摘要

A QSPR (Quantitative Structure-Property Relationship) model was derived for a set of forty 7 beta-hydroxysteroid compounds selected from PubChem database in order to assess the link between structural features and lipophilicity expressed as logP. After optimization and topological indices data collecting, the cluster of molecules was superposed onto a representative hypermolecule. Based on each molecule atoms positions, a binary vector and its weighted by mass fragments was computed for each molecule in the set. A model relating the structure with logP was identified based on the contributions of statistically significant positions of each molecule superposed on the hypermolecule and based on structural descriptors. The obtained model was validated in leave-one-out analysis as well as on training versus test analysis.

  • 出版日期2016-3-1