AI and Machine Learning Has a Bright Future in Chemistry

You are here:

Group Leader Andrew Jordan is a key member of Charnwood Discovery’s sustainability working group.

Being invited to a meeting, like AI4Green at the University of Nottingham’s GSK Carbon Neutral Laboratory for Sustainable Chemistry, enables us to ensure that we are keeping abreast of the latest developments in green and sustainable chemistry.

Here is what Andrew has to say about his learnings from the meeting:

It was fantastic to be invited back to the University of Nottingham. I have many fond memories of my postdoctoral research here. During the morning session we saw some great discussion on how artificial intelligence and machine learning might be incorporated into an electronic lab notebook (ELN) system and whether this implementation could improve reaction sustainability and safety by suggesting alternative solvents and reagents. The general consensus seems to be that an ELN system that is constantly learning and adapting would be a hugely powerful resource and could guide the next generation of chemists towards safer and more sustainable chemistry before they have even entered the lab.

Later in the day there were some great presentations from Richard Bourne, Jonathan Goodman, and Claire Adjiman. Richard presented some fantastic results from machine learning self-optimising flow reactions: a technique that can allow for two weeks’ conventional work to be carried out in just 24 hours. Somewhat controversially, he believes that 90% of all reactions will be optimised this way within 19 years. It’s hard to argue with the concept when you see these kind of results presented! (you can read more here)

Jonathan gave some really interesting insights into data generation and how we as chemists are now working in a very “data rich” environment. In the past, there was always the go-to excuse of “we haven’t got enough data”, when asked to solve a particularly challenging problem or model the behaviour of chemical systems. However, chemists working in the 21st century have never had access to more data than we do now. The real challenge is trying to process and derive meaning and understanding from the swathes of information available. To answer some of these challenges Jonathan has developed excellent computational and modelling techniques. One such example is the tool BINOPtimal, which is a free online tool that can aid in the prediction of stereochemical outcomes of many acid catalysed processes. Definitely a tool I will be adding to the toolbox!

Finally, Claire who has a passion for modelling and optimisation methods, took us through a variety of case studies in which advanced modelling techniques were applied to real world problems. They ranged from optimising carbon capture from gas streams, solving the issues of HCN generation during amide coupling, ternary phase diagrams for compound solubility, and recrystallization techniques. I’ll be sure to look further into her work on recrystallization optimisation as computer modelling at your desk can often help save many hours at the bench. (read the article here)

Overall, I can say that the day was an excellent tour de force of how computational methods can greatly assist chemists in their day-to-day research and development activities. Adoption of these methods will only serve to strengthen any research group or department. Plus, the judicious use of more advanced techniques, such as AI and machine learning, has a bright future in enabling safer and more sustainable chemistry.

Andrew Jordan

Andrew Jordan

Group Leader, Chemical Development