Prediction of toxicity of aliphatic carboxylic acids using adaptive neuro-fuzzy inference system

Abstract

Toxicity of 38 aliphatic carboxylic acids was studied using non-linear quantitative structure-toxicity<br />relationship (QSTR) models. The adaptive neuro-fuzzy inference system (ANFIS) was used to construct the<br />nonlinear QSTR models in all stages of study. Two ANFIS models were developed based upon different<br />subsets of descriptors. The first one used log ow K and LUMO E as inputs and had good prediction ability; for<br />the training set of 28 compounds 2<br />Training R was 0.86 and for the test set of 10 compounds, the corresponding<br />statistic was 2<br />Test R =0.97. Two outliers were detected for this ANFIS model and removing them improved the<br />quality of the model. Another ANFIS model was constructed based on PEOE_VSA_FPNEG and G3u<br />descriptors chosen by exhaustive search of all two combinations of calculated descriptors by Dragon and<br />MOE softwares. The later ANFIS model showed better performance than the former ( 2<br />Training R =0.92 and<br />2<br />Test R =0.90) and no outlier was detected.