Holistic Prediction of the pKa in Diverse Solvents Based on a Machine‐Learning Approach. (25th August 2020)

• Record Type: Journal Article
• Authors: Yang, Qi; Li, Yao; Yang, Jin‐Dong; Liu, Yidi; Zhang, Long; Luo, Sanzhong; Cheng, Jin‐Pei
• Is Part Of: Angewandte Chemie. Volume 132:Number 43(2020)
• Page Start: 19444
• Page End: 19453
• Publication Date: 2020-08-25
• Subjects: iBond -- machine learning -- neural network -- organocatalysts -- pKa prediction -- XGBoost; Chemistry -- Periodicals; 540
• DOI: 10.1002/ange.202008528
• Languages: English
• ISSNs: 0044-8249

Abstract

Abstract: While many approaches to predict aqueous p K a values exist, the fast and accurate prediction of non‐aqueous p K a values is still challenging. Based on the iBonD experimental p K a database (39 solvents), a holistic p K a prediction model was established using machine learning. Structural and physical‐organic‐parameter‐based descriptors (SPOC) were introduced to represent the electronic and structural features of the molecules. The models trained with a neural network or the XGBoost algorithm showed the best prediction performance with a low MAE value of 0.87 p K a units. The approach allows a comprehensive mapping of all possible p K a correlations between different solvents and it was validated by predicting the aqueous p K a and micro‐p K a of pharmaceutical molecules and p K a values of organocatalysts in DMSO and MeCN with high accuracy. An online prediction platform was constructed based on the current model, which can provide p K a prediction for different types of X−H acidity in the most commonly used solvents. Abstract : Deep learning enables the holistic p K a prediction of various types of X−H acidities in 39 types of solvents. The accuracy of the predictions is demonstrated by a mean absolute error of 0.87 p K a units.