Unrest is brewing in the world of chemistry, because a piece of research is raising eyebrows, and hackles, of those who think that it represents possibly dishonest science. The case also showcases the dilemmas in modern scientific progress, which have been illustrated before by some other famous cases. I think it would be worth it for me to elucidate the nature of the current entertaining fracas. But after a very short introduction to the art and science of organic synthesis.
* Magic in moleculeland and showdown in the house of hexacyclinol:
'Total synthesis', the multistep synthesis of complex organic molecules with practical and medicinal benefits, has been one of the cornerstones of the scientific foundation of the modern world. Look everywhere around you, and you see materials that have been manufactured tediously by organic chemists, one step and one bond at a time, since many decades. The greatest impact of this endeavor has been in the pharmaceutical industry, and most of today's important drugs would not exist if it were not for the patience and ingenuity of the organic chemist, nor could we hope to get future pharmaceutical products. Without synthesis of natural and artificial molecules, we would essentially be losers in our fight against disease.
So much for the practical aspect. The other reason why organic chemists synthesize molecules is for the sheer intellectual challenge. Building a complex molecule is like building the Eiffel tower or like painting the Sistine chapel, where not only does every bit and piece have to be put in its exact place, guided by known chemical principles, but the result, efficiency, and methods also have to be aesthetically pleasing. Many chemists are drawn to the synthesis of especially natural molecules, because they are dazzled by the beauty of nature's architectures, and want to make every effort to top nature in her magnificent constructions. Many have come close, if not surpassed, the rich and astonishing diversity of nature's creations.
So did James La Clair of the Xenobe Research Institute (?) want to synthesize a molecule called hexacyclinol, whose structural beauty and complexity only seasoned organic chemists can truly appreciate. It is a metabolite from a fungus isolated a few years ago, from a dead piece of wood in Siberia. Many of the top selling drugs of recent years have incidentally been based on molecules isolated from such obscure terrestrial and marine organisms in exotic locations, a resounding case for preserving biodiversity.
A couple of months ago, his synthesis appeared in one of the two top chemistry journals in the world. I am not a seasoned organic chemist, but when I saw the structure and synthesis, my first reaction was 'Wow'. But further reading of the paper made my jaw drop lower down. The synthesis was 30 steps, something not uncommon in today's synthetic protocols, although still a formidable feat. What was dazzling was the fact that authoring the paper was one man- La Clair. Today's syntheses are massive endeavors, usually involving at least four to five graduate students and post-docs who have toiled for months, if not years, on such a complex product. One man's crusade in synthesizing such a molecule would have been hard to believe even in the maverick days of the 30s, when rebels determined to overturn sacred cows could toil obsessively in their laboratories. Frankly though, I was not completely qualified to judge the synthesis, but I was astonished at the fact that one man had done all the work of procuring, synthesizing, and then characterising this gigantic succesion of molecules in the synthesis. Suddenly I remembered having seen La Clair in a session from the annual meeting of the American Chemical Society. I remembered his clever remarks in the session which he had chaired, which involved praising my friend's synthesis and saying that 'my friend is ahead of the bug that usually synthesizes the molecule by several steps'.
My astonishment and puzzlement at the solitary achievement of La Clair was justified when I came across Dylan's Tenderblog, where he pointed out the dubious nature of some of the steps and statements in the paper, followed by a world-class barrage of invective, hilarious comments, and astute observations. I was ROFL when I was reading these comments.
However, now, it seems that La Clair may have become A La Carte indeed. A chemist by the name of Scott Rychnovsky at the University of California Irvine, predicted that some structural data for the molecule actually matched the calculated data for a totally different molecule.
Note that the molecule which La Clair synthesized already existed. Its structural data was already known, and a structure had already been deduced from that data. What Rychnovsky did was use powerful computers and the methods of computational chemistry (my turf!) to conversely calculate the structural data from the molecule. What he got did not correspond to the observed data. Instead, he came up with a totally different structure which would correspond to the observed data. In a mammoth effort now, John Porco of Boston University has actually synthesized the alternative structure which according to Rychnovsky, should correspond to the observed data. Voila! It does.
Now what does this mean?? For one thing, it could mean that La Clair synthesized the wrong molecule. But remember, Rychnovsky's structural data calculated for La Clair's structure does not actually correspond to that structure. However, in his paper, La Clair has done the usual routine of presenting the experimental structural data for his molecule, comparing it to the original experimental data acquired when the molecule was isolated from its fungus, and then noting the exact correspondence between them, concluding that his structure is the same as the original one. This is standard and age-old scientific protocol; come up with something, then see if the data for that something matches the known data. If it does, you do have what you say you have. But again, according to Rychnovsky, La Clair's structure should NOT give the structural data which La Clair has presented. So there's a disconnect between La Clair's structure and its synthesis, and La Clair's structural data. Now, the structural data already exists (from the original isolation study) and so cannot be fabricated. Thus, by the rules of Aristotelian logic, that imperfectly perfect science, the fault lies with La Clair's synthesis. Ergo, the skeptics conclude, La Clair could not have synthesized the molecule which he claims he did. Ergo, La Clair's synthesis is not what he says it is.
Ergo, La Clair has committed scientific fraud, or that's what they are saying at least.
La Clair claims that most of his work was done in an institute named Bionic Bros' in Germany. This name sounds to me like something from an Asimov novel, a cross between a robotics company and a bagel bakery (I am thinking of Einstein Bros. of course) where an obscure genius toiling in an obscure institute with a funny name, comes up with a breakthrough to create artificial life or something similar.
Supporting the skeptics' conclusion is the dubious nature of some of La Clair's statements and experimental steps, which I would leave an experienced organic chemist to pontificate on.
To be frank, strictly speaking, the verdict is still out on the La Clair affair. La Clair himself says now that the structural data for both the molecules could be exactly the same. To my humble chemist's mind, this seems highly unlikely, given the very different structures of the two. But as it is the case, whether I or La Clair or the critics are wrong in this case, science will progress either way. No offense to La Clair. If he is right, we will be wiser anyway.
That's the good thing about doing science. Whatever happens, science always wins.
* I replicate, therefore I am:
But this case is illustrating some of the inherent problems of scientific peer review. Scientific results should crucially be testable. But who is going to go to the trouble of testing a 30 step synthesis, or any such mammoth endeavor? And mind you, dozens of such syntheses are published every month. All the reviewers can do is check for internal consistency and past conformity based on their own knowledge and experience. Nature has recently published a nice article, narrating the problems with the all-important replication of data that is paramount in the scientific method. How can you check each and everything in a paper? Even if you can, does failure to replicate mean shoddy work on your own part, or a fundamental problem in the original author's work? And more importantly as the article points out, scientific research has subtle details in the exact protocol, including elements introduced by the skill (or lack of) the experimenter. Such elements can hardly be evaluated, and are never mentioned in a paper. What if these subtleties are playing a large role in the results?
The Nature author recounts the efforts of several journals now going to be devoted to methods, as well as websites where readers can rank papers based on their own efforts in duplicating the data from those papers. As the author says though, this could have the adverse effect of having to include too many similar pieces of work, a fact that may make the journal less attractive for authors and readers alike.
Another quite different matter concerns the nagging question; Why do they do it? Is their behaviour an inevitable consequence of today's cut throat world of competition in science, the high-pressure world of publish-or-perish that is so emotionally taxing, that sometimes scientists just lose it a little and falsify their results with the hope that they won't be detected. In La Clair's case, hexacyclinol was one example among a dozen other interesting examples. But what about the celebrated case of Woo-Suk Hwang, the Korean celebrity, the scientist who became the lifeblood and then the pariah of his ilk with his work on stem cells and cloning. Surely he could not have assumed that crucial work such as his would not be subjected to the closest scrutiny possible. That scientists would not keep the midnight oil burning in their laboratories to validate his results. How can someone who does work of that calibre afford to be dishonest to any degree and think he would get away with it? What about another celebrated case, that of Jan Hendrik Schon, the trailblazing Bell Labs researcher who promised to revolutionise electronics, semiconductor and superconductor technology. Not in one but in sixteen papers did he duplicate the exact same graph. How could he think he would escape unscathed and in fact be lionized? The question defies explanation. However, if any kind of serious study found out that it is emotional strain and pressure that brings about such behaviour, then does our entire way of doing science deserve a second and serious look? Or do we just dismiss these few cases as bad apples?
I believe that psychologists should find this avenue of investigation a very fertile paradigm of study. In the 1930s, a publicised and comprehensive psychological study tried to document what kind of men and women become scientists. What is their personality, their origins, childhood influences that turned them toward science and inquiry? Maybe the time is ripe for a similar study asking a different question: What are the adverse effects of the framework of modern scientific research and peer review?
But in the end, I think that the Nature author says it best when he points out that the marvelous expositor of science Nobel laureate Peter Medawar once pointed out that all scientific papers are frauds, because they paint a false picture of science progressing in an orderly fashion from hypothesis, testing, confirmation, to theory or law! As the great philosopher Paul Feyerabend pointed out to the chagrin of others, science is an anarchic enterprise. As the Nature article enumerates, since the actual process of science is much more messier than is depicted, maybe the evaluation of the process should also be more messier. I agree. Let evaluation also be an anarchic enterprise.
The more the messier. The messier the merrier.
Idiosyncratic Thinking: a computer heuristics lecture
6 hours ago in Doc Madhattan