In this study, a general quantitative structure-property relationship (QSPR) protocol, fragments based graph convolutional neural network (F-GCN), was developed for atomic and inter-atomic properties predictions. We applied this novel artificial intelligence (AI) tool in NMR chemical shifts and bond dissociation energies (BDEs) predictions. The predicted results were comparable to experimental measurement, while the computational cost was substantially reduced, with respect to pure density functional theory (DFT) calculations. The two important features of F-GCN can be summarised as: first, it could utilise different levels of molecular fragments centered at the target chemical bonds for atomic and inter-atomic information extraction; second, the designed architecture is also open to include additional descriptors for more accurate solution of chemical environment, making itself more efficient for local properties descriptions. And during our test, the averaged prediction error of 1H NMR chemical shifts can be as small as 0.32 ppm; and the error of C-H BDEs estimations, is 2.7 kcal/mol. Moreover, we further demonstrated the applicability of this developed F-GCN model via several challenging structural assignments. The success of the F-GCN in atomic and inter-atomic predictions also indicates an essential improvement of computational chemistry with the assistance of AI tools.
The first version of the manuscript.