RESUMEN

The commercial applications of nanoparticles are growing rapidly, but we know relatively little about how nanoparticles interact with biological systems. Their value--but also their risk--is related to their nanophase properties being markedly different to those of the same material in bulk. Experiments to determine how nanoparticles are taken up, distributed, modified, and elicit any adverse effects are essential. However, cost and time considerations mean that predictive models would also be extremely valuable, particularly assisting regulators to minimize health and environmental risks. We used novel sparse machine learning methods that employ Bayesian neural networks to model three nanoparticle data sets using both linear and nonlinear machine learning methods. The first data comprised iron oxide nanoparticles decorated with 108 different molecules tested against five cell lines, HUVEC, pancreatic cancer, and three macrophage or macrophage-like lines. The second data set comprised 52 nanoparticles with various core compositions, coatings, and surface attachments. The nanoparticles were characterized using four descriptors (size, relaxivities, and zeta potential), and their biological effects on four cells lines assessed using four biological assays per cell line and four concentrations per assay. The third data set involved the biological responses to gold nanoparticles functionalized by 80 different small molecules. Nonspecific binding and binding to AChE were the biological endpoints modelled. The biological effects of nanoparticles were modelled using molecular descriptors for the molecules that decorated the nanoparticle surface. Models with good statistical quality were constructed for most biological endpoints. These proof-of-concept models show that modelling biological effects of nanomaterials is possible using modern modelling methods.

Asunto(s)

RESUMEN

A Gaussian process method (GPM) is described and applied to the production of some QSAR models. These models have the potential to solve a number of problems which arise in QSAR modeling in that no parameters have to be supplied and only one hyperparameter is used in finding the optimal solution. The application of the method to QSAR is illustrated using data sets of compounds active at the benzodiazepine and muscarinic receptors as well as the data set of the toxicity of substituted benzenes to the ciliate, Tetrahymena Pyriformis.

RESUMEN

We have used a new, robust structure-activity mapping technique, a Bayesian-regularized neural network, to develop a quantitative structure-activity relationships (QSAR) model for the toxicity of 278 substituted benzenes toward Tetrahymena pyriformis. The independent variables used in the modeling were derived solely from the molecular structure, and the model was tested on 20% of the data set selected from the whole set by cluster analysis and which had not been used in training the network. The results show that the method is robust and reliable and give results for mixed class compounds which are comparable to earlier QSAR work on single-chemical class subsets of the 278 compounds and which employed measured physicochemical parameters as independent variables. Comparisons of Bayesian neural net models with those derived by classical PLS analysis showed the superiority of our method. The method appears to be able to model more diverse chemical classes and more than one mechanism of toxicity.

Asunto(s)

RESUMEN

We describe the use of Bayesian regularized artificial neural networks (BRANNs) coupled with automatic relevance determination (ARD) in the development of quantitative structure-activity relationship (QSAR) models. These BRANN-ARD networks have the potential to solve a number of problems which arise in QSAR modeling such as the following: choice of model; robustness of model; choice of validation set; size of validation effort; and optimization of network architecture. The ARD method ensures that irrelevant or highly correlated indices used in the modeling are neglected as well as showing which are the most important variables in modeling the activity data. The application of the methods to QSAR of compounds active at the benzodiazepine and muscarinic receptors as well as some toxicological data of the effect of substituted benzenes on Tetetrahymena pyriformis is illustrated.

RESUMEN

We describe the use of Bayesian regularized artificial neural networks (BRANNs) in the development of QSAR models. These networks have the potential to solve a number of problems which arise in QSAR modeling such as: choice of model; robustness of model; choice of validation set; size of validation effort; and optimization of network architecture. The application of the methods to QSAR of compounds active at the benzodiazepine and muscarinic receptors is illustrated.

Asunto(s)

RESUMEN

A holographic neural network has been investigated for use as a discriminant. Six sets of artificial data and two data sets of infrared spectra, reduced using principal component analysis, of prepared cervical smears were analyzed by four regular discriminant methods as well as by the holographic neural network method. In all cases, it was found that the holographic neural network method gave comparable, and in some cases superior, results to the other discriminant methods. The holographic neural network method is simple to apply and has the advantage that it can be easily refined when new data become available without disturbing the original mapping. It is suggested that the holographic neural network method should be seriously considered when discrimination methods need to be applied.

Asunto(s)

RESUMEN

Cross-validated and non-cross-validated regression models using principal component regression (PCR), partial least squares (PLS) and artificial neural networks (ANN) have been used to relate the concentrations of polycyclic aromatic hydrocarbon pollutants to the electronic absorption spectra of coal tar pitch volatiles. The different trends in the cross-validated and non-cross-validated results are discussed as well as a method for the production of a true cross-validated neural network regression model. It is shown that the methods must be compared through the errors produced in the validation sets as well as those given for the final model. Various methods for calculation of errors are described and compared. The separation of training, validation and test sets into fully independent groups is emphasized. PLS outperforms PCR using all indicators. ANNs are inferior to multivariate techniques for individual compounds but are reasonably effective in predicting the sum of PAHs in the mixture set.

Asunto(s)