Computers and the Future of Chemistry

I know this blog has been dormant for a while…thought I’d get back in business with a few thoughts on today’s post over at In The Pipeline about a review in the journal Angewandte Chemie. The review and the post at In the Pipeline discuss the current state (and possible future) of computer algorithms that devise synthetic routes for organic (carbon-based) molecules. At this time there is no commercially available software that can reliably figure out a way to make any structure you give it. But the software is getting better, and it’s not unthinkable to imagine that at some point in the future, computers might even rival trained synthetic chemists in their ability to devise synthetic routes for drugs and other compounds of interest. One possible analogy is the progress chess and Go computer algorithms have made over time — they have steadily evolved from simple beginnings to the point where they can beat the world’s best players. Writing a computer program that could plan an organic synthesis is a far more complex problem than writing a computer program to play chess, although the Angewandte Chemie review at least argues the software is getting better and someday this problem also might be solved.

For me, the most interesting aspect of this is the ramifications — what this means or at least could mean for the future of chemistry (again, assuming that a reliable synthesis algorithm are possible — I’m just going to treat that assumption as a given for now, although it’s open to debate). It does not mean, I think, that computers will someday replace synthetic chemists. You’ll always need an experienced practitioner who can check the route proposed by the algorithm to make sure it’s something that would actually work — in other words, the computers will always need supervision. The software is going to be a tool, not a substitute for thought. For another, even if your software can suggest a practical synthetic route it can’t actually execute it. Organic synthesis employs such a wide variety of reactions and reagents it would be very difficult to automate completely.

What it does suggest at least to me, however, is that the future of chemistry may be different from its past in some very important ways. I don’t think it’s unfair to say organic synthesis has occupied a prominent place in the history of chemistry. Some of the 20th century’s most admired chemists were people like Robert Woodward, the man who devised a 51-step route for the synthesis of chlorophyll and routes for the synthesis of vitamin B-12, strychnine and colchicine, among others. The synthetic routes he devised for chlorophyll and b12 never saw widespread use — at the end of the  day, it’s cheaper and easier to get plants to make chlorophyll for you, at least if the alternative is a synthesis in 51 steps. But that’s aside from the point. Woodward proved you could synthesize molecules no one had believed could be synthesized by humans — a hideous natural product like vitamin B-12 among others. In the process, Woodward and others like him discovered synthetic reactions that other chemists could use to make valuable products or lifesaving drugs.

Decades after Woodward, there are still a reasonable number of academic labs that work on total synthesis of huge, ugly-looking natural products, although this area of research seems to be growing less fashionable. There are several reasons why: funding for this kind of thing is less abundant now, and the list of huge ugly-looking natural products that no one has as yet found a way to synthesize isn’t as long as it used to be. And thanks to Woodward and Corey and others like them, chemists have gotten much better at figuring out ways to synthesize complex organic compounds. Sure, there are still plenty of tough problems that take a lot of work to solve; chemistry hasn’t quite reached the point where we can confidently say, “I don’t care what it looks like, if you can draw it, you can make it”, although we’re getting there. Frankly, a lot of the time the question is about how to make something most efficiently. But the real dilemma facing modern chemistry is not how to make things — synthetic chemists can nearly always figure out how to do that sooner or later, even if the “sooner or later” part involves a lot of hard work. The real dilemma facing modern chemistry is how do we know what to make? because in that direction lies the poorly explored territory.

The best way to illustrate this is to look at drug discovery, the area where this problem is most painfully apparent, although this gets complicated because drug discovery of course is about how to find a molecule that alters a biological system, and there’s still a lot we have yet to learn about biological systems. But I think the thing that people outside of drug discovery would find most surprising is just how much trial-and-error modern drug discovery involves.

Take for example one approach to discovering new drugs, high-throughput screening (HTS): throw all kinds of molecules, a whole library of compounds, at a target and see what sticks. (It’s a little more complicated than that, yes, but that’s the essence of HTS in a nutshell). Once you’ve found something that does, modify it through trial and error and educated guesswork and test the modified versions you make until you find something that works the way you want in a test tube and in animals. This is trial and error every step of the way. Educated trial and error, trial and error informed by careful experiments and meticulous planning, but trial and error nonetheless. And it involves problems no amount of synthetic prowess will solve. The question facing drug discovery teams is not how to make the molecules they want, but how to know which ones are worth making, and why biology gives them so many nasty surprises along the way.

So my two cents is this. I think the evolution of synthesis software, assuming it continues, will reinforce a growing trend that will continue in chemistry, a shift in emphasis away from synthesis and towards other problems. Does that mean we no longer have a need for research in organic synthesis and total synthesis of natural products? Of course not. But I suspect chemistry research in the future will focus less on synthesis and more on properties as a function of structure — especially biological properties. Because let’s not forget, biology at the end of the day is chemistry too. Complex chemistry, perhaps, but chemistry nonetheless.

8 thoughts on “Computers and the Future of Chemistry

  1. Great article! It reminded me of a seminar I was at a year or two ago. It was a researcher from McGill University talking about his work on software that could predict active drug compounds. Essentially, what you’re saying is still missing from the field. I wonder how effective combining these two programs could be.

    Here’s his page.

  2. I work for Wiley and we’ve just launched ChemPlanner (, which is commercially available and is already being used by big pharma companies (happy to offer a demo and trial access if you’re interested!). ChemPlanner is a computer-aided synthesis design (CASD) tool similar to the one described in the Angewandte review. It predicts reactions and complete synthetic routes, even if they haven’t been reported before in the literature. Video explainer:

    Also, I agree with the comment that chemists won’t be replaced, and nor should we be trying to do so. The point is to develop tools to enable more extensive and creative exploration of chemical space, so you’re right to say that technology will be key in reducing the amount of trial and error involved from lead generation to synthesis.

  3. great to see you writing again you always hit on great key points or educate with a fun history story. Keep up the cool work 🙂

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s