QSAR Modeling of Acute Toxicity for Nitrobenzene Derivatives Towards Rats: Comparative Analysis by MLRA and Optimal Descriptors
Andrey Andreevich Toropov, Bakhtiyor Rasulev, Jerzy Leszczyński · QSAR & Combinatorial Science · 2007
Abstract Quantitative Structure–Activity Relationships (QSAR) have been developed for a set of 28 benzene derivatives. LD50 oral toxicity of these compounds towards rats has been modeled by Multiple Linear Regression Analysis (MLRA) based on descriptors generated by DRAGON software and by optimal descriptors approach. Twenty‐eight benzene derivatives have been split into training (n= 14) and test (n=14) sets. In the case of MLRA, a two‐variable model has the best predictive potential. Comparison of the quality of MLRA and optimal descriptor models showed that the predictive potential of a one‐variable model based on optimal descriptors is better than a two‐variable MLRA model.