The race to build a robo-chemist

Over the past year a clutch of startups have talked up their ability to automate aspects of R&D by matching software to robotic technologies. But the ambitions of some groups go well beyond these examples, with researchers aiming to create robo-chemists.

This week a Nature feature looked at the progress of some of these groups and the obstacles that stand between them and their goal of building chemical synthesis machines. When presented with a chemical structure such a machine could figure out how to synthesize it and then carry out the steps. If researchers are to realize this vision they must find solutions for three separate, difficult tasks--scan a database, plan the synthesis and make the molecule--and then make the resulting systems work in unison.

The consensus is that some, if not all, of these three steps need work before even a rudimentary synthesis machine is viable. "The hardware has always been there, but the software and data have let it down," GlaxoSmithKline ($GSK) automation researcher Yuichi Tateno told Nature. Tateno is also a member of Dial-a-Molecule, a 5-year, 60-company collaboration in Britain that is researching synthesis machines. Dial-a-Molecule has spent the past few years trying to figure out what would be needed to build a robo-chemist.

While Tateno is optimistic that the hardware capabilities are in place, In the Pipeline's Derek Lowe thinks existing systems could come unstuck when presented with the sort of steps a synthesis machine-grade algorithm would churn out. "The kinds of reactions that useful software would suggest will be things that no one has ever tried to automate, and the range of reagents that can be accommodated by the existing hardware may well cripple the software algorithms before they even get off the floor," Lowe wrote.

- read the Nature feature
- here's Lowe's take