Silly Little Country

Derek Lowe Rants

And rightly.

We did it, naturally, because we expected to make a profit out of it in the end. The whole PPAR story looked like a great way to affect metabolic disorders and plenty of other diseases as well: cancer, inflammation, cardiovascular. That is, if we could just manage to understand what was going on. But we didn't. Once we all figured out that nuclear receptors were involved and got busy on drug discovery on that basis, we didn't help anyone with any diseases, and we didn't make any profits. Big piles of money actually disappeared during the process, never to be seen again. You could ask Merck about that, or GSK (post-rosiglitazone), or Lilly, or BMS, or Bayer, and plenty of other players large and small.
No one hears about these things. We're understandably reluctant to go on about our failures in this industry, but the side effect is that people who aren't paying attention end up thinking that we don't have any. Nothing could be more mistaken. And they aren't failures to come up with a catchy slogan or to find a good color scheme for the packaging - they're failures back at the actual science, where reality meets our ideas about it, and likely as not beats them down to the floor.
Honestly, I don't understand where these they-don't-do-any-research folks get off. Look at the patent filings. Look at the open literature. Where on earth do you think all those molecules come from, all those research programs to fill up all those servers? There are whole scientific journals that wouldn't exist if it weren't for a steady stream of failed research projects. Where's it all coming from?

Of course there are those who think Universities and Government Labs create all that initial research and I do not believe it for a second. In the industry I wwrked in, software development, the academic 'research' teams were ludicrously behind the private development teams.
Most hilariously, I recall a keynote address presented at a workshop by an academic who had just figured out that there is quite a bit of nondeterminism in Java performance results, and her treatment was to deal with them statistically.
Of course, the company I worked for had figured this out many years ago and had been doing that sort of analysis for years. I would be deeply unsurprised if any of the other companies in the business were behaving otherwise. This was a revelation only to academics for the obvious reason that the right outcome actually mattered to the companies, and in no ways to acacemics.
The deepest irony is that this was presented largely to a bunch of corporate Java performance analysts, who of course said not a single word. (But likely shook their heads in wonder.)
I doubt it is any way different in the drug industry. I will wager the development guys know WAY more than the profs.
Just a bet but I am happy to make it real with someone who wants to differ.

posted by Alan Adamson @ 3:28 PM