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 Old Days in Organic Chemistry | Main | Of Mice (Studies) and Men »

November 25, 2013

Lipinski's Anchor

Email This Entry

Posted by Derek

Michael Shultz of Novartis is back with more thoughts on how we assign numbers to drug candidates. Previously, he's written about the mathematical wrongness of many of the favorite metrics (such as ligand efficiency), in a paper that stirred up plenty of comment.

His new piece in ACS Medicinal Chemistry Letters is well worth a look, although I confess that (for me) it seemed to end just when it was getting started. But that's the limitation of a Viewpoint article for a subject with this much detail in it.

Shultz makes some very good points by referring to Daniel Kahneman's Thinking, Fast and Slow, a book that's come up several times around here as well (in both posts and comments). The key concept here is called "attribute substitution", which is the mental process by which we take a complex situation, which we find mentally unworkable, and try to substitute some other scheme which we can deal with. We then convince ourselves, often quickly, silently, and without realizing that we're doing it, that we now have a handle on the situation, just because we now have something in our heads that is more understandable. That "Ah, now I get it" feeling is often a sign that you're making headway on some tough subject, but you can also get it when you're understanding something that doesn't help you with it at all.

And I'd say that this is the take-home for this whole Viewpoint article, that we medicinal chemists are fooling ourselves when we use ligand efficiency and similar metrics to try to understand what's going on with our drug candidates. Shultz go on to discuss what he calls "Lipinski's Anchor". Anchoring is another concept out of Thinking Fast and Slow, and here's the application:

The authors of the ‘rules of 5’ were keenly aware of their target audience (medicinal chemists) and “deliberately excluded equations and regression the expense of a loss of detail.” One of the greatest misinterpretations of this paper was that these alerts were for drug-likeness. The authors examined the World Drug Index (WDI) and applied several filters to identify 2245 drugs that had at least entered phase II clinical development. Applying a roughly 90% cutoff for property distribution, the authors identified four parameters (MW, logP, hydrogen bond donors, and hydrogen bond acceptors) that were hypothesized to influence solubility and permeability based on their difference from the remainder of the WDI. When judging probability, people rely on representativeness heuristics (a description that sounds highly plausible), while base-rate frequency is often ignored. When proposing oral drug-like properties, the Gaussian distribution of properties was believed, de facto, to represent the ability to achieve oral bioavailability. An anchoring effect is when a number is considered before estimating an unknown value and the original number significantly influences future estimates. When a simple, specific, and plausible MW of 500 was given as cutoff for oral drugs, this became the mother of all medicinal chemistry anchors.

But how valid are molecular weight cutoffs, anyway? That's a topic that's come up around here a few times, too, as well it should. Comparisons of the properties of orally available drugs across their various stages of development seem to suggest that such measurements converge on what we feel are the "right" values, but as Shultz points out, there could be other reasons for the data to look that way. And he makes this recommendation: "Since the average MW of approved oral drugs has been increasing while the failure rate due to PK/biovailability has been decreasing, the hypothesis linking size and bioavailability should be reconsidered."

I particularly like another line, which could probably serve as the take-home message for the whole piece: "A clear understanding of probabilities in drug discovery is impossible due to the large number of known and unknown variables." I agree. And I think that's the root of the problem, because a lot of people are very, very uncomfortable with that kind of talk. The more business-school training they have, the less they like the sound of it. The feeling is that if we'd just use modern management techniques, it wouldn't have to be this way. Closer to the science end of things, the feeling is that if we'd just apply the right metrics to our work, it wouldn't have to be that way, either. Are both of these mindsets just examples of attribute substitution at work?

In the past, I've said many times that if I had to work from a million compounds that were within rule-of-five cutoffs versus a million that weren't, I'd go for the former every time. And I'm still not ready to ditch that bias, but I'm certainly ready to start running up the Jolly Roger about things like molecular weight. I still think that the clinical failure rate is higher for significantly greasier compounds (both because of PK issues and because of unexpected tox). But molecular weight might not be much of a proxy for the things we care about.

This post is long enough already, so I'll address Shultz's latest thoughts on ligand efficiency in another entry. For those who want more 50,000-foot viewpoints on these issues, though, these older posts will have plenty.

Comments (44) + TrackBacks (0) | Category: Drug Development | Drug Industry History


1. simpl on November 25, 2013 10:55 AM writes...

Same thing with musical composition, a complex ill-defined set of rules about defining what is good. The theorists then say that if you are a good practitioner, you know when you can break the rules.So they award extra points for getting away with it.
By analogy, if you normally get a bottle of champagne for each compound that makes it into clinical, you should get two if your compound breaks the Lipinski rules.

Permalink to Comment

2. anon on November 25, 2013 10:58 AM writes...

Its an interesting paper.

Bottom line is that the observation:parameter ratio for what makes a good oral drug is so poor that any attempt to rationalise it will involve so many approximations and assumptions that they will not be statistically sound.

What do the current suite of metrics give us - they give us a set of checks that can be used to match the small molecule discovery process to current thinking.

So many of the metrics du jour align with the smaller less lipohiliic approach that is prominent at the moment.

As the fragment community are starting to demonise rule of any nature we should call them guides as opposed to rules.

The worst thing is the new over-paramerisation approach where compounds are being judged on how they map to a set of approximations, which upon failure means they wont even get assayed.

I have seen multiple programs where chemists are optimising metrics but have minor appreciation for the biologic rational of the target.

Permalink to Comment

3. John Wayne on November 25, 2013 11:07 AM writes...

The observed metric will improve, especially if that metric is simple enough for the average MBA to understand. I don't usually make sweeping generalizations about other people's professions, but it has been a hard quarter for objectivity.

Permalink to Comment

4. Anonymous on November 25, 2013 11:18 AM writes...

Great article. I'm a big fan of rules and metrics ... Especially testing and breaking them! It's about time we brought more of Kahneman's thinking on decision making into pharma, it's exactly what we need to break through the destructive force of hubris which has plagued the industry, and virtually killed it.

Permalink to Comment

5. frag_boy on November 25, 2013 11:26 AM writes...

My favourite thing is when people present data for a series of compounds in an XY plot with Ligand efficiency vs potency and discuss the correlation ...........

Permalink to Comment

6. Rock on November 25, 2013 11:30 AM writes...

It is interesting how Lipinski, once lauded, is now demonized with some frequency. Keep in mind that the origins of his "rules" started in the early 90's when access to high throughput permeability assays was virtually non-existent (yes, the rules were focused on absorption). And also don't forget that the rule is: you can break one rule. Given our current tools, all skilled medicinal chemists should have a very good understanding where high probability space is for their compounds for any particular parameter.
Lastly, people often discuss LE and LLE (my personal favorite) as an inherent property of a compound. In my view, they are best used to compare compounds during optimization (i.e. was that change in structure worth the added MW or LogP?). Sometimes the answer to the question is yes, even if it sends the metric in the wrong direction if it solves a problem. But all things being equal, I will take a compound with an LLE of 6 over one of 4 any day.

Permalink to Comment

7. Bunsen Honeydew on November 25, 2013 11:33 AM writes...

In The Pipeline Quote of the Month: "I'm certainly ready to start running up the Jolly Roger about things like molecular weight." Brilliant, Derek, just bloody brilliant!

Permalink to Comment

8. petros on November 25, 2013 11:58 AM writes...

As Derek comments the article is too brief. But did the Rule of 5 not develop as a consequence of the large amount of garbage that was being generated, and screened, as a consequence of the rush to implement combichem to meet the demands of HTS capacity?

The early, unthinking, approaches tended to generate large numbers of highly lipophilic, relatively large, compounds

Permalink to Comment

9. Anonymous on November 25, 2013 12:02 PM writes...

the closer you are to the bench, the more you agree with the author.

Permalink to Comment

10. Anonymous on November 25, 2013 12:15 PM writes...

That "Ah, now I get it" feeling is often a sign that you're making headway on some tough subject, but you can also get it when you're understanding something that doesn't help you with it at all.

...or maybe when the "it" you just got isn't the "it" you actually care about, but close enough to seem so. Everybody loves feeling smart, but that's not what we get paid for - we get paid for doing useful things (I hope).

It probbably gets more seductive when the ability to feel that you've found "it" (and to get others to feel that they have) is a requirement to get more money from people, which tends to be an expensive way to feel smart for investors (you're not paying us so you can feel smart, you're paying us to make money), and a scientific version of being a faith healer for the investment fetcher.

Permalink to Comment

11. Cellbio on November 25, 2013 12:58 PM writes...

Clicked on the link above for Kahneman's book and see that the Kindle version is only $2.99. Love this site, interested in the book, so bought it through the link, thanks Derek.

Permalink to Comment

12. Iridium on November 25, 2013 1:17 PM writes...

There are no rules.

I never understand people the work milions of hours to write up "rules" and people that spend milions of hours to show they are wrong.

Nobody with a bit of brain could think LE=0.45 is better than 0.43. And better for and than what? The message should just be...they both bind very well. Let see the other properties.

Just use your brain, logic, experimental data and chemistry! Get to know you molecules. And think 3D.

And than give luck a chance to hit you.

Permalink to Comment

13. Chemjobber on November 25, 2013 1:51 PM writes...

@11: Did the same thing just now. Thanks for suggesting it as an idea!

Permalink to Comment

14. Lyle Langley on November 25, 2013 1:59 PM writes...

@12, Iridium,

I'm not a big fan of these calculated properties, but you have to know the example you just gave is ludicrous, right? You are correct, nobody with a brain thinks that 0.45 is different than 0.43; and that is not what these parameters are telling you. What the proponents of these parameters are telling you is: compounds that have LE > 0.4 are better (for the most part) than compounds with LE ~0.25. Then all other factors most be evaluated. I have worked in multiple labs and nobody thinks the way you just described (and please don't tell me some MBA or manager tried making that argument, either). If you're not a fan of these parameters, you can make your case with a better argument than you just did.

Permalink to Comment

15. Anonymous on November 25, 2013 2:09 PM writes...

We like rules because they are simple and, as Kahneman explained, our brains are lazy so we easily accept them. But a rule is no more than a correlation which is based only on past data, however that may have been acquired, filtered and analysed, so we need to keep that in mind and question everything.

Permalink to Comment

16. Chrispy on November 25, 2013 2:25 PM writes...

I'm stuck on the IC50 > 10,000nM being the most probable description of an oral drug (from the paper). Are there a lot of oral drugs with IC50s over 10uM?

Permalink to Comment

17. jgault on November 25, 2013 2:38 PM writes...

I wish you were right, I wish no one ever made the arguement that LE 0.45 is better than 0.43, but unfortunately I have seen this kind of stupidity too many times to dismiss the behaviour as extraordinary, and to some extent I think this is the point of the discussion (Deeper thought is required on these metrics). I think Shultz has done great service to the industry by pointing out the mathmatical falicies in some the metrics we are using, good paper.
This paper reminds me of when I read the first paper on LE and thought, does this mean water is the best possible drug? Come to think of it, maybe that might be worth some thought too?!?

Permalink to Comment

18. anon0mouse on November 25, 2013 2:44 PM writes...

Interesting article and clearly a smart guy, but unfortunately all this brain power is being spent on such post hoc analyses rather than on something more important like actually discovering a drug. To me, the cottage industry of drug discovery experts (like Bernard Munoz) who find it easier and more lucrative to pontificate on the failures of others than to dare to attempt it themselves is growing at an alarming rate.

Permalink to Comment

19. Lyle Langley on November 25, 2013 2:55 PM writes...


No disrespect intended, but the statement "I wish you were right, I wish no one ever made the arguement that LE 0.45 is better than 0.43, but unfortunately I have seen this kind of stupidity too many times to dismiss the behaviour as extraordinary", is not believable. Just isn't.

Permalink to Comment

20. jgault on November 25, 2013 2:57 PM writes...

While I agree with the sentiment that doing is more difficult than discussing, I disagree that this work is unproductive. What Schultz is giving the drug discovery scientist is a tool to use against bad management decision making. Pointing out the statisitcal flaws and limitations of oversimplified metrics is likely to allow many scientists to advance their experiments into areas that some of them are not allowed to pursue currently because of the misconception that these metrics are predicitive. I applaude the effort. While Schultz and Munoz are not inventing drugs thier efforts are certainly more worthwhile than the authors of the next rule-of-5 envy paper.

Permalink to Comment

21. jgault on November 25, 2013 3:04 PM writes...

Sorry you don't believe Iridium and I, and I am happy you work with smarter people than we do.

Permalink to Comment

22. Kazoo Chemist on November 25, 2013 3:07 PM writes...


I once was a referee for a paper submitted to J. Med. Chem. In a discussion of synergy the authors drew a distinction between 0.49 and 0.50 (synergistic and additive respectfully)! As amazing as that might seem, the values were obtained by dividing values from a two-fold dilution scheme. It seems that their calculator didn't round one of the results up to 0.5. The paper never saw the light of day.

Permalink to Comment

23. Anonymous on November 25, 2013 3:10 PM writes...

@20: As a case in point, I previously set up a biotech company to develop short peptide derivatives to inhibit amyloid aggregation as drug candidates for AD. The CEO (ex-Pfizer) stopped the program and ultimately killed the company purely on the basis of Lipinski's Rules, since the MW was 650, even though the compounds were soluble in both octanol and water, and went straight through Caco-2 cell membranes. They were later shown by academic groups to be orally bioavailable and BBB permeable only after the patents were allowed to lapse and the company went into liquidation after he told potential investors the drugs could never work! I later met and told Lipinski what happened and he said the guy must be stupid for following his rules so dogmatically without using his brain.

Permalink to Comment

24. Mad Dog on November 25, 2013 3:38 PM writes...

The questions in footnote 1 are flawed for what the author wants to acheive. They need to be phrased such that "Statitistically speaking is it better for a group of 10 people to choose A or B, and then C, or D". In that case you then have statistics, granted with a low n-value. But for an individual being asked "Do you want $800 guaranteed or an 85% chance of getting $1000? Well, then I am just as irrational as the next guy; I'll take the $800, thank you!!

Permalink to Comment

25. Anony on November 25, 2013 4:12 PM writes...

Schultz is as guilty of generalization as the metric users he accuses. I don't really care if a metric represents something "real" or not. A metric is a model. What I care about is whether it allows us to see interesting trends that point to more druglike molecules. Drug discovery is not about mathematical rigor, it's about utility.

Permalink to Comment

26. Anonymous on November 25, 2013 4:18 PM writes...

@25: Want utility? Then use your brain. That's all Shultz is saying, and good thing too, because too many people forget they have a brain when they follow dogma.

Permalink to Comment

27. MoMo on November 25, 2013 6:00 PM writes...

CL would say rules will be and are meant to be broken. Then he'd probably laugh at this for its obviousness and the lack of intellect that is endemic to the Pharma industry.

Permalink to Comment

28. iridium on November 25, 2013 6:03 PM writes...


I agree with you on the value of LE. I use it all the time.

But you are wrong about the fact that "nobody thinks the way you just described".
Not later than a couple of month ago on this blog there has been a huge discussion about:
should we calculate LE using number of heavy atoms or MW?...who cares!!!

So more people than you think look at the number.
You have to try to go to a FBDD conference than you will tell me =)

Permalink to Comment

29. Lyle Langley on November 25, 2013 10:37 PM writes...

@Iridium and jgault...

If you really work at a place where the knowledge you described is not "extraordinary", and people actually believe (and defend) there is no difference between 0.45 and 0.43, then someone needs to turn out the lights, lock the doors and give any money back because the people don't know what they are doing. In fact, the people in charge have defrauded the funding agencies (investors or grants) because they don't have the ability to carry out the work. We all know there are manuscripts that have those correlations and those are usually pretty crappy papers to begin with, but that is a far cry from saying that is the rule and not exception.

But what I believe is going on here is the two are over exaggerating the pervasive attitudes to try and make a point.

And, yes, I do work with very smart and intelligent people - just like the vast majority of all posters on this site (except, of course, you two).

Permalink to Comment

30. Nysted Reagent Never Works on November 25, 2013 11:17 PM writes...

It's a relevant paper. However, I think I should write a paper demonstrating that, in the past 5 years, there is a near perfect inverse correlation between the number of metric-related papers versus the number of compounds pushed into humans. Pfizer, Novartis and AZ together have probably published more metric-related papers than result-driven scientific papers. This is a serious concern.

Permalink to Comment

31. Anonymous on November 26, 2013 1:54 AM writes...

@30: Results are so 80's!

Permalink to Comment

32. Heretic Computational Chemist on November 26, 2013 2:51 AM writes...

The reality is that the medicinal chemists are blind within the chemical space, and for a blind guy is more confortable to accept and follow a set of rules (no matter how much stupid), than think with his own brain. I didn't saying to completely discard rules, but the most important thing is to not assimilate them into a dogma, and identify when it is possible to bend them.
We are damnt scientists not priests of the Lipinsky's church.

Permalink to Comment

33. Jonas on November 26, 2013 6:01 AM writes...

Kahneman is really a true genious. But when reading his book I got the (somewhat dissapointing) impression that he would love the rule-of-five. For example, he seemed to be a great fan of the durability of marriage formula: “frequency of love-making minus frequency of quarrels.”. A very, very simple equation, which admittely does better than the average marriage counselor in predicting whether a marriage will last. But it's as crude (and useless) as the rule-of-five

Permalink to Comment

34. Jose on November 26, 2013 6:46 AM writes...

A sly allusion to Diax's Rake ?

Permalink to Comment

35. Anonymous on November 26, 2013 8:41 AM writes...

Those of us working in natural product-based drug discovery have long ignored Lipinski's rules, because Lipinski himself said natural products don't follow the rules. And there's nothing inherently unusual to the chemistry of natural products (except, sometimes, their size and complexity).

Permalink to Comment

36. Pete on November 26, 2013 12:15 PM writes...

It's worth remembering that Ro5 is simply a statement of property distributions for compounds that had been taken into Phase 2 at some point before the Ro5 article was published.

The fundamental problem with LE is that relative values of LE depend on the concentration used to define the standard state. Insights that depend on the value of the standard concentration are not thermodynamically valid since the choice of the concentration is arbitrary. I've linked a blog post in which some of this is discussed as the url for this comment.

Permalink to Comment

37. anon on November 27, 2013 6:31 AM writes...

Classical LE is properly defined by the KD determined by binding kinetics as opposed to assay potency which people use interchangeably while ignoring the issues with (p)IC50 's etc.

Permalink to Comment

38. jgault on November 27, 2013 11:42 AM writes...

Normally at this point in the discussion I would move on. However, I think a story of mine might help people understand where I am coming from. First of all, we appear to agree that the use of metrics in this way is moronic, we just seem to disagree to the extent this effects our industry. I had a development program in for preclinical review last year. We were going through the standard discovery data set presenting to managment (ADME, Efficacy, safety, Pharm Sci.) The objective being to get the company to invest in preclinical development (4-5 million dollar investment). At the end of two hours of presenting the data a senior chemist on the management team pulled out the Pfizer 3/75 paper and suggested that the compoud should not move forward because the PSA was 78 and the clogP was 3.3 and this represented a 3x increase in risk of toxicity in preclinical development (Ironically since the compound was efficacious at a target Cp of 5 nM the attrition risk was near zero, look closely at the first chart in the Pfizer paper). Since our managment team was all non-chemists except this moron they relied on his judgement to make the decision to not move forward. After 3 months of lobbying and the help of the Kenny and Novartis papers I was able to convince managment that the 3/75 paper was not to be relied on for these kinds of decisions. The compound is now successfully moving forward to the clinic, but this moron cost us 3 months and about 4 million in lost time. My point is not that this is "pervasive" but it is also not "extraordinary". Should he be fired? Absolutely! Should we pack up and disolve the company? I hope better minds can prevail, so if it is OK with you I will keep trying. Remember that most big managment decisions in pharma are not made by chemists, and they rely on us to judge risks and present them, if we use simplistic rules they will remember them and probably misues them to our demise. I would say only 5-10% of chemists severely missuse these metrics but when one of them gets into managment they can cause serious problems.

Permalink to Comment

39. DrSnowboard on November 28, 2013 4:24 PM writes...

I presented at a review committee on a series of compounds which were poorly soluble. Swedish chap pipes up 'we do not make any compound that has a predicted solubility I believe his research site has since closed.

Permalink to Comment

40. Suxin on November 28, 2013 8:52 PM writes...

All those metircs came from statistics, so they should only work well on number of the cpds, like Derek said worked for million cpds. So I think those metrics should not be taken as abolutely "cutoff", and it should give you the most likely druglike chemistry space.

Permalink to Comment

41. DrSnowboard on November 29, 2013 7:47 AM writes...

comment 39 got mangled somehow... it was meant to read 'e do not make any compound that has a predicted solubility of

Permalink to Comment

42. DrSnowboard on November 29, 2013 8:06 AM writes...

Nope, still garbage. Maybe your comments box doesn't like the word 'defaecated'

Permalink to Comment

43. Jim Rinker on December 20, 2013 2:18 AM writes...

Well if most medicinal chemists are designing drugs based on the rule of 5 MW

Permalink to Comment

44. mike b. on December 20, 2013 5:28 AM writes...

Can we just finally admit drug discovery is more of an art and dumb luck than it is a science?

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