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

« A Blogroll Update | Main | I'll Get Right On That For You, Professor »

November 17, 2009

Side Effects, Predicted?

Email This Entry

Posted by Derek

There's a new paper out in Nature that presents an intriguing way to look for off-target effects of drug candidates. The authors (a large multi-center team) looked at a large number of known drugs (or well-characterized clinical candidates) and their activity profiles. They then characterized the protein targets by the similarities of the molecules that were known to bind to them.

That gave a large number of possible combinations - nearly a million, actually, and in most cases, no correlations showed up. But in about 7,000 examples, a drug matched some other ligand set to an interesting degree. On closer inspection, some of these off-target effects turned out to be already known (but had not been picked up during their initial searching using the MDDR database). Many others turned out to be trivial variations on other known structures.

But what was left over was a set of 3,832 predictions of meaningful off-target binding events. The authors took 184 of these out to review them carefully and see how well they held up. 42 of these turned out to be already confirmed in the primary literature, although not reported in any of the databases they'd used to construct the system - that result alone is enough to make one think that they might be on the right track here.

Of the remaining 142 correlations, 30 were experimentally feasible to check directly. Of these, 23 came back with inhibition constants less than 15 micromolar - not incredibly potent, but something to think about, and a lot better hit rate than one would expect by chance. Some of the hits were quite striking - for example, an old alpha-blocker, indoramin, showed a strong association for dopamine receptors, and turned out to be an 18 nM ligand for D4, which is better than it does on the alpha receptors themselves. In general, they uncovered a lot of new GPCR activities for older CNS drugs, which doesn't surprise me, given the polypharmacy that's often seen in that area.

But they found four examples of compounds that jumped into completely new target categories. Rescriptor (delavirdine), a reverse transcriptase inhibitor used against HIV, showed a strong score against histamine subtypes, and turned out to bind H4 at about five micromolar. That may not sound like much, but the drug's blood levels make that a realistic level to think about, and its side effects include a skin rash that's just what you might expect from such off-target binding.

There are some limitations. To their credit, the authors mention in detail a number of false positives that their method generated - equally compelling predictions of activities that just aren't there. This doesn't surprise me much - compounds can look quite similar to existing classes and not share their activity. I'm actually a bit surprised that their methods works as well as it does, and look forward to seeing refined versions of it.

To my mind, this would be an effort well worth some collaborative support by all the large drug companies. A better off-target prediction tool would be worth a great deal to the whole industry, and we might be able to provide a lot more useful data to refine the models used. Anyone want to step up?

Update: be sure to check out the comments section for other examples in this field, and a lively debate about which methods might work best. . .

Comments (20) + TrackBacks (0) | Category: Drug Assays | In Silico | Toxicology


1. HappyDog on November 17, 2009 1:27 PM writes...

One of my annoyances is the anti-industry bias regularly shown by the editorial staff of Science and Nature. Another one is the repeated omission of prior art by well-known researchers, and this is one example. Studies of off-target effects (both beneficial as well as potential adverse events) have been studies in the industry for years. For example, take the work of Jenkins, Glick, and Davies at Novartis. They have reported on methods to search for off-terget effects in the literature since at least 2004.

Permalink to Comment

2. alig on November 17, 2009 1:43 PM writes...

Seems like an awful lot of work to find an off-target binding event when all we would care about is off-target pharmacology, which tox studies are a better and faster method to discover relevant pharmacological effects.

Permalink to Comment

3. SPRITY on November 17, 2009 3:35 PM writes...

The result is not surprising at all. One should not expect highly selective small molecule drugs. If there are a few more kinase inhibitors in the market, there should be a lot more off target predictions. The most incredicle thing is that the author and reviewer actually believe that "The chemical similarity approach is systematic and comprehensive, and may suggest side-effects and new indications for many drugs." If 2D structural similarity can really represent the dynamic 3D world, all the medicinal chemistry should be in the unemployment line now.

Permalink to Comment

4. Sili on November 17, 2009 7:02 PM writes...

I do realise that it's hard enough as it is to get good activity for the desired target, but I'd love to see the day where something like this is used to deliberate reduce an inconvenient sideeffect.

Permalink to Comment

5. cliffintokyo on November 17, 2009 7:36 PM writes...

This is a really good informative post; comment on previous lit enhances the usefulness.
#2 Once the software is up and running, can't see that it would be so much effort to plug in the structure and read the print out.
#3 What's wrong with having some additional info about what to be on the lookout for? Since when has tox been so good at predicting human side effects anyway? Perhaps its *not my job*. Get out of your silo and get real!

Permalink to Comment

6. Anonymous on November 17, 2009 9:56 PM writes...

they authors tested some molecules not only for convenience (one group is specializing on GPCR), but also for easiness----- most med chemists could easily deal with gpcr activities, which is just easy, OK?

see the results for non-gpcrs, micro-molar? so what? without those gpcr data(nM, magic word:-), I bet they could not even think of NATURE.

drug discovery with 2D similarity---- we all must be nuts, otherwise we could not lose jobs :-)

Permalink to Comment

7. Anonymous on November 17, 2009 10:04 PM writes...

1. they should list a table to let the readers know how similar the molecules are.

2. target sequence similarity mentioned in the paper is too naive: one should really look into the ACTIVE-SITE STRUCTURAL similarity?

3. the results may be impressive, but are also misleading. If a paper like this could be called high-profile, it would make damages for the field. Fortunately most experienced professionals in industry are not as naive as those academic folks, since they already tried everything known and available, and the ideas in this paper is at least 50 years old...

Permalink to Comment

8. Anonymous on November 17, 2009 10:54 PM writes...

And how about this one?
Nature Chemical Biology 1, 389 - 397 (2005)
Analysis of drug-induced effect patterns to link structure and side effects of medicines

Anton F Fliri, William T Loging, Peter F Thadeio & Robert A Volkmann

Permalink to Comment

9. Aspirin on November 17, 2009 11:35 PM writes...

Anon, 2D similarity methods work better than you think

Permalink to Comment

10. Evorich on November 18, 2009 7:38 AM writes...

It would have been interesting if they'd cross referenced with other studies - particularly ones done by different methods, e.g. , which used genomic expression profiles to try to get at these relationships.

All is good info to suggest other activities for current drugs - not that the industry needs more trouble on that front!

Permalink to Comment

11. Jose on November 18, 2009 7:38 AM writes...

Color me underwhelmed. As in statistics, given a big enough fishing net, something is bound to be "significant" and interesting by dumb luck alone. Is there a Bonferroni for binding values?

Permalink to Comment

12. BACE on November 18, 2009 7:58 AM writes...

Jose, it's not by dumb luck. They calculate E values which indicate whether the correlation is by chance, and for all their cases the E values are small enough to indicate that this is not so.

Permalink to Comment

13. CASE on November 18, 2009 8:10 AM writes...

BACE, you would be as realistic as me if they list the E-values of all false-positives and false-negatives... hallelujah

Permalink to Comment

14. Jose on November 18, 2009 8:39 AM writes...

OK, ok, so perhaps I need to read the paper in detail before I blather....

Permalink to Comment

15. SRPITY on November 18, 2009 10:41 AM writes...

Don't be fooled by Gaussian probability which does not work in real life. The dice in real life is rigged. The outcomes do not have equal opportunity. All small molecule drugs have cross activity. Drug molecules with less enthalpic component in the binding free energy would be less specific. It does not take a lot of dumb luck to find small molecule drugs with off target activity. The big question is whether the predicting method makes sense. In you fills a bucket with a hundred fishes, a fish will jump into your hands if you wait long enough. Does it mean standing with open arms a good way to fish?

Permalink to Comment

16. David P on November 18, 2009 12:08 PM writes...

Not a huge surprise that drugs bind to things they weren't designed to, though whether they have any effect on those things, especially in the body with protein binding and everything else going on as well is surely questionable.

Gives us a few more things to think about though (as if med chem did not have enough of those!) and might give some good pointers to things to look for early on. I hope compounds are not knocked out by a bad binding result though. I've seen plenty of compounds that showed binding to this enzyme or whatever that then showed no in vivo effect.

A useful tool, but not a replacement for actual experiments in vivo.

Permalink to Comment

17. Jose on November 19, 2009 5:10 AM writes...

I'm not seeing the leap from (flat) Tanimoto coefficients to similarity ensembles of (flat?) ligands to BLAST E-values, which apparently measures (linear) sequence similarity? If SAR is that easy to model with topological schemes, why do in silico approaches yield such atrocious S/N?

Permalink to Comment

18. ex-Pfizerite on November 19, 2009 9:05 PM writes...

What I want is a way to correlate toxicology in the (whole) dog and rat with the observed clinical side effects in man so that I can disregard idiosyncratic effects in animals that are not applicable in man and also know whether or not I will have a problem in man that does not show up in my animal tox data. The last requirement should be credited to a VP in development.

Permalink to Comment

19. TruthSeeker on November 25, 2009 5:24 PM writes...

Fliri et al. have a simular paper in early online publication at J. Med Chem. "Drug Effects Viewed from a Signal Transduction Network Perspective".
Dr. Volkmann first presented on this approach back in 2006.

Permalink to Comment

20. LeeH on November 25, 2009 7:42 PM writes...

The Shoichet paper is not surprising at all. Similar molecules tend to have similar activities. It's the principle that drives all drug discovery, whether you know (or want to admit) it or not.

FYI, if you have Pipeline Pilot you can download a model object from their user contributed examples which computes the probability of a molecule belonging to each of the pharmacological classes in the MDDR. Better than straight similarity, since the probabilities are actually calculated from multiple Bayesian classifier models. Another example of how this sort of approach has been done before, albeit without any evidence that it actually works.

Permalink to Comment


Remember Me?


Email this entry to:

Your email address:

Message (optional):

Gitcher SF5 Groups Right Here
Changing A Broken Science System
One and Done
The Latest Protein-Protein Compounds
Professor Fukuyama's Solvent Peaks
Novartis Gets Out of RNAi
Total Synthesis in Flow
Sweet Reason Lands On Its Face