Gabriela Espinosa

A Fuzzy ARTMAP Based on Quantitative Structure-Property Relationships (QSPRs) for Predicting Aqueous Solubility of Organic Compounds

Quantitative structure-property relationships (QSPRs) for estimating aqueous solubility of organic compounds at 25 degrees C were developed based on a fuzzy ARTMAP and a back-propagation neural networks using a heterogeneous set of 515 organic compounds. A set of molecular descriptors, developed from PM3 semiempirical MO-theory and topological descriptors (first-, second-, third-, and fourth-order molecular connectivity indices), were used as input parameters to the neural networks. Quantum chemical input descriptors included average polarizability, dipole moment, resonance energy, exchange energy, electron-nuclear attraction energy, and nuclear-nuclear (core-core) repulsion energy. The fuzzy ARTMAP/QSPR correlated aqueous solubility (S, mol/L) for a range of -11.62 to 4.31 logS with average absolute errors of 0.02 and 0.14 logS units for the overall and validation data sets, respectively. The optimal 11-13-1 back-propagation/QSPR model was less accurate, for the same solubility range, and exhibited larger average absolute errors of 0.29 and 0.28 logS units for the overall and validation sets, respectively. The fuzzy ARTMAP-based QSPR approach was shown to be superior to other back-propagation and multiple linear regression/QSPR models for aqueous solubility of organic compounds.

Neural Network Based Quantitative Structural Property Relations (QSPRs) for Predicting Boiling Points of Aliphatic Hydrocarbons

Quantitative structural property relations (QSPRs) for boiling points of aliphatic hydrocarbons were derived using a back-propagation neural network and a modified Fuzzy ARTMAP architecture. With the back-propagation model, the selected molecular descriptors were capable of distinguishing between diastereomers. The QSPRs were obtained from four valance molecular connectivity indices (1chiv,2chiv,3chiv,4chiv), a second-order Kappa shape index (2kappa), dipole moment, and molecular weight. The inclusion of dipole moment proved to be particularly useful for distinguishing between cis and trans isomers. A back-propagation 7-4-1 architecture predicted boiling points for the test, validation, and overall data sets of alkanes with average absolute errors of 0.37% (1.65 K), 0.42% (1.73 K), and 0.37% (1.54 K), respectively. The error for the test and overall data sets decreased to 0.19% (0.81 K) and 0.31% (1.30 K), respectively, using the modified Fuzzy ARTMAP network. A back-propagation alkene model, with a 7-10-1 architecture, yielded predictions with average absolute errors for the test, validation, and overall data sets of 1.96% (6.79 K), 1.83% (6.45 K), and 1.25% (4.42 K), respectively. Fuzzy ARTMAP reduced the errors for the test and overall data sets to 0.19% (0.73 K) and 0.25% (0.95 K), respectively. The back-propagation composite model for aliphatic hydrocarbons, with a 7-9- architecture, yielded boiling points with average absolute errors for the test, validation, and overall set of 1.74% (6.09 K), 1.25% (4.68 K), and 1.37% (4.85 K), respectively. The error for the test and overall data sets using the Fuzzy ARTMAP composite model decreased to 0.84% (1.15 K) and 0.35% (1.35 K), respectively. Performance of the QSPRs, developed from a simple set of molecular descriptors, displayed accuracy well within the range of expected experimental errors and of better accuracy than other regression analysis and neural network-based boiling points QSPRs previously reported in the literature.

A fuzzy ARTMAP-based quantitative structure-property relationship (QSPR) for the Henry's law constant of organic compounds.

Quantitative structure-property relationships (QSPRs) for estimating a dimensionless Henrys Law constant of organic compounds at 25 degrees C were developed based on a fuzzy ARTMAP and back-propagation neural networks using a heterogeneous set of 495 organic compounds. A set of molecular descriptors developed from PM3 semiempirical MO-theory and topological descriptors (second-order molecular connectivity index) were used as input parameters to the neural networks. Quantum chemical input descriptors included average molecular polarizability, dipole moments (total point charge, total hybridization, and total sum), ionization potential, and heat of formation. The fuzzy ARTMAP/QSPR correlated Henrys Law constant for -6.72 </= logH </= 2.87 with average absolute errors of 0.03 and 0.13 logH units for the overall data and the test set, respectively. The optimal 7-17-1 back-propagation/QSPR model was less accurate and exhibited larger average absolute errors of 0.28 and 0.27 logH units for the validation (recall) and test sets, respectively. The fuzzy ARTMAP-based QSPR was superior to the back-propagation and multiple linear regression/QSPR models for Henrys Law constant of organic compounds.

Fuzzy ARTMAP and Back-Propagation Neural Networks Based Quantitative Structure-Property Relationships (QSPRs) for Octanol-Water Partition Coefficient of Organic Compounds

Quantitative structure-property relationships (QSPRs) for estimating the logarithm octanol/water partition coefficients, logK(ow), at 25 degrees C were developed based on fuzzy ARTMAP and back-propagation neural networks using a heterogeneous set of 442 organic compounds. The set of molecular descriptors were derived from molecular connectivity indices and quantum chemical descriptors calculated from PM3 semiempirical MO-theory. Quantum chemical input descriptors include average polarizability, dipole moments, exchange energy, total electrostatic interaction energy, total two-center energy, and ionization potential. The fuzzy ARTMAP/QSPR performed, for a logK(ow) range of -1.6 to 7.9, with average absolute errors of 0.03 and 0.14 logK(ow) for the overall data and test sets, respectively. The optimal 12-11-1 back-propagation/QSPR model, for the same range of logK(ow), exhibited larger average absolute errors of 0.23 and 0.27 logK(ow) for the test and validation data sets, respectively, over the same range of logK(ow) values. The present results with the fuzzy ARTMAP-based QSPR are encouraging and suggest that high performance logK(ow) QSPR that encompasses a wider range of chemical groups could be developed, following the present approach, by training with a larger heterogeneous data set.

An integrated SOM-fuzzy ARTMAP neural system for the evaluation of toxicity

Self-organized maps (SOM) have been applied to analyze the similarities of chemical compounds and to select from a given pool of descriptors the smallest and more relevant subset needed to build robust QSAR models based on fuzzy ARTMAP. First, the category maps for each molecular descriptor and for the target activity variable were created with SOM and then classified on the basis of topology and nonlinear distribution. The best subset of descriptors was obtained by choosing from each cluster the index with the highest correlation with the target variable and then in order of decreasing correlation. This process was terminated when a dissimilarity measure increased, indicating that the inclusion of more molecular indices would not add supplementary information. The optimal subset of descriptors was used as input to a fuzzy ARTMAP architecture modified to effect predictive capabilities. The performance of the integrated SOM-fuzzy ARTMAP approach was evaluated with the prediction of the acute toxicity LC50 of a homogeneous set of 69 benzene derivatives in the fathead minnow and the oral rat toxicity LD50 of a heterogeneous set of 155 organic compounds. The proposed methodology minimized the problem of misclassification of similar compounds and significantly enhanced the predictive capabilities of a properly trained fuzzy ARTMAP network.

Modeling airborne benzene in space and time with self-organizing maps and Bayesian techniques

An assessment of personal exposure to airborne chemical contaminants demands for individual-specific registration of their concentrations, a procedure which is expensive and difficult to implement. An alternative approach is the calculation of a spatial concentration field in high resolution where exposure can be assigned to individuals according to their dwelling locations. Self-organizing maps (SOM) and Bayesian Hierarchical Models (BHM) were applied to model the spatial concentrations of benzene, an airborne volatile organic compound (VOC), in the urban area of Leipzig, Germany. Different performance measures (mean absolute error, coefficient of determination, etc.) were adopted to evaluate and compare the performance of SOM and BHM. Relevant input factors related to VOC dispersion were stepwise selected with the BHM. Both modeling techniques identified seasonal as well as spatial variations of benzene, with the highest concentrations occurring in winter and the lowest in summer. SOM and BHM showed that high concentrations of benzene are correlated with low distances to the city center and with the major traffic routes. Both SOM and BHM were suitable to model the spatial distribution of benzene concentrations, with the latter yielding a better overall performance using input factors selected by BHM. Beyond this specific application the suggested approaches have potential for statistical spatiotemporal modeling of other environmental parameters, an issue that is currently under rapid development.

Prediction of boiling points of organic compounds from molecular descriptors by using backpropagation neural network

The design and optimisation of industrial process require the knowledge of thermophysical properties. Available data banks can provide this information. However in specific cases, such as those related to drug activity or enviromental impact assessment, data are scarce and difficult or expensive to obtain experimentally. To overcome this lack of ready information, several thermodynamic models and correlations have been developed for a wide range of conditions. Among these models, the methods based on quantitative structure property relationships (QSPR) are promising. The basic concept of QSPR is to relate the structure of a compound with the property of interest. The compound’s structure is expressed in terms of molecular descriptors that characterise a given molecular feature. Molecular descriptors, such as the connectivity indices and the corresponding valence connectivity indices, that encode features such as size, branching, unsaturation, heteroatom content and cyclicity [1,2] are useful. For example, the first order connectivity index was used in 1982 to correlated the solubility of hydrocarbons in water [3]. The connectivity indices are based on local molecular properties and are bond-additive quantities so that in bonds of different kinds make different contribution to the overall molecular descriptors. The key step is to build the structure property relationship.