Corante

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: derekb.lowe@gmail.com Twitter: Dereklowe

Chemistry and Drug Data: Drugbank
Emolecules
ChemSpider
Chempedia Lab
Synthetic Pages
Organic Chemistry Portal
PubChem
Not Voodoo
DailyMed
Druglib
Clinicaltrials.gov

Chemistry and Pharma Blogs:
Org Prep Daily
The Haystack
Kilomentor
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
ChemBark
Realizations in Biostatistics
Chemjobber
Pharmalot
ChemSpider Blog
Pharmagossip
Med-Chemist
Organic Chem - Education & Industry
Pharma Strategy Blog
No Name No Slogan
Practical Fragments
SimBioSys
The Curious Wavefunction
Natural Product Man
Fragment Literature
Chemistry World Blog
Synthetic Nature
Chemistry Blog
Synthesizing Ideas
Business|Bytes|Genes|Molecules
Eye on FDA
Chemical Forums
Depth-First
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
FuturePundit
Aetiology
Gene Expression (I)
Gene Expression (II)
Sciencebase
Pharyngula
Adventures in Ethics and Science
Transterrestrial Musings
Slashdot Science
Cosmic Variance
Biology News Net


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


Economics and Business
Marginal Revolution
The Volokh Conspiracy
Knowledge Problem


Politics / Current Events
Virginia Postrel
Instapundit
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

« 36 Basic Types of Reactions? | Main | Ariad's Ponatinib Runs Into Big Trouble »

October 9, 2013

The 2013 Nobel Prize in Chemistry

Email This Entry

Posted by Derek

The 2013 Nobel Prize in Chemistry has gone to Martin Karplus of Harvard, Michael Levitt of Stanford, and Arieh Warshel of USC. This year's prize is one of those that covers a field by recognizing some of its most prominent developers, and this one (for computational methods) has been anticipated for some time. It's good to see it come along, though, since Karplus is now 83, and his name has been on the "Could easily win a Nobel" lists for some years now. (Anyone who's interpreted an NMR spectrum of an organic molecule will know him for a contribution that he's not even cited for by the Nobel committee, the relationship between coupling constants and dihedral angles).

Here's the Nobel Foundation's information on this year's subject matter, and it's a good overview, as usual. This one has to cover a lot of ground, though, because the topic is a large one. The writeup emphasizes (properly) the split between classical and quantum-mechanical approaches to chemical modeling. The former is easier to accomplish (relatively!), but the latter is much more relevant (crucial, in fact) as you get down towards the scale of individual atoms and bonds. Computationally, though, it's a beast. This year's laureates pioneered some very useful techniques to try to have it both ways.

This started to come together in the 1970s, and the methods used were products of necessity. The computing power available wouldn't let you just brute-force your way past many problems, so a lot of work had to go into figuring out where best to deploy the resources you had. What approximations could you get away with? How did you use your quantum-mechanical calculations to give you classical potentials to work with? Where should be boundaries between the two be drawn? Even with today's greater computational power these are still key questions, because molecular dynamics calculations can still eat up all the processor time you can throw at them.

That's especially true when you apply these methods to biomolecules like proteins and DNA, and one thing you'll notice about all three of the prize winners is that they went after these problems very early. That took a lot of nerve, given the resources available, but that's what distinguishes really first-rate scientists: they go after hard, important problems, and if the tools to tackle such things don't exist, they invent them. How hard these problems are can be seen by what we can (and still can't) do by computational simulations here in 2013. How does a protein fold, and how does it end up in the shape it has? What parts of it move around, and by how much? What forces drive the countless interactions between proteins and ligands, other proteins, DNA and RNA molecules, and all the rest? What can we simulate, and what can we predict?

I've said some critical things about molecular modeling over the years, but those have mostly been directed at people who oversell it or don't understand its limitations. People like Karplus, Levitt, and Warshel, though, know those limitations in great detail, and they've devoted their careers to pushing them back, year after year. Congratulations to them all!

More coverage: Curious Wavefunction and C&E News. The popular press coverage of this award will surely be even worse than usual, because not many people charged with writing the headlines are going to understand what it's about.

Addendum: for almost every Nobel awarded in the sciences, there are people that miss out due to the "three laureate" rule. This year, I'd say that it was Norman Allinger, whose work bears very much on the subject of this year's prize. Another prominent computational chemist whose name comes up in Nobel discussions is Ken Houk, whose work is directed more towards mechanisms of organic reactions, and who might well be recognized the next time computational chemistry comes around in Sweden.

Second addendum: for a very dissenting view of my "Kumbaya" take on today's news, see this comment, and scroll down for reactions to it. I think its take is worth splitting out into a post of its own shortly!

Comments (103) + TrackBacks (0) | Category: Chemical News


COMMENTS

1. KC Nicolaou on October 9, 2013 7:57 AM writes...

Damnit.

Permalink to Comment

2. Rafael Najmanovich on October 9, 2013 8:23 AM writes...

Another name that should be remember here is that of Harold Scheraga...

Permalink to Comment

3. RegularReader on October 9, 2013 8:23 AM writes...

As a former computational chemistry TA, this is exciting news! Congratulations to the winners!

FYI: One of the winners is known for coma-inducing seminars...

Permalink to Comment

4. Curious Wavefunction on October 9, 2013 8:24 AM writes...

Always nice to wake up and see your own field recognized. Well-deserved, although I would have included Norman Allinger one way or another. Also definitely an award recognizing a field rather than just individuals.

Permalink to Comment

5. Bob Sacamano on October 9, 2013 8:29 AM writes...

The Nobel proof-readers let one get by in the linked document.

Page 5: 'Praiser-Parr-Pople' should read 'Pariser-Parr-Pople.'

Permalink to Comment

6. Anonymous on October 9, 2013 8:33 AM writes...

Bob - you need to get out more

Permalink to Comment

7. JTA on October 9, 2013 8:35 AM writes...

I'm just curious, what are examples of scientific problems that the recipients have solved using "multiscale models"? Or woud you say that this prize was given rather for laying the theoretical framework (force fields, etc) for methods that may deliver new science today or in the near future?

Permalink to Comment

8. Anonymous on October 9, 2013 8:39 AM writes...

From the Yahoo News version of the story:

"The Nobel jury said the tool is "universal", helping pharmaceutical engineers to design new drugs or engineers to make cleaner energy sources or smarter manufactured products."

http://news.yahoo.com/karplus-levitt-warshel-win-nobel-chemistry-prize-101235340.html

Permalink to Comment

9. a on October 9, 2013 8:47 AM writes...

Also, Bill Goddard might be feeling ticked too.

Warshel, how can I put it, his reputation for, putting it mildly, 'prickliness' precedes him..........

I guess the chip on his shoulder will be lifted somewhat.

Permalink to Comment

10. Anonymous on October 9, 2013 9:05 AM writes...

@7: The specific problem they solved is: How to model complex molecular dynamics accurately and efficiently given limited computing power.

Permalink to Comment

11. DAD on October 9, 2013 9:14 AM writes...

Agree with post 9...Goddard was there very early and has made enormous contributions at all levels from analytical theory to computational methods and frameworks to software applications to full application across a variety of academic efforts and industries. The winners all have a biology bent to their work, so maybe Sweden will one day recognize multi-scale computational work in non-biological systems like materials and reaction chemistry....(Houck/Goddard?)...

Also agree that the three chosen here have not oversold their craft (even if the Nobel committee has in their write-up...;)

Permalink to Comment

12. annon on October 9, 2013 9:38 AM writes...

Nobel prize for an area that is oversold and often badly used. Sad, sad, sad.

Permalink to Comment

13. Anonymous on October 9, 2013 9:48 AM writes...

Seems to be a big year for theory vs experiment & discovery, given this and the Physics prize.

Permalink to Comment

14. Anon on October 9, 2013 9:55 AM writes...

@1 Ha!

Permalink to Comment

15. luysii on October 9, 2013 10:02 AM writes...

Amazing! He's the 7th Nobelist present in the Harvard Chemistry department from 60 - 62 when I was a grad student there. Even better these guys weren't sitting there resting on their laurels, but in the process of actively creating them.

They are Woodward, Corey, Bloch, Lipscomb, Hoffmann, Gilbert and now Karplus. It should be noted that 3 of them were Hitler's gift to the USA -- Bloch, Hoffmann, Karplus -- all of whose parents fled the Nazi's because of they were Jews.

The place was full of European expats back then, and Don Voet (whose parents also got out of Europe for the same reason) used to say that the universal scientific language was broken English.

I wonder if 30 or so years hence we'll have all sorts of muslim nobelists, given the carnage going on there presently. Hopefully, we will.

P. S. Even though Gilbert was a physics prof, he was hanging around the chemistry department a lot of the time back than, and his Nobel work was definitely chemical.

Permalink to Comment

16. leftscienceawhileago on October 9, 2013 10:04 AM writes...

"Scientists who took chemistry into cyberspace win Nobel Prize"

...

Permalink to Comment

17. JTA on October 9, 2013 10:20 AM writes...

@10: What is an example of a complex molecular dynamics problem that has been addressed accurately? For example in the simulation of protein dynamics the consensus is that the models are not accurate enough and the simulations are not exhaustive (ie, efficient) enough.

To me, merely implementing "complex molecular dynamics" is not an achievement of such fundamental nature that it would warrant a NP of its own. The prize would of course be warranted if the technique is useful in answering real chemical or biological questions. But to my understanding QM/MM is not there yet. Maybe the NP committee has faith that QM/MM will become more useful in the future, and is giving the prize now for the theory behind it?

Permalink to Comment

18. anthony nicholls on October 9, 2013 10:34 AM writes...

Oh, please, Derek. I'm disappointed in you. Yes, perhaps Karplus deserved one for his many contributions, but MD? Really? Science is about making predictions that experiment confirms, not finding systems that sometimes agree with simulations, which is what MD does. It's a technique of essentially zero utility in pharma and almost no use outside pharma, which damages attempts to do real science by its patina of being 'real', i.e. heaven forbid you don't actually move atoms around and still get thermodynamic averages. It is not well-deserved, it's a disaster for those of us who would like to see computational chemistry be a predictive science. Finally, on your comment about Levitt and others tackling proteins because they were such good scientists- what rot! They did that because it was sexy and they could make movies that got grant money. One thing is was not was good science.

Permalink to Comment

19. coupled cluster on October 9, 2013 10:41 AM writes...

Most of Ken Houk's work uses other people's functionals such as Becke's B3LYP or Truhlar's M06 and little else. That's like using Heck or Suzuki coupling to make blablablamycins and demanding a Nobel prize for it.

Permalink to Comment

20. JH on October 9, 2013 10:41 AM writes...

I just hope that they give a prize to Carl Djerassi while they still can. He pioneered some computational tools and algorithms for chemical structure prediction AND he's done a mountain of fundamental research on dozens of other topics. I think that it would be nice if he were in the running.

Permalink to Comment

21. DAD on October 9, 2013 10:55 AM writes...

@Anthony...agree with your criticism of my post. I should have prefaced my whole post with "if Sweden is going to award a prize for this area..." I was looking at this a bit more optimistically...the award is not not a disaster as long as it's viewed as important early contributions that only barely started the path toward computational chemistry as a predictive science. You have more wisdom than I to judge that perception, and you may have stronger feelings that some of the work actually derailed the path toward a truly predictive science (there a predictive exceptions, so it's hard to state the case that it's all worthless). Synonymous with your take, one could easily argue that with modeling being so decisively predictive in other fields, such as Aeronautical Engineering, that the "meso-scale" modeling in our field has not cracked the nut it was was designed to crack...and hence should not be considered Prizeworthy.

Permalink to Comment

22. Anonymous academic on October 9, 2013 11:00 AM writes...

@9: I've never even met Warshel, but just reading some of his papers was enough to give me the same impression. I'm not an expert in the field, but I could have sworn that some of them amounted to "our simulations prove these experiments wrong".

Permalink to Comment

23. Myma on October 9, 2013 11:10 AM writes...

For those not familiar, the winners directly (or indirectly) (or with their groups) created the theory and the actual code behind CHARMm, which is used as a basis for computational chemistry on many many other software platforms as well as used by itself.
Interestingly, since I looked it up this morning, there _is_ a wikipedia page for CHARMm, which lists the Nobel winners as well as the other shoulda/coulda/woulda's people mention in other comments. If anyone is a wikipedia author/editor (which I am not), that page could perhaps be updated to mention the Nobel prize.

Permalink to Comment

24. Curious Wavefunction on October 9, 2013 11:14 AM writes...

Anthony, you have a point but I don't quite share your cynicism. Science is not just about prediction, it's about explanation (David Deutsch says this very well in his latest book). MD simulations can often point the way toward other experiments that need to be done by shedding light on plausible operating factors. Perhaps they can suggest conformational flexibility of certain side-chains affecting ligand binding which can be tested by mutagenesis, perhaps they can point to loop movements which might affect crystal packing and hamper crystallization. In one of my own projects I have used them to rationalize differences between in vitro and cell activity. Maybe they don't always help in a big way, but they can certainly help in many little ways.

I don't always see MD simulations as predictive tools, I see them as suggestive tools which point to a new experimental direction or constrain the number of known possible explanations. In any case, I think the Nobel Prize this year was more of a lifetime achievement award for the general field of computational chemistry than for any specific technique. It really makes a statement that the field as a whole is being widely employed in diverse fields of chemistry.

Permalink to Comment

25. Anonymous on October 9, 2013 11:24 AM writes...

I'd rather see a Nobel awarded for masturbation, because it's just as useless, but at least more interesting/enjoyable.

Permalink to Comment

26. RC on October 9, 2013 11:27 AM writes...

@1 is an LOL.

Permalink to Comment

27. Per-Ola Norrby on October 9, 2013 11:35 AM writes...

To answer some who say that QM/MM hasn't acomplished anything yet; I'm competing with the technique, using another method, but I still admire the results, the answers we can get about the reasons for reaction selectivity (one of humanities toughest challenges if we want to build a sustainable society). An example I know well is Maseras explaining why the AD (2001 Nobel) can reach such extreme selectivities, and when. Plenty of examples in enzyme mechanisms also. If we don't learn the underlying causes, how can we ever make rational improvements? And build a less wasteful society? And many of these problems have a size and complexity that makes QM/MM a preferred method

Permalink to Comment

28. Anonymous on October 9, 2013 11:47 AM writes...

@24,27: Any fool or tool can explain things retrospectively, but that's not science (nor useful) unless it makes testable predictions that prove to be correct.

Permalink to Comment

29. Per-Ola Norrby on October 9, 2013 12:02 PM writes...

@28 So go into for example the Maseras papers on AD and look at the predictions and results. I was impressed. You can't just say it hasn't been done, because there are examples. But we're usually not allowed to publish predictions unless the experimental verification is in the same paper. I can give additional examples from my own work if you wish, but only QM or MM, not QM/MM. I only ever used that for rationalization myself (still science, I'd say, new understanding)

Permalink to Comment

30. Anonymous on October 9, 2013 12:17 PM writes...

@29: What protein structures has QM/MM predicted to fold from just its primary sequence? Or what is the most complex reaction or rearrangement it has predicted, except for basic local energy minimization?

Permalink to Comment

31. pipeline_junkie on October 9, 2013 12:23 PM writes...

I recall a comment by Harry Gray at a lecture... something like... "years ago, we would ask the computational chemists their predictions then go with the complete opposite. Back then, that was a sure win. Nowadays, you ask them and their right 50% of the time. That's how much better computational chemistry is getting."

Permalink to Comment

32. luysii on October 9, 2013 12:33 PM writes...

Well if Houk ever gets the Nobel, he'd be #8 -- see comment #15. Although an undergraduate in '60 - '62, he participated in the Woodward seminars, and was already quite impressive

Permalink to Comment

33. MarcelSwart on October 9, 2013 12:55 PM writes...

Note, the prize is for QM/MM, not for MD!
How to effectively separate this into meaningful portions in an efficient way is what they received the prize for. How to get to the geometries (either through MM geometry optimization, Monte Carlo or Molecular Dynamics) used for it is explicitly left out:
"The importance of the work of the laureates is independent of what strategy is used for the choice of studied configuration(s). The prize focuses on how to evaluate the variation in the energy of the real system in a accurate and efficient way for systems where relatively large geometry changes or changes in electronic configuration in a smaller part of the studied system is strongly coupled to a surrounding that is only weakly perturbed."
http://www.nobelprize.org/nobel_prizes/chemistry/laureates/2013/advanced-chemistryprize2013.pdf

Permalink to Comment

34. anon the II on October 9, 2013 1:06 PM writes...

As an organic chemist, I, like Derek, have known about Karplus from his famous equation since I was an undergraduate 38 years ago. Oddly enough, I've never heard of the other two. How did that happen? I also would have thought Allinger would have been in there. It was good to see Scheraga get a mention, though a mention in the Pipeline comments isn't quite a Nobel. Maybe one day.

As for #18 Anthony's comments on MD. I tend to agree. MD has largely been a distraction and has probably hindered implementation of better methods of conformational searching.

Permalink to Comment

35. dearieme on October 9, 2013 1:10 PM writes...

"@24,27: Any fool or tool can explain things retrospectively, but that's not science ...": maybe not, but it seemed to be a large part of chemistry when I was an undergraduate.

Permalink to Comment

36. Anonymous on October 9, 2013 1:10 PM writes...

Interesting to read your first sentence, Derek.

The official NobelPrize.org site mentions:

Martin Karplus Université de Strasbourg, France and Harvard University, Cambridge, MA, USA.

The usual bias...;-)

Permalink to Comment

37. Sili on October 9, 2013 1:13 PM writes...

I just hope that they give a prize to Carl Djerassi while they still can. He pioneered some computational tools and algorithms for chemical structure prediction AND he's done a mountain of fundamental research on dozens of other topics. I think that it would be nice if he were in the running.
To Hell with that. He deserves the Peace Prize for The Pill. But of course noöne will dare give him it.

Just look what junk they gave the Medicine/Physiology prize for last year. Who needs more babies?

Permalink to Comment

38. Anonymous on October 9, 2013 1:13 PM writes...

The closest I ever came to a Nobel Prize, was renting an apartment from John Pople's (1998 Chemistry Prize) brother back in my graduate school days (Bristol, UK)!

Permalink to Comment

39. Curious Wavefunction on October 9, 2013 1:19 PM writes...

#28: If you think explanation is not an important part of science then I am afraid you don't have a very good idea of what science is. Before prediction comes explanation: Einstein had to explain known facts about gravity before he could predict the perihelion of Mercury. Darwin had to explain known facts about species before he could predict how new ones are formed. If you think only fools do explanation then surely you must think that Einstein and Darwin were fools. And of course, explanation is still better than no explanation at all.

The example I cited above was in fact a predictive example. MD predicted the kinds of functional groups - including linker lengths - that would stabilize the loop. The predictions were validated by synthesis and the synthetic chemists appreciated it. The main problem with computational chemistry in my experience comes from both computational and non-computational chemists who overstate the utility of the methods and then complain when reality does not meet their unrealistic expectations. Computational chemistry is a tool, just like any other tool in chemistry, and you cannot blame a tool if its practitioners don't use it judiciously.

Permalink to Comment

40. Anonymous on October 9, 2013 1:31 PM writes...

@39: Then I say everything works by act of God, or magic. They explain everything. Is that good enough "science" for you?

Permalink to Comment

41. Anonymous on October 9, 2013 1:38 PM writes...

PS. Or do you expect me to make testable predictions with those explanations?

Permalink to Comment

42. Curious Wavefunction on October 9, 2013 1:48 PM writes...

No, that is not science because science deals with material entities, not supernatural ones. By the way that is precisely the point that David Deutsch makes in his book; you can invoke black boxes ("God" for instance) and make apparently testable and fairly accurate predictions, but it does not lead you any closer to science, let alone understanding. That is why prediction, while important, is not the Holy Grail of science.

Permalink to Comment

43. H2L on October 9, 2013 1:50 PM writes...

Why are most people in this thread talking about molecular dynamics (MD)? This award is for the combination of molecular mechanics and quantum mechanics (QM/MM). There are plenty of examples where QM/MM has been used to predict reaction rates, enzyme mechanisms, etc. These predictions have been later validated experimentally. While QM/MM is not a panacea, it definitely has a predictive domain of applicability.

I tend to agree with the people that say MD has not been fully proven yet. That is to say, predicting things like protein folding pathways and kinetics of biological processes are very hard and the methods have to improve before they are reliable enough to truly impact drug discovery. But I digress...let's get back to the 2013 Nobel Prize in QM/MM.

Permalink to Comment

44. Anonymous on October 9, 2013 1:55 PM writes...

@42: Your definition of what is supernatural vs science seems a little subjective. Please could you explain the difference.

Permalink to Comment

45. Anonymous on October 9, 2013 2:01 PM writes...

PS. Surely evolution, relativity and quantum mechanics are all just constructs of the mind, just like religious faith, and magic, or little green men for that matter. The difference, and only difference between them is in their ability to make testable predictions!

Permalink to Comment

46. KG on October 9, 2013 2:04 PM writes...

I have been in the field for a long time, and I know, and have interacted with, all of these individuals. I think it's nice that the field of compchem is recognized outside of pure quantum mechanics. And, unlike some of the commenters here, I don't ascribe to the view that MD is nonsense only good for retrospective analysis. To take only the most obvious example, min/MD is part-and-parcel every xray and NMR structure determination done in the past ~two decades.

Obviously, there are plenty of studies that have used MD to prove, mostly, that you can use government research money and computer cycles to unfold proteins with bad force fields, or to make molecules jiggle around with no real predictive payoff. But there are also ample prospective studies now in the literature (and not in the literature when it comes to use in pharma) that demonstrate the usefulness of macromolecular simulations.


I agree that if the intent here is to give the prize for QM/MM calculations, then it's probably pretty premature. But though that's the stated purpose--because the Nobel committee doesn't like to give out "lifetime achievement" awards--the reality would seem to be different. In particular, Karplus' award would assuredly be more for a body of work, than for his specific contributions to QM/MM calculations.

As for individual unpleasantness: Well, I have to agree there. I've interacted and/or worked with most everyone from these early generations of comp chemists, and one of the three inducted here is, well, not the most generous or sweetest of their (or any) generation. The second has a prickly exterior, but a more generous interior. And the third seems to have a second Ph.D. in sitting in talks oblivious to the speaker while he taps away at his keyboard, but is otherwise fine.

To be honest, given the political nature of the Nobel Prize, I'm a little surprised that some of these individuals made it through the process. But I suppose the answer there lies in the fact that while they can be unnecessarily acidic and unpleasant, only one could possibly be accused of being a dreaded self-promoter, and even he is far from the worst of the breed.

And I agree with some of the other comments that mention others who have made huge, lasting, contributions to the field who somehow missed out here. What can you say?

Permalink to Comment

47. Anonymous on October 9, 2013 2:09 PM writes...

@46: Too much time spent talking to computers vs people?

Permalink to Comment

48. watcher on October 9, 2013 2:20 PM writes...

#46: What is said on this side really does not matter, now, does it?

Permalink to Comment

49. Pig Farmer on October 9, 2013 2:57 PM writes...

@38: the closest I ever got to a Nobel Prize was reading a book by Francis Crick.

Permalink to Comment

50. Nobel Fries on October 9, 2013 3:10 PM writes...

The closest I got to a Nobel Prize was by living in a country where Nobel Laureate Obama was president.

Permalink to Comment

51. sgcox on October 9, 2013 3:14 PM writes...

Ok, While I normally agree with Curious Wavefunction (~42), here I part the ways. MD is a BlackBox (God ?) with no predictive functions. Example: you run MD on protein A for N femtoseconds. Can you now predict what will happen to protein B after M femtoseconds of MD _BEFORE_ running MD of protein B for M femtoseconds.
No? So you are left no wiser after this hmm, exercise.. So how can you call it a science, however defined ?
Notice, I am not even invoking that awful, inconvenient and always annoying entities called experiment results...
As to Einstein and Darwin spirits here, just please, no.

Permalink to Comment

52. Hardass Slavedriver on October 9, 2013 3:22 PM writes...

@1: Kiss my @ss KC, I'll get the Nobel before you do.

Permalink to Comment

53. John Chodera on October 9, 2013 3:27 PM writes...

@18: Ant: Even though there's been a Nobel for "retrodictive" molecular modeling, there's still one to be had for *predictive* molecular modeling!

Permalink to Comment

54. Curious Wavefunction on October 9, 2013 3:27 PM writes...

#51: But it depends on what you are trying to do, right? If A and B are similar enough and you are trying to predict what happens in a gross way to a certain loop then you might well be successful. If you are trying to predict whether a particular phenylalanine residue in the active side might have its dihedral angle gauche instead of anti then MD might woefully fail. As with many other techniques (computational or experimental), the degree of prediction depends on the question you are asking and the quality of the data you are starting out with.

Let's consider another predictive MD example. Recently Houk, Baker and others used MD to look at the structural integrity of active sites in de novo designed enzymes. Basically they found out that to a decent extent they could predict which ones would be good catalytic sites vs which ones would not essentially based on how much the active site residues rattled around. "Looser" binding sites made for bad catalysts. Again, an example of a narrowly defined problem that could be usefully addressed by MD.

Permalink to Comment

55. Anonymous on October 9, 2013 3:30 PM writes...

This Nobel is for predicting which experiments are wrong. LOL

Permalink to Comment

56. sgcox on October 9, 2013 3:35 PM writes...

Actually I have to correct myself here. The inability to predict one set of calculations from other set is irrelevant - nobody cares. But I still struggle to understand the worthiness of reading any of MD papers - What did I actually learn ? Any worthy knowledge or pathetic narrative of 0 and 1 redistributions in computer memory which leaves me no wiser about what will happen in my lab tomorrow ?

Permalink to Comment

57. sgcox on October 9, 2013 3:59 PM writes...

~54. Curious Wavefunction.

Yes, that was a good paper. Baker has a real spark.
I think me, and some other people here, are a bit harsh which is understandable given the hype associated with NP. Nobel prize is often given not for brilliant science but for sme useful tools for all of us:
GFP
PCR
etc.

Yes, Comp. Chem, has its use for sure.

I just resent the pathos of calling it a triumph of "chemistry done in cyberspace"

Permalink to Comment

58. MarcelSwart on October 9, 2013 4:24 PM writes...

@51 (sgcox): "Example: you run MD on protein A for N femtoseconds"

fs = 10^(-15) s, unless you're interested in photochemistry, J. Phys. Chem. A 2001, 105, 3583-3590, http://dx.doi.org/10.1021/jp002955+, you don't do MD with fs, and even less for proteins

In my old days 5 ns would be considered a long time, nowadays protein MD simulations of 50 ns are considered short, and microseconds (10^(-6) s) are at the order of the day, especially with DE Shaw's machine (a candidate for a future MD NobelPrize?)

Permalink to Comment

59. Sweden Calling on October 9, 2013 4:45 PM writes...

Sweden here. Bad misstake. Sorry. QM/MM, MD are useless in real life. We're taking the prize back and would like to offer it to Lipinski for his rule explaining the past and influencing the future. At least he is a nice guy...

Permalink to Comment

60. Typical I guess on October 9, 2013 4:47 PM writes...

I find it interesting (and disappointing) that the comments associated with this article seem (far) worse than usual for this blog. Maybe it shows the serious divisions in the field between practitioners of different approaches or maybe it's the result from large-scale penetration by "normal" commenters into a usually esoteric affair.

Reads like Hacker News or worse on a bad day, it does.

Permalink to Comment

61. Lunar landing on October 9, 2013 5:27 PM writes...

@ #60

What?

Permalink to Comment

62. Philip Larkin (deceased) on October 9, 2013 5:34 PM writes...

From a high window in purgatory above Hull University, a miserable Dead Poet reads @25 and @37 loud and clear.

Bit rich that while computer graphics gets the glittering prize, the Norse Gods turn a perennial blind mind to the bit of practical organic chemistry that opened the door to the Annus Mirabilis of 1963, when according to a Dead Poet, "life became a brilliant breaking of the bank, a quite unlosable game."

Bit of a con much of the time, computer graphics, if you ask a Dead Poet. Follow Derek's link to the prose of the Norse Gods and they sure ain't live poets either...

Permalink to Comment

63. Yet another anonymous on October 9, 2013 5:37 PM writes...

Best comment so far is #25...

I don't know if anyone has read the press release: http://www.nobelprize.org/nobel_prizes/chemistry/laureates/2013/press.html

The last paragraph is: "Today the computer is just as important a tool for chemists as the test tube. Simulations are so realistic that they predict the outcome of traditional experiments."

The end of traditional experiments then, according to the Nobel committee... I would be interested to hear from industrial researchers on how useful they have found MM/QM techniques in practice.

Permalink to Comment

64. Anonzymous on October 9, 2013 6:55 PM writes...

Comments like #25 lower the quality of discourse on this site. Funny is not the same as illuminating.

I think there's some good points in this thread about both the strengths and limitations of computational methods. The press release from the Nobel site seems to me to exaggerate their importance, which is what you would expect a press release from the Nobel site to do.

A couple of years back I worked with a computational guy who was using both MM and QM (although not QM/MM) to predict the binding of some compounds. I still remember what he said - "I can't really tell you what to make, but I can tell you what not to make" - and this seems to me to encapsulate at least one useful aspect of simulations: ruling out possibilities and narrowing down the options.

Permalink to Comment

65. ScienceNerd on October 9, 2013 8:30 PM writes...

Another person long time in the molecular modeling field here. And I think #46 gets a lot of the particulars right, although I don't think a discourse on the various peculiarities of the men is relevant. (Which is not to say that that person doesn't hit those pretty close to the head).

Ultimately, I think MM or QM/MM is baked into all kinds of practical veins of science these days. And to focus on the question of whether a long MD simulation is yet valuable is kind of silly and misses the point. Does it work in every case? No. But then again, it doesn't have to. It's a tool, and a darn important and useful one sometimes.

I think this is a case where the Nobel committee is simultaneously recognizing a substantive and looming body of work by one person (Karplus) and also tipping their hat to a still emerging method set that has already demonstrated usefulness (QM/MM). And I think it's pretty clear that they haven't acted rashly here; this isn't Mössbauer, act 2.

Permalink to Comment

66. Bosco Ho on October 9, 2013 9:00 PM writes...

Like many others here, I think this focus on MD as a predictive tool is misplaced. MD is important for explanatory purposes.

However, what I haven't seen mentioned here is the that MD is focused on simulating dynamics, which is important for linking structure to thermodynamics. If you take into account the theoretical equivalency between a Gibbs ensemble and an infinite Poincare trajectory, then MD provides the connective tissue between thermodynamics/statistical dynamics and molecular mechanisms.

Without MD, it's very hard to connect up macro wetlab measurements such as calorimetry and binding assays, to static molecular structures, such as x-ray crystal structures, and highly detailed QM calculations. You need MD simulations to apply the arsenal of statistical mechanics to understand ensemble measurements, such as NMR couplings and phase changes in all sorts of biochemical measurements.

Permalink to Comment

67. MD on October 10, 2013 2:27 AM writes...

#66 MD provides an explanation of the process of binding, but it does not mean this is an accurate one. Free energy perturbation techniques based on MD are essentially unable to predict binding energy. I challenge anyone to provide references in which this has been achieved prospectively.

If a model cannot predict, then the associated explanation is very likely to be a mathematical artifact with little connection to studied system. The reasons go from the known unknowns (comprehensively sampling a system with a thousand degrees of freedom is unfeasible) to the unknown unknowns (go figure).

Despite enormous methodological challenges, this field still assumes that if one has a theoretically exact treatment of the problem and sufficient computational resources the problem is solvable. And yet I am still to see convincing experimental validation backing up these grand claims.

Permalink to Comment

68. A. Postdoc on October 10, 2013 2:34 AM writes...

Seriously, Ant, you forget Martin's predictive work on rhodopsin done by your own postdoc advisor? He predicted how retinal would work before anybody had the structure, even when they got the structure they forgot about Martin & Barry's contributions. Levitt made a predictive model of tRNA before anybody else, using MD to prove it was stable, unless I'm incorrect. Who knows what Warshel did, but if you ask him he did it better and before anybody else and he cited only his own papers to do it.

Curious Wavefunction and Bosco Ho, you guys seriously need to remember what science is. Explaining the past (poorly) and never making a prediction might be good enough for your jobs (what are those again?), but those of us engaged in science have to do both.

I hope to join Ant, John Chodera and others in making real predictive contributions to science, in the grand tradition of today's winners, even though their predictions are often ignored or forgotten.

Permalink to Comment

69. Christophe Verlinde on October 10, 2013 6:21 AM writes...

Only PROSPECTIVE studies can demonstrate the value of computer modeling simulations. Unfortunately, 99% of the publications in the field are about RETROSPECTIVE studies, and sadly they are hyped as if they were PROSPECTIVE.
Yet Karplus, a pupil of Pauling, did not hype his methods, and he spawned most of the current computational simulation experts. Warshel was a grad student of Karplus, Levitt a post-doc, and both are scientific descendants of Schneor Lifson. If Lifson would have been alive in 2013 the Nobel would have gone to Lifson and Karplus.
As to Allinger he is a scientific descendant of Donald Cram. He has been the best in refining force-fields - an extremely valuable, but under-appreciated undertaking, for static properties but has mostly shied away from dynamics.

Permalink to Comment

70. Anonymous on October 10, 2013 6:28 AM writes...

@69: By your argument, shouldn't the prize have gone to Adam and Eve?

Every development in science is made by individuals standing on the shoulders of giants before them.

Permalink to Comment

71. Curious Wavefunction on October 10, 2013 7:17 AM writes...

#68: I mentioned at least three cases where MD was predictive, and if you think explanation in science is not important then I am afraid you don't have a very good idea of how science has been done until now. In any case, I am ready to move on from this discussion since the prize was given out mainly for QM/MM, not MD simulations.

Permalink to Comment

72. Anonymous on October 10, 2013 7:52 AM writes...

#71: Any theory can be developed to explain past data, but that's just the start, and the easy part. A *good* theory must also explain (i.e., predict) future data. That's the tricky bit, and QM/MM often falls short of this basic requirement.

Permalink to Comment

73. newnickname on October 10, 2013 8:33 AM writes...

Misc replies:

(a) Around the same time I was getting the Karplus Eq in undergrad chem lab (Silverstein, Bassler and Morrell), I used Karplus and Porter for P Chem; I still see it some college bookstores as a supplementary text.

(b) @58 DE Shaw: Should Shaw win the Nobel for Chem or Economics? :-)

(c) A well known comp chemist published a paper that showed his model was pretty good (correct ~50% of the time). In "reply", Fritz Menger then published something similar using a random number generator to show that he could predict the same reaction with the same success rate. (I like Menger!)

(d) Predictive utility: Didn't Tack Kuntz have that neat paper in Science or Nature that showed how they used Dock to predict a strong inhibitor; they made the compound and it was a nmolar super-active hit; but the X-ray showed that it had bound in a completely different way to a site nowhere near the predicted binding site! They didn't predict the correct binding but they predicted a "hit"! Darn ... I can't remember the protein or activity.

(e) I think of QM/MM like combi chem. It's something to have in the tool box. Sometimes it can help but it's not the main game.

Permalink to Comment

74. ScienceNerd on October 10, 2013 8:48 AM writes...

#69: "If Lifson would have been alive in 2013 the Nobel would have gone to Lifson and Karplus."

I think this is true.

Permalink to Comment

75. Curious Wavefunction on October 10, 2013 8:53 AM writes...

#72: I agree that a good theory should both explain and predict; what I am saying is that prediction does not always imply understanding, or it may imply partial understanding at best. A great analogy is provided by David Deutsch in his book "The Beginning of Infinity" where he points to a magician doing tricks. After watching the magician for long enough you can predict what will happen next, but this does not mean you understand what's going on behind the curtains and under the hood. You can often predict simply based on pattern recognition or rough rules of thumb without understanding much about the inner workings of a system. Nobody doubts that prediction is very important, but this belief that prediction must necessarily imply good understanding of a system and is the only thing that matters in science is flawed. I am actually quite familiar with that Menger anecdote and I agree that in that case the "well known comp chemist" did come up with a flawed model. I also think that people are far more likely to point to failures rather than successes in fields other than their own. That doesn't mean that we need to stop criticizing the failures, but we also need to be fair, point out the successes and work together to improve the field.

Permalink to Comment

76. Anonymous on October 10, 2013 11:51 AM writes...

#75, Well put!

Permalink to Comment

77. Chemistrybruce on October 10, 2013 1:37 PM writes...

#38 The closest I ever came to a Nobel was building models with Don Cram's old CPK set (shh, I still have a methane).

Permalink to Comment

78. sgcox on October 10, 2013 2:11 PM writes...

What is the state of affairs in a free energy perturbation field ? When I was a student (20 years ago, damn) it was kind of a hot topic. I tried it few times to predict if H->Me will improve Kd for my compounds and gave up. But surely it should be improved since. Curios pubmed search reveals nothing truly exiting.

Permalink to Comment

79. Curious Wavefunction on October 10, 2013 2:19 PM writes...

#78,Short answer: To my knowledge it still has a long way to go. It usually works with very small changes in structure, and how "small" these changes need to be seems to be very idiosyncratic and system-dependent. Some systems give results that are just all over the place. Plus it's still very computationally expensive. However it's at least becoming user friendly now (which also increases the potential for abuse, but oh well).

Permalink to Comment

80. SS on October 10, 2013 4:27 PM writes...

A lot of rant on this site is about the utility of MD. However without force fields we would not be able to predict conformation of molecules reliably. FFs formed the basis for all the computational techniques we use today. The quality of molecular models are lot better today because of robust FFs. Prospective predictions today are not quantitatively accurate yet. With access to GPUs and CUDA versions of MD codes, we may be able to experiment more and add value to predictive ability. Time will only tell! This year's Nobel prize winners definitely led us to scratch our heads. not be able to predict conformation of molecules reliably. FFs formed the basis for all the computational techniques we use today. The quality of molecular models are lot better today because of robust FFs. Prospective predictions today are not quantitatively accurate yet. With access to GPUs and CUDA versions of MD codes, we may be able to experiment more and add value to predictive ability. Time will only tell! This year's Nobel prize winners definitely led us to scratch our heads.

Permalink to Comment

81. sgcox on October 10, 2013 5:06 PM writes...

Wow !

Permalink to Comment

82. ScienceNerd on October 10, 2013 7:36 PM writes...

@78: The field of free energy perturbation (free energy calculations via MD or MC sampling) started in around 1986 (the equations are older, but '86 is about when people thought computing power was sufficient to implement them). The first round of papers suggested all kinds of amazing accuracy with really short simulations for small changes. ~30 years later, and we are, at best, just about where we can actually calculate those same free energy changes with the precision claimed in the original papers. Just about.

Permalink to Comment

83. Neo on October 11, 2013 3:26 AM writes...

#79, #88. so free energy perturbation allows you to predict delta_G of binding of any complex with reasonable accuracy? I find interesting, but not surprising, that no one is providing supporting references with prospective studies...

The Nobel Prize should only have been awarded to MM/QM or MD if these were generally able to provide reliable system explanations leading to accurate predictions. We are anywhere near that. I am personally sad that hype has beaten science.

Permalink to Comment

84. Julien Michel on October 11, 2013 3:57 AM writes...

#83 Neo, I pasted below a couple of references of studies that report a mix of retro/pro-spective predictions. Accuracy varies and there is a lot of work to do to even be able to anticipate when the methods will or won't work well, but it would be unfair to say that the field is unable or unwilling to pursue prospective studies.

Blind Prediction of Charged Ligand Binding Affinities in a Model Binding Site. G. J. Rocklin, S. E. Boyce, M. Fisher, I. Fish, D. L. Mobley, B. K. Shoichet, and K. A. Dill, J. Mol. Biol. (2013) in press

Predicting Ligand Binding Affinity with Alchemical Free Energy Methods in a Polar Model Binding Site S. E. Boyce, D. L. Mobley, G. Rocklin, A. P. Graves, K. A. Dill, B. K. Shoichet. . Mol. Biol. 394: 747-763 (2009).

"Predicting absolute ligand binding free energies to a simple model site," D. L. Mobley, A. P. Graves, J. D. Chodera, A. C. McReynolds, B. K. Shoichet and K. A. Dill, Journal of Molecular Biology 371(4):1118-1134 (2007).

In silico improvement of b3-peptide inhibitors of p53•hDM2 and p53•hDMX Michel, J. ; Harker, E. A. ; Tirado-Rives, J. ; Jorgensen W. L. ; Schepartz, A. J. Am. Chem. Soc., 131 (18), 6356 -6357, 2009

Permalink to Comment

85. Anonymous on October 11, 2013 4:58 AM writes...

Never mind prediction, the real value of modeling is to show that experimental results are wrong most of the time.

Permalink to Comment

86. Anon on October 11, 2013 6:36 AM writes...

I get a chuckle whenever anyone starts the argument that MD has tangible scientific merit. When I hear someone say that, I learn that they are either: A) not a key decision maker in competitive research; or B) have limited ability as a critical thinker.

At best MD is a "nice to have." You could toss MD out the window and there is not a single research program that would blink an eye at its departure.

That's how valuable a research tool it is.

Just my $0.02 cents...

Permalink to Comment

87. A Scientist on October 11, 2013 10:52 AM writes...

I assume all the comments about "experiments being wrong" (#85,55) are mockery toward theory and not serious. Those people should remember that the most famous discovery in molecular biology (the DNA double helix) came about via perceptive choices, motivated by theory, about which experimental "facts" to *ignore*. For the full account see Judson, The Eighth Day of Creation; for a summary that emphasizes this particular point see http://philsci-archive.pitt.edu/3486/1/SSchindler_DNA.pdf.

Permalink to Comment

88. Anon on October 11, 2013 10:55 AM writes...

The DNA double helix discovery was, more than anything else, a triumph of model building. Watson and Crick did no new experiments but cleverly interpreted data from multiple experimental observations using basic theory.
#86: MD was at best a tiny part of this year's Nobel Prize citation.

Permalink to Comment

89. Anthony Nicholls on October 11, 2013 12:40 PM writes...

It's been gratifying to see plenty of posts agreeing with my take on the usefulness of MD- I usually have to assume I'm the (only) crank in the room! And it's been nice to see CuriousWavefunction defend his views- as we are both in Boston we are planning a meet up at the OpenEye offices in a few weeks where I'll put my case against MD more formally and completely and we can debate it all civilized like. I think there is some great work being done with MD by people trying to make it predictive, e.g. Pande, Chodera, Shirts, Mobley but then these are scientists who are the first to question its accuracy. And I'd love to see the promise become reality- it'd make my job much easier! What I mostly object to is the "well, sometimes it works, sometimes it doesn't" attitude. Well, so do the much simpler QSAR, docking, 2D similarity, pharmacophores etc. And you don't see them getting a Nobel Prize (don't start!).

Some brief comments on comments:

Thanks to Marcel (#32) for pointing out the award is for QM/MM- this made me less angry and simply more perplexed. I haven't followed QM/MM much so perhaps it's made some great strides when I wasn't looking. I am completely unaware of any, as far as I can tell much of enzyme catalysis is still 'understood' by waving of hands and lots of talk about 'key protons'.

I'd have agreed with CW (#39) about the importance of 'explanation" or "understanding" until I read Bishop and Trout's "50 Years of Successful Predictive Modeling Should be Enough: Lessons for the Philosophy of Science. Philosophy of Science 68". (Thanks, Kim!) Now I think appeals to 'understanding' may be seriously flawed, but it's a deep topic.

Thanks #68! Though you are wrong on all counts. In talking to Barry about his work with Martin he admitted that actually they got the right answer (shifts in visual pigment) for completely the wrong reasons. As for Levitt and tRNA, it is impossible he used 'native' MD to get the structure without essentially biasing to the known answer and he could not have made it stable without an ansatz such as turning off all the charges. As he predated Darden's game-changing Ewald sums there is no way tRNA was stable under simulation- even proteins were not stable to simulation back then.

As for the references given by Julien in #84, these are mostly the same piece of work by Shoichet. When he first presented them at an ACS meeting I noticed that essentially he had put aside a fraction of his results to reach his final conclusion that MD was being predictive. With all the data it was a coin-flip as to whether a compound was more or less active. When I challenged him after his talk on this he said, and I am not making this up, "That's true- but I wanted to be encouraging to the simulationists". Poor lambs. All that grant money to play on big computers apparently isn't enough encouragement. And really, 25+ years, thousands of papers, conservatively over a billion dollars of government research dollars and that's it?

I entirely agree with John's comment (#53) that there is still a prize for an actually predictive theory- in fact, perhaps a silver lining of this farce is that it may open the door to such. Binding energies would deserve one. Protein folding will probably now get one (Pande, Shaw, even Baker). Force-fields (Allinger, Jorgensen, Hagler, ever a forgotten Lifson student) ought to have had one by now. So perhaps Derek is right- yay for the field, if for the wrong reasons.

Permalink to Comment

90. ScienceNerd on October 11, 2013 8:58 PM writes...

@89: I'm surprised you give a pat on the back to those you assert are tying to "make MD predictive" because, you assert "they question its accuracy." Funny, having worked with many of the "names" in the field and aware of the work of all the others, I have seen that most are fully aware of the limitations of MD. I don't think that the labs you cite have a monopoly on that, and in fact I'd argue the opposite: That most labs that use _and develop_ MD techniques (a list much longer than the one you site) are completely aware of when MD might work, when it won't, and what kinds of work is necessary to try to push the field forward.

I would add that there isn't a name either on your list, or on the canonical list of titans of the field, who hasn't been associated with MD work that, in retrospect, probably reflects some element of fortuity. But that's how science works, and to expect otherwise--to expect that a fully elaborated protocol for MD with a fully validated force field and a compute engine sufficient to push us into the simulation realm appropriate for the questions we are interested in to appear full grown from the head of Zeuss before any of this appears in the literature--is beyond silly. The progression of science is a series of zigs and zags, apparent successes later revealed to be failures and other apparent failures that lead to major paradigm shifts.

Name a macromolecular theory that passes your test of being prospectively predictive, applicable to systems of practical interest, and (presumably) fully passing the coin flip test on a consistent prospective basis that doesn't have considerable antecedents in the literature that are somehow (or fully) flawed.

"We can do better" is hardly an argument for dismissing the "cretins" who laid the groundwork for the field. And I would suggest that without said groundwork, the company you run would have precious little to sell, as many (perhaps most) of your products reflect ideas that were previously discussed in the literature, even if you have added your own twists.

There are few bigger scientific skeptics than myself. But I also realize that you have to step back and look at the bigger picture, and watch the gradual progress. NONE of us is immune to occasional scientific folly, and I would dare say that a lot of the names that you besmirch in your rant either by name or by exclusion have contributed dearly to the progress of science.

Permalink to Comment

91. A. Postdoc on October 12, 2013 12:07 AM writes...

@Ant, I love being wrong! It is great to see Barry (though perhaps not Martin) admit their paper was wrong. And a big fault of MD is certainly that the researchers using it are biased.

As for this: Protein folding will probably now get one (Pande, Shaw, even Baker).

I'd say none of those people deserve it. Pande has just now started to admit that Walter Englander's foldon ideas were right, only because after decades of simulations and billions of PS3 hours he got the same answers as Walter got years ago with experiments. Shaw has contributed nothing but even more money to make computers run biased and hopelessly misplaced MD faster. I'm not sure what Baker has done, his latest Nature paper is a crock, he 'rationally designed' proteins to bind ligands, but he needed lots of screening of millions of proteins to find a few that actually bound.

Permalink to Comment

92. Gabriel Rocklin on October 12, 2013 1:08 AM writes...

@90: My read of Tinberg/Baker et al. is that paragraphs 3-4 state that a 12 uM binder (Figure 2) was identified straight out of the computer as one of 17 (paragraph 3) designs tested. Further affinity maturation improved the affinity and selectivity. Am I missing something, or is your description that testing "millions and millions" of designs was necessary to find a binder mistaken?

Permalink to Comment

93. Gabriel Rocklin on October 12, 2013 1:11 AM writes...

above comment was directed at #91, not #90 obviously.

Permalink to Comment

94. Chemcat on October 12, 2013 11:52 AM writes...

@60: I agree. The amount of mudslinging and vulgarity found in some comments of this post is not what I have come to expect from this blog. Here's to hoping that what's "in the pipeline" for future comments sections will promote spirited and respectful conversation!

Permalink to Comment

95. sgcox on October 14, 2013 2:50 PM writes...

#94 & #60
This is simply because it touched some raw nerves. If you look closely, all (well, almost all) comments came from posters who are clearly very well qualified and informed in the field. There are simply no trolls here. My personal, albeit mild, hostility to the computational chemistry is based on a personal experience of testing hundreds and hundreds of "rationally deigned" inhibitors which turned up turds. And of course, eventually one will come out true by the shear number of fails and get trumped up as a proof of a genius of the computer person who run essentially a random number generator but still got all the credits in the end. Please, tell me you did not see it before.

Permalink to Comment

96. Joaquin Barroso on October 14, 2013 6:46 PM writes...

Wow! what a lengthy discussion! By construction its almost impossible for any given prize to be 'fair' and 'comprehensive' at the same time; and by human nature it is flawed to please most people. The Nobel prize is often regarded as the highest honor a scientist could get (lets be honest, who among us during undergrad school never dreamed of getting one?) yet it gets bashed year after year by scientist themselves! This is called the Nobel prize because its awarded by the Nobel committee based on some criteria of their own, there is no universal criteria or metric entirely objective, just as there is no exact density functional despite the fact that its existence is proved by Hohenberg and Kohn (bashing in 1, 2,...) .
Anyway, I for one feel excited every year to find out who the Nobel laureate is because it makes me spontaneously go and read about their work and thus learn some more chemistry; and this year in particular, I feel glad that the field, as Derek wrote, gets some recognition through -some- of its most prominent developers.
Now about MD being a science or not... I will write a post on my blog soon...

Permalink to Comment

97. Anon on October 15, 2013 1:08 AM writes...

#91, A Postdoc: "I'm not sure what Baker has done".

I think that statement tells us how well-qualified you are to comment on the state of the field.

Permalink to Comment

98. Li on October 17, 2013 1:18 AM writes...

I see a strong similarity here between many of the apologists and those who justify police bringing in (and of course paying for) psychics to help with missing child cases. Just sayin' - it works sometimes, so it must be science... Mustn't it?

Permalink to Comment

99. OSDDlinux on October 20, 2013 11:40 AM writes...

I think trend is changing world wide, in 2008 first time in India S. S. Bhatnagar Award (Biological science) goes to a bioinfomatician (G P S Raghava) http://en.wikipedia.org/wiki/Gajendra_Pal_Singh_Raghava
. This is welcome change in attitude, both experimental & theoretical science are equally important.

Permalink to Comment

100. surajjena on October 24, 2013 8:26 AM writes...

Curiosity waves function & proteins folding way quantum biology still it asked question how it is evolved in computational stimulating scheme of astro computational biotech.

Permalink to Comment

101. surajjena on October 24, 2013 8:27 AM writes...

Curiosity waves function & proteins folding way quantum biology still it asked question how it is evolved in computational stimulating scheme of astro computational biotech.

Permalink to Comment

102. Anon on October 28, 2013 9:53 PM writes...

#98: Except that psychics don't work and there's no causal connection.

Permalink to Comment

103. Frederic on October 29, 2013 10:13 AM writes...

I am amazed that experimentalists are usually more critical with simulations that theoreticians are critical with experiments. However, theory and simulation are very helpful to rationalize what sometimes look like nonsense explanations of nature's laws by people who directly deal with them in the lab and not through numbers and equations. Providing explanations and questioning assumptions already proves the utility of the simulation approach, with all its limitations.

I also would like to mention that getting a Nobel prize is somehow a matter of chance and possibly a reward for scientists communication skills. I see at least one person who has had key contributions to deserve the prize: Wilfred van Gunsteren in Zurich. Other possible European whose name has not appeared yet on this page: Michele Parrinello in Zurich, too. These two persons have developed powerful free energy sampling methods and coupling tools which are routinely used in computational bio-physics/chemistry groups.

Permalink to Comment

POST A COMMENT




Remember Me?



EMAIL THIS ENTRY TO A FRIEND

Email this entry to:

Your email address:

Message (optional):




RELATED ENTRIES
The Worst Seminar
Conference in Basel
Messed-Up Clinical Studies: A First-Hand Report
Pharma and Ebola
Lilly Steps In for AstraZeneca's Secretase Inhibitor
Update on Alnylam (And the Direction of Things to Come)
There Must Have Been Multiple Chances to Catch This
Weirdly, Tramadol Is Not a Natural Product After All