Computer-generated vaccines

Tracy P. Hamilton chem013 at uabdpo.dpo.uab.edu
Sat Nov 20 12:01:11 EST 1999


I am just wandering in for the day, so don't expect any further response.

Pierre wrote in message <814hkq$965$1 at news2.isdnet.net>...
>I agree with Tyson. Computer are more and more helpful. but we are still
not
>able to calculate reliably the 3D structure of a single protein. Thus,
>saying that one day we will be able to calculate a vaccine with a
>supercomputer is science fiction. May be we will. May be not. Notice also
>that it is a reductionist/determinist dream that might be totally
irrelevant
>to biology. For example, will the same supercomputer be able to calculate
>the future evolution of man ?

Progress is being made in predicting the 3-D structure of proteins.  By the
end
of next century that will not be a problem.  However, there are more
problems than
that in fighting disease.

The least of which is finding out which protein is being affected by the
disease.

Computers are used heavily now in what is called rational drug design:

Approach 1 - active site not known

1) Take a known drug that works.
2) Build a model of the active site where the molecule
determines its shape and charge distribution of your model site.
For example, if the drug has a hydrophobic chain, there probably is
a hydrophobic pocket at a corresponding position in the active site.
3) Make changes to the drug.  From the effects, one can refine the model
shape and charge distribution of the active site.

Approach 2 - active site structure known

1) Make a theoretical combinatorial library of molecules that bind to
the site.
2) Compute the binding of each.
3) Examine the shapes and charge distributions of the best, try different
types of drugs that will give similar ones.

What can be done quite nicely (although there is room for better
quantitative accuracy that will also come) is computing of binding
to an active site.

However, just because a molecule has the best binding ability does not
mean it is the best drug.  It also has to *get to* the protein (or DNA, or
RNA),
not be metabolized, the treatment has to depend on strength of
binding to the active site (best for competitive inhibition), and it should
not
bind to other active sites as well, which can cause side effects.

It is a long, tedious process.

One way to try to reduce the side effects is to introduce rigidity into the
molecule.  This would work best where a flexible molecule takes one
shape in binding to the active site desired, and another in binding to the
site which causes side effects.  If you can make a drug with the rigid
conformation close to the first, it will not bind as well to the second.
Rigidity can be introduced primarily by using rings instead of chains.

Combinatorial chemistry relies more on automation rather than
*calculations*,
although computer data analysis and process control is required.

I have heard this approach called irrational drug design.  There is no clear
advantage of irrational to rational.  The major drug companies use both.

Experiments will always have to be done in the foreseeable future,
although theory will help in both the screening of compounds AND
in the study of the basic processes.  Theory is also a major component
of the latter.

Tracy P. Hamilton








More information about the Immuno mailing list