ABSTRACT

QSAR (Quantitative Structure Activity Relationship) modelling was performed on a dataset of 90 sodium-dependent glucose cotransporter 2 (SGLT2) inhibitors. The quantitative and explicative evaluations revealed some of the subtle and distinguished structural features that are responsible for the inhibitory potency of these compounds against SGLT2, such as less possible number of ring carbons at 8 Å from the lipophilic atoms in the molecule (fringClipo8A) and more possible value for the sum of the partial charges of the lipophilic atoms present within seven bonds from the donor atoms (lipo_don_7Bc). Multivariate GA–MLR (genetic algorithm–multi linear regression) and thorough validation methodology out-turned a statistically robust QSAR model with a very high predictability shown from various statistical parameters. A QSAR model with r² = 0.83, F = 51.54, Q²_LOO = 0.79, Q²_LMO = 0.79, CCC_cv = 0.88, Q²Fⁿ = 0.76–0.81, r²_ext = 0.77, CCC_ext = 0.85, and with RMSE_tr < RMSE_cv was proposed. This QSAR model will assist synthetic chemists in the development of the SGLT2 inhibitors as the antidiabetic leads.

摘要

对 90 种钠依赖性葡萄糖协同转运蛋白 2 (SGLT2) 抑制剂的数据集进行 QSAR（定量结构活性关系）建模。定量和解释性评估揭示了导致这些化合物对 SGLT2 具有抑制效力的一些微妙和显着的结构特征，例如距离分子中亲脂性原子 8 Å 的碳环数量较少（fringClipo8A）等等来自供体原子 (lipo_don_7Bc) 的七个键中存在的亲脂性原子的部分电荷总和的可能值。多元 GA-MLR（遗传算法-多元线性回归）和彻底的验证方法产生了一个具有统计鲁棒性的 QSAR 模型，从各种统计参数显示出具有非常高的可预测性。QSAR 模型[R ²  = 0.83，˚F = 51.54，Q ² _LOO  = 0.79，Q ² _LMO  = 0.79，CCC _CV  = 0.88，Q ² ˚F ^Ñ  = 0.76-0.81，- [R ² _EXT  = 0.77，CCC _EXT  = 0.85，并用RMSE _TR  <提出了RMSE _cv。该 QSAR 模型将帮助合成化学家开发 SGLT2 抑制剂作为抗糖尿病药物的先导。