W1 AM Nuts and Bolts of Logistics and Data Management for Data-Intensive Programs
Wednesday, 16 November 2005: 8:00 AM - 11:40 AM in Ballroom 1

(KOM-1118-046543) Study on (Quantitative) structure-activity relationships of acute ecotoxicity based pattern recognition methods using fragment.

Komatsu, Eiji¹, Shiraishi, Hiroaki ¹, Yoshioka, Yoshitada², ¹ National Institute for Environmental Studies, Japan, Tsukuba, Ibaraki, Japan² Oita University, Oita, Japan

ABSTRACT- The quantitative structure-activity relationship (QSAR) has become a powerful alternative theoretical tool for the prediction of chemical toxicity based on molecular descriptor. Thus, the application of QSAR is expected assisting for data gap filling in the assessment of chemicals under law concerning the evaluation of chemical substances. However, the inventory of actual evaluated substances has a long list and a large diversity of chemical structure, then, it is difficult to predict the physico-chemical property and toxicity of such chemicals with using the QSAR. This study deals with classification for acute toxicity prediction based on the fragment and descriptors related with mode of action. We built several QSAR models and investigated to define the domain identification of QSAR modeling in considering for functional cluster, modes of toxic action and relationship between descriptor variables and toxicity. This technical approach was applied for the acute fish toxicity by using a data set of more than 600 chemicals which was investigated or selected from literature. The result showed established QSAR model on some classified categories was described relationship between the toxicity and the logarithm of the octanol-water partition coefficient (log P). However, the prediction of the toxicity of categories which have a lot of reactive chemicals was found to require a multiple regression model using of additional descriptors or non-linear models e.g. an artificial neural networks model. As for validation result of the approach proposed by this study, the comparison evaluation is executed with existing QSAR models of aquatic toxicity.

Key words: QSAR, Aquatic toxicity, Mode of action, Classification