Hi Chandrika,
protein structures are clustered into a limited number of energetically
favoured stable folds (a few thousand). I.e. proteins adopt the same fold with
certain modification if they belong to the same protein family, and even
sometimes proteins with different function adopt the same general fold. This is
the basic idea of comparative or homology modeling: we can model a wide range
of proteins if at least one structure is known from the given fold family.
However comparative modeling is not able to model a fold starting form a
template that belongs to another fold. In your practical approach probably one
of the followings will happen: A., as you move away sequentially in each step
from your original template you accumulate more and more errors in your new
models. Finally you will have a structure which will have nothing to do neither
with the original template nor with the desired new fold. The number of
possibilities of new conformations increase exponentially as you are moving
away structurally from the original template. Outcome B ., is that finally you
will thread a fully diverged sequence (the sequence of the final target) onto a
backbone which resembles very much the original template with certain
fluctuations at some parts. These are my guesses, the outcome will depend on
the comparative modeling program you apply: whether the program copies in a
very conservative manner the original backbone of the core or they allow some
fluctuations and use optimisation at the end. Modeller is rather the second
one.
Andras
Chandrika Mulakala wrote:
>
> Hi,
>
> I have question that may sound a little naive. It has more to do with homology
> modeling than modeller itself.
> Here's the question:
>
> Since, through homology modeling it is possible to model the structure of a
> molecule, given the structure of another molecule that is fairly similar,isn't
> it possible theoretically, then, to be able to model the structure of any
> protein in that manner? Why isn't it possible to used a modelled structure as
> a template for further modelling?
>
> For example: we could start with an accurate structure of some protein
> molecule, change a few residues and then model the mutant. Then use the mutant
> as a template, change some more residues and then model it. In this way isn't
> it possible, theoretically, to model any protein? What are the practical
> problems with this idea?
>
> Thanks,
> Chandrika
--
,
Andras Fiser, PhD # phone: (212) 327 7206
The Rockefeller University # fax: (212) 327 7540
Box 270, 1230 York Avenue # e-mail:
New York, NY 10021-6399, USA # http://salilab.org/~andras