A closed-loop, autonomous molecular discovery platform driven by integrated machine learning tools was developed to accelerate the design of molecules with desired properties. Two case studies are demonstrated on dye-like molecules, targeting absorption wavelength, lipophilicity, and photo-oxidative stability. In the first, the platform experimentally realized 312 unreported molecules across three automatic iterations of molecular design-make-test-analyze cycles while exploring the structure–function space of four rarely reported scaffolds. In each iteration, the property-prediction models which guided the exploration learned the structure–property space of diverse inexpensive scaffold derivatives realized through using multi-step syntheses. Conversely, the second study exploited property models trained on a chemical space with pre-existing examples to discover 6 top-performing molecules within the structure-property space. By closing the molecular discovery cycle of prediction, synthesis, measurement, and model retraining, the platform demonstrates the potential for integrated platforms to automatically understand a local chemical space and discover functional molecules.
We demonstrate the usefulness of general atom- and bond-level DFT descriptors to enhance the performance of neural networks for general reaction condition prediction. We treat reaction condition prediction as a multi-class classification task and report the performance of neural networks trained on 59,512 reactions with 283 distinct reaction condition classes and varying input embedding compositions. We show that by combining structural and general DFT descriptors in optimized ratios, models with input size up to 15% smaller than their purely structural counterparts can provide comparable recall, top-1 and top-3 accuracies. Moreover, we report improvements of up to 6%, 7% and 9% in weighted F1 score, top-1 accuracy and weighted recall, respectively, for neural networks trained on combined general DFT and structural descriptors when compared to purely structural models with equivalent architectures and input sizes. Remarkably, these results were achieved using a training set containing 267 times fewer data points than the one used for generating and embedding structural descriptors, despite both embedding strategies being similar unsupervised learning algorithms.
Automation and digitalization solutions in the field of small molecule synthesis face new challenges for chemical reaction analysis, especially in the field of high-performance liquid chromatography (HPLC). Chromatographic data remains locked in vendors hardware and software components, limiting their potential in automated workflows and data science applications. In this work, we present an open-source Python project called MOCCA for the analysis of HPLC-DAD (photodiode array detector) raw data. MOCCA provides a comprehensive set of data analysis features, including an automated peak deconvolution routine of known signals, even if overlapped with signals of unexpected impurities or side products. We highlight the broad applicability of MOCCA in four studies: (i) a simulation study to validate MOCCAs data analysis features; (ii) a reaction kinetics study on a Knoevenagel condensation reaction demonstrating MOCCAs peak deconvolution feature; (iii) a closed-loop optimization study for the alkylation of 2-pyridone without human control during data analysis; (iv) a well plate screening of categorical reaction parameters for a novel palladium-catalyzed cyanation of aryl halides employing O-protected cyanohydrins. By publishing MOCCA as a Python package with this work, we envision an open-source community project for chromatographic data analysis with the potential of further advancing its scope and capabilities.