Digitization and validation of a chemical synthesis literature database in the ChemPU

Abstract

Despite huge potential, automation of synthetic chemistry has only made incremental progress over the past few decades. We present an automatically executable chemical reaction database of 100 molecules representative of the range of reactions found in contemporary organic synthesis. These reactions include transition metal–catalyzed coupling reactions, heterocycle formations, functional group interconversions, and multicomponent reactions. The chemical reaction codes or χDLs for the reactions have been stored in a database for version control, validation, collaboration, and data mining. Of these syntheses, more than 50 entries from the database have been downloaded and robotically run in seven modular ChemPU’s with yields and purities comparable to those achieved by an expert chemist. We also demonstrate the automatic purification of a range of compounds using a chromatography module seamlessly coupled to the platform and programmed with the same language.

The way to AI-controlled synthesis: how far do we need to go?

“…They employed this more mature automatic platform to process 53 kinds of chemical reactions, which are from 103 kinds of reactions that are encoded by χDL, and obtained good results and high yields. This work makes it possible to replicate chemical synthesis experiments from previous literatures, and also much improves the performance of current AI-controlled synthesis system 73 . The real jobs of chemists are just to rewrite or to modify codes with C++, Python, or even their own machine languages.…”

“…In summary, the AI-controlled synthesis system could quickly and accurately control reaction conditions such as temperature and pressure to handle sensitive chemicals and dangerous reactions more safely [74][75][76] . Fortunately, some groups are trying to overcome these challenges from the flaws of the current AI-controlled synthesis system and have made some achievements 35,47,69,73 . Beyond all doubt, with the development of AI algorithms of the software and engineering technology of the hardware, we believe that chemists will move toward the goal of AI-controlled synthesis faster.…”

Open data and algorithms for open science in AI-driven molecular informatics

“…The developments of the past years demonstrate the potential of data-driven machine learning applications in the field of molecular informatics in an impressive manner [5,51,55]. An obvious requirement to benefit from this development is the availability of open structured experimental data [11,12].…”

“…Similar model architectures were successfully used to generate specific synthesis instructions [53] and to determine the yield [54] of a given chemical reaction formula. Recently, Rohrbach et al demonstrated the translation of synthesis protocols in the literature into a standardized chemical language which could then be executed by their automated synthesis system [55]. Again, the described advances are exemplary cases of the synergy of deep learning-based models and the availability of training data.…”

Autonomous Chemical Experiments: Challenges and Perspectives on Establishing a Self-Driving Lab

“…While a small set of couplings under a limited range of conditions can be useful for specific applications, it is a much greater challenge to build a self-driving lab that can make molecules of arbitrary structure. However, this outstanding challenge is too complex to be addressed here and continues to see significant ongoing developments that lead to expanded capabilities. ,,,− In summary, to run smoothly, a self-driving lab must be controlled by robust algorithms and code. Human cognitive capabilities can usually handle such tasks with ease; however, their automation can be very difficult.…”