摘要

Multivariate data analysis (MVDA) was applied on a set of 160 indole-based PPAR-gamma agonists for modeling receptor binding (pKi) and gene transactivation (pEC(50)) data. A pool of descriptors based on easily interpretable physicochemical, molecular and constitutional properties, including drug-like characteristics was used. A PLS model with satisfactory and robust statistics was produced for pKi data, while the inferior quality of pEC(50) model reflected the higher complexity associated with transactivation process. In both PLS models, lipophilicity, molecular weight and the number of halogens exerted a significant positive effect in activity, while molecules should preferably be compact and less flexible. An activity-activity model was further established using multiple regression analysis, which improved the interrelationship between pEC(50) and pKi.

  • 出版日期2009-8