About this Author
DBL%20Hendrix%20small.png College chemistry, 1983

Derek Lowe The 2002 Model

Dbl%20new%20portrait%20B%26W.png After 10 years of blogging. . .

Derek Lowe, an Arkansan by birth, got his BA from Hendrix College and his PhD in organic chemistry from Duke before spending time in Germany on a Humboldt Fellowship on his post-doc. He's worked for several major pharmaceutical companies since 1989 on drug discovery projects against schizophrenia, Alzheimer's, diabetes, osteoporosis and other diseases. To contact Derek email him directly: Twitter: Dereklowe

Chemistry and Drug Data: Drugbank
Chempedia Lab
Synthetic Pages
Organic Chemistry Portal
Not Voodoo

Chemistry and Pharma Blogs:
Org Prep Daily
The Haystack
A New Merck, Reviewed
Liberal Arts Chemistry
Electron Pusher
All Things Metathesis
C&E News Blogs
Chemiotics II
Chemical Space
Noel O'Blog
In Vivo Blog
Terra Sigilatta
BBSRC/Douglas Kell
Realizations in Biostatistics
ChemSpider Blog
Organic Chem - Education & Industry
Pharma Strategy Blog
No Name No Slogan
Practical Fragments
The Curious Wavefunction
Natural Product Man
Fragment Literature
Chemistry World Blog
Synthetic Nature
Chemistry Blog
Synthesizing Ideas
Eye on FDA
Chemical Forums
Symyx Blog
Sceptical Chymist
Lamentations on Chemistry
Computational Organic Chemistry
Mining Drugs
Henry Rzepa

Science Blogs and News:
Bad Science
The Loom
Uncertain Principles
Fierce Biotech
Blogs for Industry
Omics! Omics!
Young Female Scientist
Notional Slurry
Nobel Intent
SciTech Daily
Science Blog
Gene Expression (I)
Gene Expression (II)
Adventures in Ethics and Science
Transterrestrial Musings
Slashdot Science
Cosmic Variance
Biology News Net

Medical Blogs
DB's Medical Rants
Science-Based Medicine
Respectful Insolence
Diabetes Mine

Economics and Business
Marginal Revolution
The Volokh Conspiracy
Knowledge Problem

Politics / Current Events
Virginia Postrel
Belmont Club
Mickey Kaus

Belles Lettres
Uncouth Reflections
Arts and Letters Daily
In the Pipeline: Don't miss Derek Lowe's excellent commentary on drug discovery and the pharma industry in general at In the Pipeline

In the Pipeline

« The Industrial Diels Alder, Revisited in Detail | Main | Yuri Milner's Millions, And Where They're Going »

February 28, 2013

IBM's Watson Does Drug Discovery?

Email This Entry

Posted by Derek

I saw this story this morning, about IBM looking for more markets for its Watson information-sifting system (the one that performed so publicly on "Jeopardy". And this caught my eye for sure:

John Baldoni, senior vice president for technology and science at GlaxoSmithKline, got in touch with I.B.M. shortly after watching Watson’s “Jeopardy” triumph. He was struck that Watson frequently had the right answer, he said, “but what really impressed me was that it so quickly sifted out so many wrong answers.”

That is a huge challenge in drug discovery, which amounts to making a high-stakes bet, over years of testing, on the success of a chemical compound. The failure rate is high. Improving the odds, Mr. Baldoni said, could have a huge payoff economically and medically.

Glaxo and I.B.M. researchers put Watson through a test run. They fed it all the literature on malaria, known anti-malarial drugs and other chemical compounds. Watson correctly identified known anti-malarial drugs, and suggested 15 other compounds as potential drugs to combat malaria. The two companies are now discussing other projects.

“It doesn’t just answer questions, it encourages you to think more widely,” said Catherine E. Peishoff, vice president for computational and structural chemistry at Glaxo. “It essentially says, ‘Look over here, think about this.’ That’s one of the exciting things about this technology.”

Now, without seeing some structures and naming some names, it's completely impossible to say how valuable the Watson suggestions were. But I would very much like to know on what basis these other compounds were suggested: structural similarity? Mechanisms in common? Mechanisms that are in the same pathway, but hadn't been specifically looked at for malaria? Something else entirely? Unfortunately, we're probably not going to be able to find out, unless GSK is forthcoming with more details.

Eventually, there's coing to be another, somewhat more disturbing answer to that "what basis?" question. As this Slate article says, we could well get to the point where such systems make discoveries or correlations that are correct, but beyond our ability to figure out. Watson is most certainly not there yet. I don't think anything is, or is really all that close. But that doesn't mean it won't happen.

For a look at what this might be like, see Ted Chiang's story "Catching Crumbs From the Table", which appeared first in Nature, and then in his collection Stories of Your Life and Others, which I highly recommend, as "The Evolution of Human Science".

Comments (32) + TrackBacks (0) | Category: In Silico | Infectious Diseases


1. patentgeek on February 28, 2013 3:19 PM writes...

Eventually, AIs will progress to proposing de novo novel compounds for evaluation, and some will be successful. I hope I'm around for the gnarly questions on inventorship under US law that will occur when that happens. "...beyond our ability to figure out." Heh-heh.

Derek's recommendation re Ted Chiang is strongly recommended!! His "The Merchant and the Alchemist's Gate" is a modern-day Arabian Nights tour de force.

Permalink to Comment

2. josh bloom on February 28, 2013 3:47 PM writes...

I'll take "Lipinski Violators" for 400, Alex.

Permalink to Comment

3. oldtimer on February 28, 2013 4:10 PM writes...

Hmmm, teams of bright PhDs with tons of experience, and insight wade through a great deal of poorly validated (clinically) biological data and progress compounds to the clinic where for the most part they fail. What makes you think that Watson or any other expert system will do better? It is not the sifting of the data that is the issue, it is the quality of and validity of the data.

Permalink to Comment

4. Foolery on February 28, 2013 4:22 PM writes...

There is another Slate article here:

The program, Eureqa, has actually come up with an analysis of cellular behavior that holds true and is currently unexplainable.

Permalink to Comment

5. anon the II on February 28, 2013 4:23 PM writes...

This is a crock. First of all, if those pansies at IBM were real men, they would have made that stupid Watson parse the damn questions like the two humans had to do. The technology is certainly there for a camera to read the board, do the ICR and feed the answers into Watson's search engine. But Noooooo. They got the answer fed from a text file straight into Watson's feeble brain. So while Jennings and that other sap are trying to read the first character in the answer, Watson was already scanning the databanks and slamming the button. I was embarrassed for science to watch that crap and hear it called competition.

When the battle comes, I'm siding with the humans.

Permalink to Comment

6. gippgig on February 28, 2013 4:32 PM writes...

This would probably work reasonably well for things drugs all have in common, such as finding compounds with low toxicity, good pharmacokinetics, etc. Predicting compounds with totally new modes of action would probably require much more detailed information on how cells function at the molecular level then is available. Maybe in a few decades...

Permalink to Comment

7. Curious Wavefunction on February 28, 2013 4:41 PM writes...

Watson: I am sorry Dave, I cannot let you synthesize that compound.

Permalink to Comment

8. Curious Wavefunction on February 28, 2013 4:42 PM writes...

Watson: I am sorry Dave, I cannot let you synthesize that compound.

Permalink to Comment

9. Mad Dog on February 28, 2013 4:54 PM writes...

@7: Are you implying Watson is one of the Pfizer "designers"?

Permalink to Comment

10. darwinsdog on February 28, 2013 4:56 PM writes...

The day a computer does anything more than tell the programmer the same thing they could have calculated themselves but just in a faster manner is the day I...hey wait...uh, Google what do you mean 'self aware'..I didn't ask...easy now...[EXTERMINATE !, EXTERMINATE !]

Permalink to Comment

11. metaphysician on February 28, 2013 5:10 PM writes...

It may not be yet, but it is a legitimate matter of discussion for the future. I would say that it should only qualify as a "designer" for the purposes of patents if it also qualifies as a *person*, and is thus as much an employee of the company as any human. Of course, that is its own giant massive ball of wax. . .

Permalink to Comment

12. josh bloom on February 28, 2013 5:10 PM writes...

@Curious-- VERY funny

Permalink to Comment

13. matt on February 28, 2013 5:28 PM writes...

Yes, but can Watson avoid embarrassing gaffes or prove that overexposure to antioxidants is contributing to cancer rates?

Permalink to Comment

14. ex-GSK on February 28, 2013 5:34 PM writes...

Bet Watson could shift through the Sirtris data and suggest the company was worth somewhat less than the $750M GSK spent for them.... But then, so did the due diligence team at GSK. You have to actually base decisions on data to care.

Permalink to Comment

15. Anonymous on February 28, 2013 6:32 PM writes...

Well if John Baldoni thinks it's worth investigating, it must be good!

Permalink to Comment

16. hibob on February 28, 2013 7:18 PM writes...

We've " been there" ever since genetic algorithms were introduced: completely impenetrable code and decision making. The jump is when the questions getting correctly answered are ones we have no understandable way of solving, as opposed to being merely difficult.

Permalink to Comment

17. Insilicoconsulting on February 28, 2013 10:12 PM writes...

Look upon it as a very efficient form of search, given the right parameters(human experts), the right data and background knowledge .

It may outperform human intuition in terms of lesser false positives and negatives, but we would still be the ones defining the context and prior knowledge fed to the beast.

Permalink to Comment

18. Anonymous on March 1, 2013 4:58 AM writes...

John Baldoni.......! A lot of people @GSK are rolling around the floor laughing! Perhaps the Great Baldoni will invest in a Watson and sack a few scientists to make room for it.

Permalink to Comment

19. John Spevacek on March 1, 2013 8:55 AM writes...

I'm struck by the fact that this may soon create "obvious" solutions for the pharma industry, meaning no more patents - "It's obvious to ask Watson and then try what he suggests." After all, this is well within the capabilities of the iconic PHOSITA (Person Having Ordinary Skill In The Art)

Permalink to Comment

20. TheGreatBaldoni on March 1, 2013 9:28 AM writes...

Would the real Baldoni please stand up?
Is it this one?
or this one?

Permalink to Comment

21. anon on March 1, 2013 10:25 AM writes...

would all the junk research out there be included too? and would Watson know to ignore it?

Permalink to Comment

22. Woos? on March 1, 2013 10:29 AM writes...

Let Watson seat at any GSK portfolio review meeting and he will perform like everybody else in term of probability of success.

Permalink to Comment

23. watcher on March 1, 2013 10:57 AM writes...

My bet is the Watson found what any good chemist and biologist team would do when looking under the lamp post. Can it actually move the vision? Unlikely.

Permalink to Comment

24. simpl on March 1, 2013 11:06 AM writes...

agree with 16
We saw this with visual control of vial production. There was no way the first generation of machines could match human skill. But over thirty years, the makers have made enough progress that the job is now better done by machine.

Permalink to Comment

25. jimrandomh on March 1, 2013 12:37 PM writes...

My guess is that the 15 candidate drugs it suggested were, in fact, mentioned as candidates by papers in the dataset. That is, in fact, a pretty good way of finding candidate drugs (as long as no one else got there first), but it'd be a case of Watson acting like the fancy search engine that it is, and not like the AI its PR claims.

Permalink to Comment

26. Luis Pedro Coelho on March 1, 2013 1:29 PM writes...

Watson is probably well into the black box mode.

I think we need to start to become comfortable with the fact that computers can have amazing intuitions: correct inferences that are impossible to rationalize.

Permalink to Comment

27. MoMo on March 1, 2013 3:18 PM writes...

Not impressed. They fed the scientific literature, primarily chemical and biological abstracts, and came up with only 15 new anti-malarials? One can do the same with Scifinder and use Tanimoto similarity searches to do the same thing. In about 15 minutes and using a laptop.

Good waste of time, Watson.

Permalink to Comment

28. DCRogers on March 1, 2013 4:03 PM writes...

Beware of "uninterpretable" AI models.

I'm reminded of a story (perhaps apocryphal) of a university that wanted a filter for the mountains of applicants. They used Machine Learning to build a complex AI model that did as good a job sifting the candidates as did their human experts, in a fraction of the time.

Punch line: they got sued for sex discrimination. They said, hey, not possible, we're using AI! Belatedly looking into the complex model, they found to their chagrin logic something like:


Permalink to Comment

29. metaphysician on March 2, 2013 10:05 AM writes...


Of course, that opens some very interesting questions: what if an objective, unbiased AI *does* turn up logic like that in its statistical assessments? Is it discrimination if it can be actuarially supported?

Permalink to Comment

30. hibob on March 2, 2013 12:08 PM writes...

"Is it discrimination if it can be actuarially supported?"

It could still be discrimination if underperformance by women at the university was due to actions by that university as opposed to the nature of the applicants/students.

Legally allowed? Sex falls under intermediate scrutiny, but the Supreme Court normally puts it at the end of intermediate scrutiny that's closest to strict scrutiny.
Probably not.

Permalink to Comment

31. A. Postdoc on March 4, 2013 11:33 AM writes...

A total crock. Science by press release is awful. Drawing attention to this diminishes all the efforts of people doing real virtual screening with actual followup testing. Derek, I think you should try to highlight some of those.

Permalink to Comment

32. MIMD on March 5, 2013 1:06 PM writes...


"To Save Everything, Click Here: The Folly of Technological Solutionism"

Permalink to Comment


Remember Me?


Email this entry to:

Your email address:

Message (optional):

The Last Post
The GSK Layoffs Continue, By Proxy
The Move is Nigh
Another Alzheimer's IPO
Cutbacks at C&E News
Sanofi Pays to Get Back Into Oncology
An Irresponsible Statement About Curing Cancer
Oliver Sacks on Turning Back to Chemistry