not readily available to most researchers," said Gronbech-Jensen. "Plus, we want to do thousands of simulations to compile statistical information on how the molecules behave, how often they assume one final shape over another. This all argues for having a computational method that will work on a workstation."
The research team overcame this obstacle by introducing a mathematical constraint on the atomistic bonds, effectively ruling out the rapid oscillations of individual bonds from the calculation, while retaining other important movements, such as rotations of amino acids around the backbone of the protein chain. They plan to calibrate and test their simplified model by comparing those results to full-scale simulations run on the CM-5.
Still another challenge is that the most common surrounding medium of molecules is water. The hydrogen atoms in water molecules can cut or form bonds with atoms scattered along the protein, which in turn affects the molecule's final conformation. When water is included in a simulation, more parameters are added to the calculations, effectively negating the advantages gained by using the simplified atomistic picture. To overcome this, the researchers plan to use the CM-5 to define effective "potentials"--a number that represents the sum total of interactions between the various atoms in the protein chain and the surrounding water --which will represent an average effect. Once calculated, their values can be plugged into any simulation.
"If our more advanced simulations are successful, for the first time we will be able to make realistic predictions or compare our results with experiments," said Gronbech-Jensen, adding that no one has yet bridged this gap between simulation and laboratory observation. The team eventually hopes to develop a commercial computer tool for designing molecules with specific functions.
©1995 - 2021, AMERICAN PHYSICAL SOCIETY
APS encourages the redistribution of the materials included in this newspaper provided that attribution to the source is noted and the materials are not truncated or changed.