EVALUATION OF THE MODEL PREDICTION TOXICITY (LD50) FOR SERIES OF 42 ORGANOPHOSPHORUS PESTICIDES

Abstract

Structure-Toxicity Relationships have been studied for a set of 42 organophosphorous pesticides (OPs) through multiple linear regression (MLR) and artificial neural networks (ANN). A model with three descriptors, including: total lipophilicity [log (P)], widths radicals R1 [(LR1)] and R2 [(LR2)] has achieved good results in phase Training and phase prediction of toxicity [log LD50 (lethal dose 50, Oral rat)]. The linear model (MLR: n=40, r²=0.86, s=40 and q2 = 0.66) and non-linear model with a configuration [3-6-1] (ANN: r²=0.95, s=0.73 and q2 = 0.17) have proved very successful and complementary. The selected descriptors indicate the importance of lipophilicity and widths radicals R1 and R2 in the contribution of the toxicity of pesticides derived from OPs used in this study. This information is relevant for the design of a new model of non-toxic pesticides OPs.

Keywords

• multiple linear regression (MLR)
• artificial neural networks (ANN)
• organophosphorous pesticides (OPS)
• LD50
• descriptors