|
|
|
|
|
One of the holy grails for simulation scientists in the
chemistry and biology fields is gaining a detailed
understanding of the mechanisms involved in the folding
of a protein molecule. This phenomenon is commonly known
to occur as part of defining the functional work a protein
does in a biological system, but precisely how it happens has
remained something of an unsolved mystery. A robust
computational approach should be able to provide some
explanations, but at several hundred atoms for the protein
itself and several thousand more for any solvent that
may need to be studied, the cost has been prohibitive.
In addition, the time scale on which this happens is
quite large in computational terms. To overcome these
limitations, it has been necessary for researchers to
employ the absolute state-of-the-art in both algorithms
and computing resources.
Obviously, it would be wonderful to be able to conduct very detailed quantum chemistry studies on these species, but the thousands of atoms in such a system quickly rules out that possibility. Even traditional molecular dynamics (MD) methods, which are designed to handle systems of large molecules or with large numbers of smaller molecules, easily fails at this task because of computational limitations. Protein folding is generally believed to occur on a scale of milliseconds to seconds, and traditional MD is usually good for a nanosecond or so of simulation in a reasonable amount of time. Obviously, new methods have been required to approach problems of this magnitude. There have been attempts to force the "virtual protein" (protein model) to complete its folding much faster than it does in reality, thus providing some insight into the problem in a more tenable period of time. Other algorithmic advances have been employed in approaching this problem, too, but probably the most significant advance has been the full exploitation of our nation's massively-parallel computers. Utilizing several to several hundred processors to perform a calculation can provide a very prominent speed-up, and make simulation applicable Go To Page: 1 2
The copyright of the article Simulation Studies of Protein Folding
in Scientific Computing is owned by . Permission to republish Simulation Studies of Protein Folding
in print or online must be granted by the author in writing.
|
|
|
|