Silicon Investor (SI) -- The First Internet Community

STOCKTALK

We've detected that you're using an ad content blocking browser plug-in or feature. Ads provide a critical source of revenue to the continued operation of Silicon Investor. We ask that you disable ad blocking while on Silicon Investor in the best interests of our community. If you are not using an ad blocker but are still receiving this message, make sure your browser's tracking protection is set to the 'standard' level.

Biotech / Medical : ArQule -- Ignore unavailable to you. Want to Upgrade?

To: ahhaha who wrote (178)	2/6/1998 1:33:00 PM
From: Peter Silsbee	Respond to of 399

>> QED. I'm sorry, what did I just prove? We were discussing the dimensionality of the search space. You claim that SBD input, as well as feedback from screening, increase the dimensionality, I claim both decrease it. My point of view is this: we pick a space (all combinations of some particular set of atoms or groups or however they do it. I'm not a chemist.) of interest. An exhaustive search of the space is utterly out of the question. Not enough time in the anticipated future of the universe. So we do two things: first, we ask the SBD group to tell us in which subspace(s) we should be looking. Bingo, the dimension is reduced. Sure, it is possible that some good solutions (in the "problem-solving" sense, not the chemistry sense) have been rejected at this stage, but this will happen with a pure SBD process as well. Next, we generate a few hundred thousand compounds and do some screening. Results of this screening are used to narrow the search space further. Dimensionality is down again. You raised the concern (if I understood you correctly) that molecules which are structurally similar can be functionally quite different. The "directed array" concept assumes that structurally similar molecules are, with non-negligible probability, functionally similar as well. Any experts out there know to what degree this is true? >> The "thoughtful" above is art. That exceeds the purview of ARQl's intent. They need guidance from the lead fabricator to artfully introduce constraint parameters to stay within functionality or within liability. Perhaps, rather than 'combichem is bad (fair, whatever) science,' you really mean that it is bad art? Computers (a metaphor for automation in general) are lousy at composing music, but they can be put to use proving theorems, optimizing mechanical designs, and solving a wide range of problems, often using stochastic techniques. The process may not be a thing of beauty to look at (computer-generated theorem proofs can even be somewhat depressing in their methodical, tedious, thoroughness...), but that doesn't change the validity of the result. Why insist on art? If* the result is valid (a safe, effective drug), that's good enough for me. I really don't mind if it was discovered by a tribe of monkeys spinning a prayer wheel. *Admittedly, an important "if." PLS P.S. Hey, I'm dying of suspense. What do the theoreticians say about practicality?