According to the science magazine Nature, chemists are heralding a new artificial intelligence platform as a significant milestone. The platform has the potential to accelerate the process of drug discovery, and it should be able to make organic chemistry more efficient.
The new platform is designed to help chemists to plan the syntheses of small organic molecules. Traditionally, chemists use the process of retrosynthesis, which is an established problem-solving technique whereby target molecules are recursively transformed into increasingly simpler precursors. The goal of retrosynthetic analysis is structural simplification.
The new development is based on the premise that computer-aided retrosynthesis is possible, and here artificial intelligence can assist. The algorithm behind the intelligence is based on Monte Carlo tree search and symbolic artificial intelligence. In computer science, Monte Carlo tree search is a heuristic search algorithm for some kinds of decision processes, most notably those employed in game play. It can, however, also have a scientific application.
With the new platform, the researchers successfully combined Monte Carlo tree search with an expansion policy network to guide the search, plus a filter network to pre-select the most promising retrosynthetic steps. To make the deep neural networks effective, the researchers trained them in every reaction ever published in organic chemistry (some 12.4 million). Verification was undertaken using a double-blind AB (or ABX) test. An ABX test is a method of comparing two choices of sensory stimuli to identify detectable differences between them.
According to the lead scientist, Pablo Carbonell, from the University of Manchester: “What we have seen here is that this kind of artificial intelligence can capture this expert knowledge.”
The tool is different to previous attempts because the AI platform is capable of learning from the data alone and it does not require people to input rules for it to use.
The artificial intelligence study has been published in the science journal Nature. the peer reviewed research paper is titled: “Planning chemical syntheses with deep neural networks and symbolic AI.”
