I think combinatorial chemistry has a lot in common with epidemiology.
Both build a database by measuring everything under the sun and then
hunt for causality amidst a mass of correlations.
In some sense this is the same approach taken by physicians (both
ancient and modern) when confronted with a disease state whose mechanism
is unknown. They try everything, and look to see what works. It seems
only fitting that this combinatorial work is, today, largely carried out
by robots.
Mechanism and causation are closely related; hence, I cast my vote for
mechanism over combinatorial chemistry. Science without imagination is
pretty boring. Combinatorial chemistry may identify an important drug,
but it will fall into disuse as soon as someone discovers a means of
combining the robot's efficiency with a mechanistic directive.
The same hubris cited by Athel in Nature Biotechnology is evident in
some of what's written about the promise of the human genome project. As
one example, consider the vast resources being expended to fill the cDNA
and EST databases with sequences from "normalized" tissue libraries.
Since the tissues chosen are all made up of multiple cell types, the
normalization procedure assures that no cell-specific expression
information can ever be extracted from these databases. Consequently,
the integrative biologist, intent on understanding the metabolic
machinery of a single cell type, will have no more information when the
EST database is complete than he or she has now. This is the sort of
legacy left by combinatorial science.
Bob
--
Robert D. Phair, Ph.D. rphair at ix.netcom.com
BioInformatics Services http://www.webcom.com/rphair
Partnering and Outsourcing for Computational Biology
