Prediction of Chromatography Conditions for Purification in Organic Synthesis Using Deep Learning

Abstract

In this research, a process for developing normal-phase liquid chromatography solvent systems has been proposed. In contrast to the development of conditions via thin-layer chromatography (TLC), this process is based on the architecture of two hierarchically connected neural network-based components. Using a large database of reaction procedures allows those two components to perform an essential role in the machine-learning-based prediction of chromatographic purification conditions, i.e., solvents and the ratio between solvents. In our paper, we build two datasets and test various molecular vectorization approaches, such as extended-connectivity fingerprints, learned embedding, and auto-encoders along with different types of deep neural networks to demonstrate a novel method for modeling chromatographic solvent systems employing two neural networks in sequence. Afterward, we present our findings and provide insights on the most effective methods for solving prediction tasks. Our approach results in a system of two neural networks with long short-term memory (LSTM)-based auto-encoders, where the first predicts solvent labels (by reaching the classification accuracy of 0.950 ± 0.001) and in the case of two solvents, the second one predicts the ratio between two solvents (R2 metric equal to 0.982 ± 0.001). Our approach can be used as a guidance instrument in laboratories to accelerate scouting for suitable chromatography conditions.

Toward microfluidic continuous-flow and intelligent downstream processing of biopharmaceuticals

“…130 In addition to soft sensors, ML algorithms are also crucial components for constructing smart downstream processing platforms. ML models have vastly been employed to create accurate predictive models for mechanistic modeling 131 and parameter optimization [132][133][134][135][136] in the chromatography process, for enhancing efficiency and product quality while reducing experimentation time and cost. ML models have also been intensively used for development of membranes for UF 136,137 and APTS.…”

Toward an Integrated Machine Learning Model of a Proteomics Experiment

“…Prediction of how proteins and peptides behave in electrophoretic or chromatographic separations is important in both analytical and preparative contexts, including when analyzing proteomics data from experimental workflows including multiple dimensions of separations or when optimizing preparative methods used to purify recombinant proteins, synthetic peptides, and other products. 69 Such predictive models can also be used to optimize chromatographic fractionation in proteomics workflows, even if the optimization target is very different than when purifying a single component, for example adjusting chromatographic conditions to distribute the proteins or peptides evenly between fractions with minimal overlap. 70 , 71 …”

Machine learning in bioprocess development: from promise to practice

“…In this review we will thus mainly discuss the application of ML in bioprocess development, particularly in upstream processes. ML is also advancing in downstream processing, where corresponding techniques are developed for specific technologies such as chromatography [26][27][28][29][30][31], but also for complex purification pipelines of specific products such as antibodies [32] or inclusion bodies [33]. Since available literature is highly diverse and vast enough to be covered in a separate review, ML for downstream processing is not discussed in detail here.…”