Structural biology should be computable.
It's been known for over 40 years that
protein structures are completely determined by their amino acid sequences.
For almost all protein structures and all protein-protein complexes,
the experimentally observed structures and
conformations are almost certain to correspond to global free energy minima.
So, it should be possible to predict the structures of proteins and
protein-protein complexes readily by
identifying the global free energy minima for a polypeptide chain,
in case of protein structure prediction,
or identifying the global free energy minima for two proteins coming together,
which will be the prediction of protein complexes problem.
If we could do this, it would both be a fundamental test of
our understanding of macromolecular and interactions,
and it would also be huge practical relevance as
the cost of determining protein structures computationally,
would be a small fraction of the cost of
current experimental methods such as X-ray crystallography, and NMR spectroscopy.
But, as you know, today,
structural biology is not computed,
it is primarily experimental science.
And what I'm going to tell you about today is progress
towards making structural biology computable.
The work I'm going to tell you about today is carried out
with a computer program being developed in my group,
and groups of people left my group in
the last seven years that has the following structure.
We have a model of the energetics of inter and intramolecular interactions,
which allows us to compute the energy of
conformation of a protein or a protein- protein complex.
And given that model,
we can do one of two things.
We can either do a prediction problem,
in which we're given for example,
the sequence of a protein and asked to find
the lowest energy structure for that sequence,
that would correspond to the Ab initio structure prediction problem.
We can also take the structures of two proteins,
and try and find the lowest energy docked arrangement,
that would be the protein-protein docking problem.
In these cases, we're given the sequence or
the structures and trying to find the lowest energy conformation.
Now, the inverse problem is the design problem,
where we're given a structure and we want to find the lowest energy sequence.
So for example, that would be the problem designing a new protein structure,
a sequence that would fold to give a new structure or the problem of
given a protein-protein complex designing an interface between the two proteins,
which will allow them to bind to each other tightly.
This approach has been extended to protein ligand interactions.
So for example, ligand docking,
the design of new enzymes and to protein DNA interactions.
In particular, the design of new DNA binding proteins with new specificities,
which is something that we've had a fair amount of success with lately.
However in this lecture, I'm going to focus on the prediction and design of
protein structure and protein-protein interactions.