Table S1.

PCA loadings score for active and inactive DPP4 inhibitors

Descriptor	Active			Inactive
	PC1	PC2	PC3	PC1	PC2	PC3
MW	0.849	0.195	0.402	0.849	−0.376	0.224
RBN	0.638	0.431	−0.025	0.664	−0.017	0.476
nCIC	0.297	−0.241	0.739	0.489	−0.651	0.077
nHDon	0.578	−0.231	−0.204	0.341	0.605	0.289
nHAcc	0.637	0.527	0.013	0.893	0.160	0.112
ALogP	0.061	0.267	0.564	0.164	−0.813	0.203
TPSA	0.693	0.268	−0.239	0.815	0.405	0.094
Qm	0.445	0.410	−0.542	0.601	0.483	−0.144
Energy	−0.242	−0.765	0.102	−0.654	−0.264	−0.234
Dipole moment	0.361	−0.617	−0.261	0.439	0.091	−0.634
HOMO	−0.361	0.637	0.380	−0.211	−0.352	0.698
LUMO	−0.585	0.732	0.073	−0.559	0.344	0.671
HOMO–LUMO	−0.575	0.452	−0.413	−0.463	0.608	0.258

Notes: The bold values represent the highest loadings scores at the current PC, compared to other PCs. For instance, MW has a higher loading score of 0.849 at PC1, compared to PC2 (0.195), and PC3 (0.402).

Abbreviations: ALogP, Ghose–Crippen octanol–water partition coefficient; HOMO, highest occupied molecular orbital; HOMO–LUMO, energy gap between the HOMO and LUMO states; LUMO, lowest unoccupied molecular orbital; MW, molecular weight; nCIC, number of rings; nHAcc, number of hydrogen bond acceptors; nHDon, number of hydrogen bond donors; PCA, principal component analysis; Q_m, mean absolute charge; RBN, rotatable bond number; TPSA, topological polar surface area.