How do chemists design new molecules?

09 April 2019


this is a picture of the periodic table



How do scientists go about making new molecules?


Sam wanted to know how we make new compounds and molecules, so Chris Smith put the question to chemist Lee Cronin...

Lee - Okay. There are three ways of going about it. So if you think about chemists they're a bit like architects when designing say a house, but we're restricted by the bonds that we can use, and these bonds are how the atoms get held in a given molecule. The first one is if we know roughly what the molecule looks like, say it’s already existing drug and antibiotic, what we can do is draw that up and then add on the other parts you want. Say you want an extension onto the house, as it were, we can extend on some other atoms and what we would then do, we would then draw that molecule on a computer and then the organic chemist would then -what we would call disconnect it - we would take it apart on certain ways and there are some really good rules we could use. And then go to the laboratory and actually knock that molecule together in a flask, so you actually do reactions to join those bonds up in order and then purify each step. And then, hey presto, you should get a molecule. That's if you know where to start.

The second way I can think of is by trial and error. We just make a whole load of molecules and check what they do. Are they good electrolytes for batteries, or are they good drugs, or are they good pigments? And then go from the property. The other way is to use a computer and do some physics and then say right, I want a molecule that has say I'm going to make an LED with this molecule, so it's going to emit light. So I'd work out what colour that would be and then I would look at how far the electrons have to move and then go backwards. And then the computer would literally spit out a blueprint. The fascinating thing, this thing called inverse design this is really the cutting-edge. Chemists are really beginning to dream molecules in the computer and then take it through the process of molecular architecture.

Chris - But you were in the news in the last year because you also made a robot that could do this and learn as it did it so you could work out from its own mistakes and what did and didn't work; how to make better molecules better make them faster?

Lee - Yeah. I was trying to be coy about selling my own stuff. But in my own lab...

Chris - I'm not just planting that, that wasn't a planted question. I did actually read the paper that you published on this and it seems like it would save enormous amounts of chemists, enormous amounts of frustration and hair loss trying to go down blind endings. Because of course, we are only to make these things because to have those rules to make these molecules because people have relentlessly pursued all these avenues to work out what does and doesn't work?

Lee - So that kind of trial and error method of making molecules step-by-step used to be done by hand and it was kind of an artisan process. And what we've done in my lab is written a programming language that can deploy those steps as you need - a bit like how you would write some software now. And then what we do is reversion control so the idea is that we can send a code to other people to introduce on demand.

Chris - And that means you get precisely the right conditions to get the right molecule?

Lee - Yes.


Add a comment