On Wed, 28 Mar 2001, Chris Phoenix wrote:
> As to the protein folding problem, several years ago Drexler was saying
> that you don't have to solve the entire protein folding problem in order
> to design proteins that will fold the way you want them. Does anyone
> know if this logic has proved correct, and if anyone is working along
> these lines?
That wouldn't happen until you had the protein folding problem at least
partly solved. If you're working on folding, why would you want to work
on nanotech which won't be realized anytime soon, instead of biotech,
which will. However, it may be soon, or we may already be at the point
where one could begin to do fundable research on protein design for
nanotechnology, using existing folding prediction methods.
At the APS meeting this month, I saw examples of 150-kD proteins folded
to topological accuracy (contacts generally within a nm or a few of
positions determined separately by XRD or NMR). These predictions were
greatly aided by kinetic modeling based on free-energy channeling. From
the conformational constraints imposed by "contact order", or the number
of backbone positions lying between two amino acids residues in physical
proximity in the folded protein, one calculates an entropy for topolgical
configurations which can be used to compute the free-energy surface.
The topological results were very good; what was surprising was the
short-range inaccuracy. It was as if one had NMR data to a few-nm
resolution. However, these methods address only topological or low-res
geometrical order; to get a final structure one should do high-accuracy
molecular dynamics, as is done to refine XRD structures in the angstrom
> For protein folding, can't you stack the deck in your favor by making
> the protein much more stable than most biological proteins seem to be?
> In other words, design it to fold so firmly that you don't have to
> simulate it?
Yeah, you can use highly-stable subunits if you can identify them, and you
may be able to use nonnative residues and chemistry to make stronger
bonds, but all of this will only make the simulation problem easier, it
won't eliminate it. If you have a deeper well, you don't have to
calculate to as high accuracy, but you still have to calculate.