Quantitative Structure Activity Relationship Modeling of Environmentally Important Diphenyl Ether Herbicides Using MLR and PLS

A quantitative structure - activity relationship (QSAR) study has been applied to a data set of 33 diphenyl ether herbicide with their inhibition data on protoporphyrinogen oxidase (PPO) enzyme. Diphenyl ethers (DPEs) are a class of herbicides that strongly and competitively inhibit the protoporphyrinogen oxidase (PPO) enzyme. PPO is the last common enzyme in the biosynthetic pathway to heme and chlorophyll. Two linear correlating models, multiple linear regression (MLR) and partial least squares (PLS) regression methods were used. Stepwise regression as a variable selection method was employed to develop a regression equation based on 26 training compounds, and predictive ability was tested on 7 compounds reserved for that purpose. A multi-parametric equation containing four structural descriptors with good statistical impact was obtained. The squared regression coefficients of prediction for the MLR and PLS regression methods were 0.95 and 0.94 respectively. The physicochemical meaning of the descriptors and their relations to the dependent variable are discussed. Keywords: Protoporphyrinogen oxidase, Diphenyl ether herbicide, QSAR, MLR, PLS, Theoretical descriptors