Aquatic Toxicity Assessment of Esters Towards the Daphnia magna Through PCA-ANFIS

作者：Asadollahi Baboli M^*

来源：Bulletin of Environmental Contamination and Toxicology, 2013, 91(4): 450-454.

DOI：10.1007/s00128-013-1066-8

• Aquatic toxicity
• Ester
• MIA-QSTR
• Daphnia magna
• Principal component analysis
• Adaptive neuro-fuzzy inference systems

The widespread production of esters combined with their ability to migrate in different compartments, makes their environmental toxicity important. In this background, the multivariate image analysis-quantitative structure-toxicity relationship (MIA-QSTR) method coupled to principal component analysis-adaptive neuro-fuzzy inference systems (PCA-ANFIS) was applied to assess the toxicity of esters to Daphnia magna. In MIA-QSTR, pixels of chemical structures (2D images) stand for descriptors, and structural changes account for the variance in toxicities. The ANFIS procedure was capable of correlating the inputs (PCA scores) with the toxicities accurately. The PCA-ANFIS also was statistically validated for its predictive power using cross-validation, applicability domain and Y-scrambling evaluation procedures. The satisfactory results (R (p) (2) = 0.926, Q (LOO) (2) = 0.887, R (L25 %O) (2) = 0.843, RMSELOO = 0.320 and RMSEL25 %O = 0.379) suggests that the QSTR model could be proposed as an alternative method for aquatic toxicity assessment of esters allowing possible application in the European Union regulation REACH.

• 出版日期2013-10