Studies on two types of PTP1B inhibitors for the treatment of type 2 diabetes: Hologram QSAR for OBA and BBB analogues.

Hologram quantitative structure-activity relationships (HQSAR) analysis were conducted on two series of PTP1B inhibitors, 39 2-(oxalylamino) benzoic acid (OBA) analogues and 60 benzofuran and benzothiophene biphenyls (BBB) analogues. The optimal HQSAR model of the OBA analogue has q(2)=0.592 and r(2)=0.940, while the optimal HQSAR model for the BBB analogues shows q(2)=0.667 and r(2)=0.863. Two models were employed to predict the biological activities of two test sets. For OBA analogues, the optimal model was validated by an external test set of six compounds with satisfactory predictive r(2) value of 0.786. For BBB analogues, the optimal model shows satisfactory predictive r(2) value of 0.866 for an external test set of 10 compounds. The contribution maps derived from the optimal HQSAR models are consistent with the biological activities of the studied compounds. Two virtual combinatorial libraries were designed and screened by the optimal HQSAR models and potential candidates with high predictive biological activities were discovered. This work may provide valuable information for future design of more promising inhibitors for PTP1B.