DeepSynthesis | AI Chemical Platform

Please Cite:
Qi Yang, Yao Li, Jin-Dong Yang, Yidi Liu, Long Zhang,* Sanzhong Luo,* Jin-Pei Cheng. Holistic Prediction of the pK_a in Diverse Solvents Based on a Machine Learning Approach. Angew. Chem. Int. Ed. 2020, 59, 19282-19291.

1. Abbreviations of solvents: H₂O: Water; DMSO: Dimethyl sulfoxide; EtOH_50%: Ethanol:Water=50:50; AN: Acetonitrile; MeOH: Methanol.

2. Prediction methods: XGBoost with RMSE=1.79 and r²=0.918 (80:20 train test split).

3. Experimental data: experimental data comes from the sub-database of iBonD, for more details please click http://ibond.nankai.edu.cn.

4. Recommended pK_a ranges: H₂O: -2~16; DMSO: 5~35; EtOH_50%: 0~20; AN: 5~30; MeOH: 0~18.

5. Note: for special molecule's pK_a which is out of the solvent leveling range, the experimental value is unreliable, so we exclude these values from training set. For these molecules, this model will return unreliable results.