Discovering New Chemistry with an Autonomous Robotic Platform Driven by a Reactivity-Seeking Neural Network

Abstract

We present a robotic chemical discovery system capable of navigating a chemical space based on a learned general association between molecular structures and reactivity, while incorporating a neural network model that can process data from online analytics and assess reactivity without knowing the identity of the reagents. Working in conjunction with this learned knowledge, our robotic platform is able to autonomously explore a large number of potential reactions and assess the reactivity of mixtures, including unknown chemical spaces, regardless of the identity of the starting materials. Through the system, we identified a range of chemical reactions and products, some of which were well-known, some new but predictable from known pathways, and some unpredictable reactions that yielded new molecules. The validation of the system was done within a budget of 15 inputs combined in 1018 reactions, further analysis of which allowed us to discover not only a new photochemical reaction but also a new reactivity mode for a well-known reagent (p-toluenesulfonylmethyl isocyanide, TosMIC). This involved the reaction of 6 equiv of TosMIC in a “multistep, single-substrate” cascade reaction yielding a trimeric product in high yield (47% unoptimized) with the formation of five new C–C bonds involving sp–sp2 and sp–sp3 carbon centers. An analysis reveals that this transformation is intrinsically unpredictable, demonstrating the possibility of a reactivity-first robotic discovery of unknown reaction methodologies without requiring human input.

Artificial neural networks and data fusion enable concentration predictions for inline process analytics

“…Our attention was drawn to convolutional neural networks (CNNs), which have been previously applied for NMR data. 12 b , c ,19…”

Challenges in lab-on-a-chip technology

“…Beyond applications-led lab-on-a-chip, the field of intelligent microfluidics is also emerging, using machine learning for the monitoring and control of microfluidic systems providing new challenges and opportunities to accelerate chemical exploration and synthesis at reduced costs. These concepts will underpin a new generation of high throughput discovery tools associated with digital chemistry (Caramelli et al, 2021), developing new methods in important areas such as new medicines synthesis and screening, as well as in, e.g., advanced materials discovery.…”

Sharing is Caring: Guidelines for Sharing in the Electronic Laboratory Notebook (ELN) Chemotion as applied by a Synthesis-oriented Working Group

“…The digital revolution is transforming chemistry in an unprecedented way: computational chemistry allows the detailed description of reaction mechanisms [1] , methods of machine learning enable retrosynthesis and the prediction of material properties [2,3] , and software designed to control chemical reaction processes, such as a Chemputer to perform syntheses and their optimization in automated feedback-loops. [4,5] While analytical data are already mainly generated and stored digitally, many scientists still use handwritten laboratory notebooks to record the experiments related to their obtained analytical data. [6][7][8] Recording experiments in an analogous way leads to several disadvantages, like limited accessibility for scientists outside the laboratory, possible misinterpretation due to illegibility, or the missing connection to the digitally stored data.…”