Qsar Study Of Phenol, Aniline Compounds, Quantitative Structure Toxicity

The molecular descriptors of 36 phenols and anilines were calculated by AMI quantum chemical computation with software ChemOffice6.0, which including such electiric parameters as logKow, MR, HOP, TE, Ele.E et al. and the orbit energy parameters as ELUMO, EHOMO- And it was discussed with the relationships between the 24/zECso to Daphnia magna with the each parameter.The QSAR model was built with the multiple linear step regression method, and it was tested for the stability with other sample compounds. To show its un-linearity between these parameters, the QSAR model has been got by BP artificial neural networks with its suitable net structure( 8:8:1). Based on the different methods, the compare between the different models has been taken, the result was shown that the QSAR model with ANN owned better prediction ability than MLR. Otherwise the toxicity mechanism to the phenols and anilines was discussed in details.