« Here and There |
Main
| A Sensitive Guy »
September 15, 2005
Pretty Much the Reason You'd Think
Posted by Derek
I wrote the other day about having a hypothesis in mind when you make new drug analogs (as opposed to just trying a few to see what happens.) A colleague of mine and I were talking about this, and he offered a suggestion about why some people are much more "by the book" than others when it comes to running an analoging program. It's all, he said, about covering your anatomy.
And I think he's right. If you confine yourself to just the sorts of structures that people have had success with before, no one can question your selection. And it's also true that people who insist on a hypothesis behind every new analog also have a strong inclination toward the hypotheses that are already thought to be important (or at least fashionable.) If upper management gets convinced that the smell of your compounds is a key component for clinical success, you can expect these folks to make sure that everything gets made with the sniff test in mind.
So if you're one of these people and your project runs into the ditch - most do, you know - you're still going to be OK. You did everything the way that everyone else thinks that it should be done, and you have concrete reasons to point to if anyone asks you why a particular compound was made. It's an insurance policy.
But if you have some weird ones in there, though, or tried some things that aren't in the usual playbook, the tendency will be to blame those for anything that went wrong. Well, actually, people will blame the person who thought up or OKed these things in the first place - you. I'm not arguing that projects switch over to complete wild-blue-yonder mode, not any more than I think a pot of soup needs a pound of pepper in it. But for seasoning, I think that a drug discovery program needs a small (but real) effort to make some things just because no one's made them. These are the kinds of ideas that go on to breed new projects themselves.
I once came across an analogous effect in the financial world. An article I read about the money management industry pointed out that many fund managers tend to cluster together for safety. If everyone's going up and down roughly in tandem, there's not as much room for irate customers who demand to know why you aren't keeping up. If you buy a million dollars of IBM and it goes down, the author pointed out, people will ask "What's wrong with that IBM?" But if you buy a millions dollars worth of Zarkotronics and it tanks, people will ask "What's wrong with you?"
Comments (19)
+ TrackBacks (0) | Category: Who Discovers and Why
POST A COMMENT
- RELATED ENTRIES
- › Shire's Replagal Problems: An Inside Look?
- › Bungled Structure, And How
- › Roche Closes Nutley, Once Its US R&D Home
- › The Next Five Years in the Drug Industry
- › A Kinase Inhibitor Learns Something New
- › The Good Ol' Friedel-Crafts
- › Merck's Madagascar Marketing Mess
- › Scientific Literacy: Where Do You Stop?
College chemistry, 1983
The 2002 Model
After 10 years of blogging. . .
1. daen on September 16, 2005 3:29 AM writes...
I came across a game theory puzzle a few days ago. Pick a number between 0 and 100. The one who gets closest to 2/3 of the average wins a prize. What should be your strategy in picking the number? The answer depends very much on who you're playing against.
Permalink to Comment2. Petros on September 16, 2005 4:52 AM writes...
Your comments are very apt, and perhaps highlight the need for companies to embrace medicinal chemists with both philosophies.
The best example I am aware of is of nifedipine (Adalat) which has made billions for Bayer and Pfizer over the years.
It was synthesized by a chemist in Wuppertal but was of a structural type not approved by management. This chemist apparently built up a significant collection of such compounds which he submitted for testing just as he was about to retire!
A few years later a goodly portion of Bayer's medicinal chemistry capacity was devoted to making related dihydropyridine structures.
Permalink to Comment3. rob cyran on September 16, 2005 6:20 AM writes...
Excellent point. If anything you underestimate the pernicious effects of clustering for safety. To use your finance example, consider what happened during the dotcom boom. lots of financial managers realised the valuations of companies made no sense. But they bought anyway, because it could take several years for rationality to reassert itself and their bonuses depended on quick results.
The effects of clustering is also pretty clear in academia. There are lots of PHD candidates who choose to make slight tweaks to existing theory. In economics, I actually heard people say they knew an underlying theory was rubish, but since so many other people have used it, a little tweak upon it would result in a dissertation!
Permalink to Commentissue of clustering is also obvious in academia. The economics department I studied in was full of people who had made slight tweaks to well established models.
4. Kay on September 16, 2005 7:05 AM writes...
You are zeroing in nicely on a strong probable cause for the demise of big pharma. Big pharma has had the money to take giant risks (they were paid to do so via an unwritten agreement with the public), but they mostly did the safe stuff. Rough sailing ahead.
Permalink to Comment5. Don on September 16, 2005 8:40 AM writes...
Funny thing, The IBM example was the first one that I thought of upon reading your acccount of the risk aversion of target selection. I think the old broker's slogan was "No one ever got fired picking IBM." For a medicinal chemist the anlagous saying might be, "No project leader was ever terminated for making a 4-fluorophenyl."
Permalink to Comment6. Tom Bartlett on September 16, 2005 8:58 AM writes...
Don is right on here, though the modern equivalent are "CYA" IT drones that force MSFT on everybody even though Chemdraw, Office, and just about anything else that's any good is also on Mac.
Kay also makes a good point. I have done Big Pharma and Tiny Pharma, and you get more bang for the buck innovation-wise in start-ups. Of course, you have WAY less bucks, so I wouldn't sell my GSK/PFE/BMS stock.....
Permalink to Comment7. Derek Lowe on September 16, 2005 9:07 AM writes...
Nope, you'll never get any flak for 4-fluorophenyl, or sticking on a morpholine instead of a dimethylamino, or switching phenyl to thiophene. And there's nothing wrong with any of those, of course. It's just that if that's all you do, you can hope to be pleasantly surprised, but you'll never be amazed.
Permalink to Comment8. jim on September 16, 2005 11:27 AM writes...
Another thing that this mindset can affect in patentability, don't you think? As things stay with the cookbook approach, the odds that someone else is going to try the same thing drastically increase. You might discover a good compound, but it may not be yours.
Permalink to Comment9. Sigivald on September 16, 2005 1:48 PM writes...
Don: That statement is more traditionally applied to people buying computer systems (especially in the mainframe days).
Nobody ever got fired for buying IBM. I suspect some people did get fired for buying, say, Mexican clones of a VAX.
Permalink to Comment10. Tom Bartlett on September 16, 2005 1:55 PM writes...
" As things stay with the cookbook approach, the odds that someone else is going to try the same thing drastically increase. "
Disagree. If you start with a novel template, it doesn't matter whether you do methyl, ethyl, propyl or not; the novelty comes from the base structure.
Permalink to Comment11. Ramesh on September 16, 2005 1:57 PM writes...
To echo the earleir comment about economists clustering together, I had the misfortune of going after something off the beaten path both methodologically and substantively. Not only was it harder to find committee members and longer to finish, very few people really understood it; as a result, attempts to get it published failed.
Yet, several years later, I see similar approaches start to develop. I am not a practicing economist any more (my analytics help sell legal drugs), and therefore am not so close to the literature, but, nonetheless, will wager that nonlinear and combinatorial methods will become more prevalent in the economists' arsenal.
Permalink to Comment12. Michael T on September 16, 2005 2:20 PM writes...
In a similar vein, I have always thought the hiring practices of 'big pharma' were motivated, at least in part, by anatomy covering. For example, if a hire turns out to be a bad one, one can say, "how was I to know? They were from Corey's group!" One cannot say that if the hire was from no-name's group.
Permalink to Comment13. jim on September 16, 2005 2:37 PM writes...
Tom. You're right in that you're OK if your template is novel, but what if your best or only lead is from Chembridge or some other commercial library? I know of one instance in particular where 3 companies were pursuing the same lead series simultaneously based on such a hit. It wasn't until 3-4 years later that the truly novel chemistry started to come in to play.
Permalink to Comment14. SRC on September 16, 2005 3:01 PM writes...
Ask Jeff Vinik about that.
Permalink to Comment15. Linkmeister on September 16, 2005 4:11 PM writes...
Don, that slogan you cite about "picking IBM" may have been a broker's, but it applied equally well to IT/DP people. If you were looking for new computer hardware, trying to persuade management that something other than IBM was a good supplier was like tooth extraction.
Permalink to Comment16. Tom Bartlett on September 16, 2005 4:25 PM writes...
" but what if your best or only lead is from Chembridge or some other commercial library? "
yea. Been there; done that. Pharmacia vs Astra-Zeneca on IKK, etc.
Permalink to Comment17. Joe on September 16, 2005 8:00 PM writes...
Drug development seems more expensive, and riskier than ever, YET
we keep hearing its the age of biology and that computational tools, automation, genome sequencing, etc etc were supposed to transform this industry.
It doesn't add up. Why?
Permalink to Comment18. Psychic Chemist on September 16, 2005 9:56 PM writes...
Asking the question "what is your hypothesis" makes sure that lazy people cant hide behind "the monomer was commercially available" (meaning, I am too lazy to do anything else) excuse for more than a couple of years - or maybe longer depending on where you are.
Permalink to Comment19. weirdo on September 20, 2005 2:55 PM writes...
"YET we keep hearing its the age of biology and that computational tools, automation, genome sequencing, etc etc were supposed to transform this industry. It doesn't add up. Why?"
You don't hear that much from Big Pharma, do you? You hear it from the dink companies trying to sell those technologies. Which we have all tried. They only get you to the hard parts (5-day tox, 4-week tox, Phase I) a little faster, and not necessarily with a better molecule. I think Derek has blogged on this topic a number of times before . . . .
Permalink to Comment