DeepSynthesis | AI Chemical Platform

Brief Introduction to the BDE Prediction Tool:

1. Purpose and scope: This tool predicts BDEs for noncyclic bonds in organic molecules consisting of C, H, O, N and other Group IVA, VA, VIA, VIIA elements of first five periods. 1) Prediction of unsaturated bond is not recommended; 2) Prediction of an ion pair is not supported, please try with either its cation or anion only. 3) Prediction of free radical/ionic radical is not supported.

2. Prediction methods: LightGBM and SPOC descroptors with RMSE=1.82, r²=0.980 and Mean Absolute Errors (MAE) = 1.03 (95:5 train test split); For SPOC descroptors, please refer to our previous work about pK_a prediction and SPOC.

3. Dataset: The BDE dataset for training contains more than 4000 Experimental BDEs and 290000 calculation BDEs. 1) Experimental data: experimental data comes from the sub-database of iBonD, for more details please click http://ibond.nankai.edu.cn. 2) Calculation data: please refer to Peter C. St. John and co-workers' dataset.

4. Citation: Yidi Liu, Yao Li, Qi Yang, Jindong Yang, Long Zhang*, Sanzhong Luo*. Prediction of Bond Dissociation Energy for Organic Molecules Based on a Machine-Learning Approach. Chin. J. Chem. 2024 42, 1967-1974.