Computational Chemistry, Experiment, and History

(McBride's Chemical Philosophy)

1) The future of experimentation (N.Y. Times)
2) Will computation replace experiment?
3) Chemical humility
4) Relevance of chemical history
5) The $64 Question

Modern molecular mechanics or quantum mechanics packages like Spartan or ChemBioOffice, can gain unwarranted authority from the allure of computer graphics. Consider the following portion of a news story from The New York Times (October 8, 1996):

. . . Indeed, in some ways the new approach [computation] is more efficient [than experimentation]. The pharmaceutical industry, for instance, has found a way to make and test in a given week thousands of chemical compounds as potential prescription drugs, a far cry from the two or three a day that was the upper limit only a short time ago. That's the focus of scientists at Merck like Paula Fitzgerald, a biophysicist.

Occupying one of the many narrow offices off a long, gray corridor at Merck's laboratory in Rahway, N.J., Dr. Fitzgerald models the interaction of enzymes indentified by university scientists and chemical compounds generated by Merck. She is seeking to determine what mixture might block a disease-causing enzyme. The enzyme and chemicals appear as blue, red and green filaments floating intricately in three-dimensional depth on her ink-black computer screen.

"You draw the compounds that you model and test from a computerized library," she said. "Before you had to actually mix the chemicals to see how a compound worked."

Whatever you think of the Times's colorful writing style, and regardless of whether Dr. Fitzgerald was accurately quoted, there is the crucial question of whether computation and graphics should serve as helpmates to experimental studies, suggesting profitable new leads and helping to interpret results, or whether, as suggested in this article, the computer has replaced, or is about to replace, experiment.

In fact computational chemistry has made such strides, progressing from calculating molecular structure, to simulating reaction paths, to including consideration of surrounding solvent, that its practitioners often refer to their calculations as "experiments".

Do you want to know what I think? I think the calculations most worth doing are those that inform our intuition and make us see chemistry in a new way. Those that just provide numbers for calculating an equilibrium constant or a rate can be very useful, but they don't interest me as much. I think that enormous strides are being made in computational chemistry, both quantum mechanical and molecular mechanical, and of course computers are getting more and more powerful. I can easily imagine that within 10 or 20 years there will exist virtual reality models that will allow well-heeled students and researchers to don a pair of glasses and mechanical gloves that will make them think that they are handling molecules that have proper resistance to deformation, and even proper response to HOMO/LUMO mixing between molecules that are "reacting" (this will require quantum mechanical, not just molecular mechanical computation). Including all of the surrounding molecules for a reaction in a condensed phase (liquid or solid) will be a lot tougher, but it's just a matter of time. Such "calculations" will certainly revolutionize our organic chemical intuition (less futuristic calculations are already doing so).

Do I think that such phenomenal models will replace the need to "actually mix the chemicals to see how a compound worked" ? No Way!

I think the day that we abandon working with the real thing is the day we stop learning what is really interesting and new. It would be a sign of naïve chemical hubris.

---

My viewpoint on the need for a measure of chemical humility can be illustrated by a true story told by Yale professor Harry H. Wasserman:

In 1945 Wasserman had just returned from service in the army during World War II and had resumed graduate research in organic chemistry at Harvard under the legendary Robert Burns Woodward, a chemical prodigy, now 28 years old, who would become, in the view of most, the greatest organic chemist of the 20th Century. Woodward would win the Nobel Prize in Chemistry in 1965. He certainly would have shared the 1981 prize had he not died prematurely two years earlier. Woodward was not a person inclined to don a cloak of false modesty.

Wasserman's project concerned the synthesis of artificial penicillin, and at one juncture he needed to convert S-benzylpenicillamine, an incredibly precious compound of which Eli Lilly had given Woodward 25 g, into a 5,5 fused-ring thiazolidinelactam. Wasserman had tried the reaction on a 50 mg sample, and it worked like a charm. So Woodward suggested that Wasserman scale the reaction up over the weekend. On Saturday he confidently scaled up by a factor of 200, reacting 10 g of the S-benzylpenicillamine, almost half of the remaining supply. Instead of colorless crystals, he recovered only a deep yellow, intractable oil.

He was devastated and conflicted. Should he wait until Woodward came to lab, or should he bite the bullet and telephone to announce that his naïve blunder in using so much material had resulted in a catastrophe? No one who has known Prof. Wasserman would be surprised that he chose the proper course and decided to telephone.

Reaching Prof. Woodward at home, Wasserman reported that he had attempted the reaction on a larger scale, but that it was an utter failure, and that precious starting material had been destroyed.

Woodward asked, "How much of the S-benzylpenicillamine did you use?" Wasserman gulped and answered, "10 grams."

There was a pause. Then Woodward admonished, "Ten grams! Harry! Even I don't know all the things that could go wrong in that kind of reaction."

If Woodward couldn't imagine what all could go wrong, I doubt that a computer algorithm will ever be able to do so. I think we must remain firmly rooted in real experiments.

Some students asked after class whether I wish I had lived in the 19th Century, when so much was happening in organic chemistry. Of course not. No doubt more is happening now, and the development of computational chemistry is only one example. Besides, the 19th Century lasted a long time. It would be a great place to visit, but I wouldn't want to live there.

It is fun and valuable to study the chemistry of that period, because it gives deeper insight into our own chemistry:

- We can see reports of careful experimental observations that are not colored by our current fashions and theoretical prejudices. This can suggest important experiments that still need doing.

- From the standpoint of current knowledge we can see which attitudes were fruitful and which weren't for the development of science.

- We can see how the smartest, hardest-working people could go wrong and fail to understand things that seem really obvious in retrospect, or, if they did understand, couldn't communicate, or were squelched.

- We can see that the big shots almost always thought what they were doing was the most important, and that typically they thought they had THE answer and needed only to convince everyone else to do what they wanted them to do in order to fill in the few remaining gaps in a comprehensive view of nature. Often it was young, independent nobodies who led science into the future.

Studying the past gives us perspective on the present and keeps our thinking fresh.

Finally, here's an interesting Question:

In holding an only limitedly optimistic view of the prospects for computational chemistry vis-à-vis experiment, am I being a latter day Kolbe? Do I just lack imagination, as Kolbe did so obviously (in retrospect)?