Daniel Domine

A General QSAR Model for Predicting the Toxicity of Organic Chemicals to Luminescent Bacteria (Microtox® test)

Abstract A large data set of Microtox® toxicity results was used to derive a general QSAR model. Chemicals were described by means of a modified autocorrelation method. The autocorrelation vectors were generated from atomic contributions encoding the hydrophobicity and molar refractivity of the molecules. A three-layer backpropagation neural network was used to design the model. The obtained results were compared with those obtained from a principal components regression analysis.

Estimating pesticide field half-lives from a backpropagation neural network.

The field half-lives of 110 pesticides were modelled using a backpropagation neural network (NN). The molecules were described by means of the frequency of 17 structural fragments. Before training the NN, different scaling transformations were assayed. Best results were obtained with correspondence factor analysis which also allowed a reduction of dimensionality. The training and testing sets of the NN analysis gave 95.5% and 84.6% of good classifications, respectively. Comparison with discriminant factor analysis showed that a backpropagation NN was more appropriate to model the field half-lives of pesticides.

Chemometrical evaluation of the PAH contamination in the sediments of the Gulf of Lion (France)

Abstract The concentrations of fluorene, dibenzothiophene, phenanthrene, anthracene, retene, fluoranthene, pyrene, chrysene, triphenylene, benz[ a ]anthracene, benzo[ k ]fluoranthene, benzo[ a ]fluoranthene, benzo[ b ]fluoranthene, benzo[ j ]fluoranthene, benzo[ e ]pyrene, benzo[ a ]pyrene, perylene, indeno[1,2,3- c,d ]pyrene, dibenz[ a,h ]anthracene, benzo[ g,h,i ]perylene, 1-methylfluorene, 2-methylfluorene, 3-methylfluorene, 4-methylfluorene, 1-methyldibenzothiophene, 2-methyldibenzothiophene, 3-methyldibenzothiophene, and 4-methyldibenzothiophene were measured in superlayer and underlayer deep sea sediments at 15 different sampling sites in the Gulf of Lion (Mediterranean Sea, France). The level of contamination of these 28 polycyclic aromatic hydrocarbons (PAH) in the sediments of the Rhone river was also recorded for comparative purposes. The results obtained were analyzed by means of the nonlinear mapping (NLM) method. A nonlinear map for the samples and another for the chemicals were generated. They were interpreted by means of multiple graphical displays on which were represented various quantitative and qualitative information. From this approach, it was possible to obtain a better knowledge of the PAH contamination of underwater canyons and deep sea fans in the Gulf of Lion. It also allowed the examination of the relationships between the PAH structures and origins and their concentrations in the different samples under study.

Prediction of Partition Coefficients (LOG P oct) Using Autocorrelation Descriptors

Abstract A backpropagation neural network model, implemented in AUTOLOGP (Version 4.0), was developed for estimating the n-octanol/water partition coefficient of organic molecules from their structure described by means of a modified autocorrelation method. The advantages of the autocorrelation method, which allows the description of any kind of molecules by means of computerized molecular descriptors presenting a physicochemical meaning, were emphasized through the simulation performances obtained with AUTOLOGP (Version 4.0) and from a comparative study involving two regression models derived from various topological indices.

Autocorrelation modeling of lipophilicity with a back-propagation neural network

Abstract From a training set of 7200 chemicals a back-propagation neural network (BNN) model was developed for estimating the n -octanol/water partition coefficient of organic molecules. Chemicals were described by means of a modified autocorrelation method. The advantages of the autocorrelation method were emphasized through the analysis of the simulation performances of the model and from a comparative study involving another BNN model [Quant. Struct. Act. Relat. 16 (1997) 224–230] using a large number of variables (atoms and bonds) derived from connection matrices.

Estimating n-Octanol/Water Partition Coefficients from the Autocorrelation Method

Abstract A log P model was derived from a stepwise regression analysis based on a training set of 800 organic molecules presenting highly diverse structures. Chemicals were described by means of the autocorrelation method using the structural fragmental values of Rekker. Our approach was shown to be simpler and more efficient than the classical method of Rekker which generally requires the use of correction factors for calculating log P values.

Multivariate Analysis of the Input and Output Data in the Fugacity Model Level I

QSAR (Quantitative Structure-Activity Relationship) analysis is a well established tool in pharmacology for optimising series of bio-active molecules or understanding a mechanism of activity (Fuller and Marsh, 1972; Fukunaga et al., 1976; Murray et al., 1976; Bonjean and Luu Duc, 1978; Fernandez et al., 1978; Di Paolo et al., 1979; Glennon et al.,1979; Thijssen, 1981; Basak et al., 1983; Reed et al., 1984; Carotti et al.,1985; Noel-Artis et al., 1985; Ray et al., 1985; Goghari et al., 1986; Kawashima et al., 1986; Motoc et al., 1986; Seiler et al., 1986; Werbel et al., 1986; Zeelen, 1986; Basak, 1987; Taillandier and Domard, 1987; Takahashi et al., 1987; Hansch et al., 1989; Dearden, 1990). The technique is statistically-based and aimed at extracting the maximum information from biological data on compounds of known structure and physicochemical properties. Historically, QSAR data analysis was dominated by the use of regression analysis, but since the early 1970s multivariate methods have been increasingly infiltrating the field (Hansch et al., 1973; Moriguchi and Komatsu, 1977; Henry and Block, 1980; MassartLeen and Massart, 1981; Grassy et al., 1982; Broto et al., 1984; Dunn et al., 1984b; Franke, 1984; Coats et al., 1985; De Flora et al., 1985; de Winter, 1985; Dufton, 1985; Laass, 1985; Leavitt and Mass, 1985; Schaper and Seydel, 1985; Benigni 1986, 1989, 1990; Benigni and Giuliani, 1986, 1987, 1988a, b; Berntsson and Wold, 1986; Lewi, 1986; McFarland and Gans, 1986, 1987; Stouch and Jurs, 1986; Enslein et al., 1987; Mager, 1988; Moreau et al., 1988; Vogt, 1988; Jerman-Blazic et al., 1989; Clementi et al., 1989; Ford et al., 1989; Grassy and Bonnafous, 1989; Livingstone and Rahr, 1989; Rose et al., 1990).

Occupational exposure modelling with ease.

This article presents a validation exercise performed from eight practical case studies on EASE (version 2.0), a knowledge-based system allowing to estimate the workplace exposure to chemicals. Our results show that EASE represents a valuable simulation tool in occupational hygiene. However, it requires to be refined and extended to more realistic and precise situations to be easily used in practice.

Nonlinear neural mapping analysis of the adverse effects of drugs.

Numerous drugs have been identified as presenting adverse effects towards the driving of vehicles. A large set of these drugs was compiled and classified into ten categories. Nonlinear neural mapping (N2M) was used to derive a typology of these molecules and also to link their adverse effects to therapeutic categories and structural information.

Modeling the Environmental Fate of Atrazine

Mathematical simulation models of fate and transport of chemicals have been identified by researchers and regulators as potentially valuable tools for improving the understanding of the environmental behavior of chemicals which may be released to the environment as a consequence of routine (i.e., normal manufacturing, use, disposal) and non-routine (e.g., accidental spillage) events. In this context, CHEMFRANCE, a regional fugacity model level III, which calculates the environmental distribution of organic chemicals in 12 defined regions of France, or France as a whole, has been designed. The aim of this study is to show that CHEMFRANCE provides valuable simulation results for understanding the environmental fate behavior of atrazine.