Frank Fuerst <ffrank at rz.uni-potsdam.de> wrote:
:klenchin at REMOVE_TO_REPLY.facstaff.wisc.edu (Dima Klenchin) wrote:
::>"deodiaus" <deodiausNOSPAM at gnu.ai.mitNOSPAM.edu> wrote:
:>:All the CPU throughput won't help if you don't first solve the mathematics
:>:correctly.
::Ever heard of numerics?
::[...]
:>: I am trying to understand how to tackle this protein folding problem.
:>:Does anyone know of any good resources about how to solve the problem from
:>:first principles. I have read people's papers on methods based on
:>:sophisticated computational dynamics, but those avoid the approach I was
:>:looking for.
:>:>The problem, as I see it, is compounded by the fact that many
:>protein folds (probably most? - famous RNAse seems to be an
:>exception) do not necessarily represent global minimum solution
:>(umm, this is not my field so I don't know the "correct term" to
:>use but the idea is that they are not in the lowest energetic
:>minimum state), just one of those local minima that are deep
:>enough for the fold to be "stable".
::It is generally asumed that proteins usually do indeed fold to the
:global minimum of Gibb's free enthalpy, and this is experimentally
:well supported for many of them (though there may be exceptions - but
:I'm currently not aware of any).
This is interesting. How can it be experimentally supported
without ability to calculate all possible minima and know exactly
which one is global? I supposed calorimetry can be used but there
you'd need to be able to create all possible alternative folds -
this isn't going to happen because in such experiement you
are limited to what protein does under your conditions (and this
means not necessarily all and not necessarily same as inside
the cell).
Sure, many small proteins do renature readily into native state,
but still many do not. The bigger the size, the bigger the problem.
Take actin. Not only it cannot be renatured, it cannot even be
folded properly in E.coli - there are _specialized_ actin-specific
chaperones that fold it in the cell. I thought it is generally
accepted that chaperones are actively involved in the folding
process of many (most?) eukariotic proteins. If so, I can
easily imagine that they burn ATP not only to accelerate reaction,
but also to push it into alternative pathway.
Then there is my humble experience: PIP-5 kinase expressed
in E.coli in inclusion bodies "renatures" relatively easily into
some structure, which is perfectly soluble, compact (not a
random coil), but which has nothing to do with the native protein -
no activity, no tri- and hexamers formation. (This is not due to
post-translational modifications because small amount of
enzyme that happen to be soluble in E.coli is perfectly active).
:On the other hand, solving the Schrödinger Equation for a molecule as
:complex as a protein is not feasible - it is impossible to solve it
:analytically, and numerics get extremely complicated and
:time-consuming. I don't think that we'll be able to even solve the
:Schrödinger Equation for small peptides in the next decades, so better
:not bother about proteins...
::>If so, then even if we solve all the math and computer problems
:>that exist, we still cannot tell which one of many possible
:>folds represents native protein.
::This shouldn't be a major problem.
I think it is. As a rule, in the absence of crystal we have no clue
about structure based on sequence (homologs with structure
don't count). Why do you think it isn't?
:>IMHO, the solution is again
:>in experiments, not in paper theory - we need to understand more
:>about just how and based on what proteins do fold in the cell.
::This is right, though "theoretical experiments", i.e. molecular
:calculations, be it classical molecular dynamics or (semi-)quantum
:mechanical stuff, are also very useful. Kind of "dialog" between
:experiments and theory.
:Yep, of course. The cross-talk is essential. My only point was
that the bottleneck is at experimental stage now.
- Dima
