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

toxicity of 38 aliphatic carboxylic acids was studied using non-linear quantitative structure-toxicityrelationship (qstr) models. the adaptive neuro-fuzzy inference system (anfis) was used to construct thenonlinear qstr models in all stages of study. two anfis models were developed based upon differentsubsets of descriptors. the first one used log ow k and lumo e as inputs and had good prediction ability; forthe training set of 28 compounds 2training r was 0.86 and for the test set of 10 compounds, the correspondingstatistic was 2test r =0.97. two outliers were detected for this anfis model and removing them improved thequality of the model. another anfis model was constructed based on peoe_vsa_fpneg and g3udescriptors chosen by exhaustive search of all two combinations of calculated descriptors by dragon andmoe softwares. the later anfis model showed better performance than the former ( 2training r =0.92 and2test r =0.90) and no outlier was detected.