On Wed, Jan 10, 2007 at 07:55:16PM +0530, Aditya Kapil wrote: > Late stage failure is due more to companies pursuing commercially safe, > 'old' targets that they have worked with for decades. Regulatory authorities
Approval process has become much longer, and costlier. Risk of litigation after approval has gone way up. Odds for each new molecule -- regardless of its origins, there are no safe molecules, whether small or large, just think cytokine storm -- are getting worse. Financially, this means the current process will result in less drugs to markets, and elephant weddings, which would result in effective monopolies. This is not good for the market, and not good for the people. Some serious innovation is in order to build up a new process which brings up drug output rate back to a sustainable level. > in the same time have become more stringent. So a side effect profile > acceptable in the 70's is not now. I'd expect the newer technologies such as > RDD, IT-based pathway analysis, systems biology etc to make things better. Sure, but they're not better, yet. It might take a long time, and you have to get all parts of the in machina process just right. This isn't easy, and if you're just looking at something as old and stuffy as a Lipinski druglike/leadlike screen of an existing empirical database. What could be easier, right? Until you look at ClogP algorithms, which are mostly proprietary, and realize they're complete shit. > In combination with conventional techniques of course. Surely this iterative > combo is better than 'hit and miss' methods involving plain old screening. The point is that the machine process is setting you up for a list of duds which might even look good in the screens. I presume many naive heads rolled when they found it out the hard way. > And that too of chemicals that may not 'fit' newer targets at all. If you don't know what the structure is, and are just looking for activity (let's say you've got a small but killer library full of natural substances -- not just huge dumb combinatorial soup which only looks good on paper) you can figure out what works, and what not. Without even knowing what your target is. > I agree to an extent. But the backlash is also due to VC's not understanding > what they were investing in. The yr 2000 euphoria in biotech was a knee-jerk Regardless of the reasons, the result is a profound chill. Not unlike the AI winter, which still hasn't ended. > reaction to the dotcom bust. The dotcom bust wasn't just a dotcom bust, but a minor (the majority of which is yet to come, probably starting as early as this year) readjustment of overvalued stock *across the board*. The reason why pharma got disillusioned with computational methods is that they suck. Not in theory, in theory they're just great. In practice. -- Eugen* Leitl <a href="http://leitl.org">leitl</a> http://leitl.org ______________________________________________________________ ICBM: 48.07100, 11.36820 http://www.ativel.com 8B29F6BE: 099D 78BA 2FD3 B014 B08A 7779 75B0 2443 8B29 F6BE
signature.asc
Description: Digital signature