Database composition can affect the structure-activity relationship prediction

A model-free ensemble method for class prediction with application to biomedical decision making

OBJECTIVE

A classification algorithm that utilizes two-dimensional convex hulls of training-set samples is presented.

METHODS AND MATERIAL

For each pair of predictor variables, separate convex hulls of positive and negative samples in the training set are formed, and these convex hulls are used to classify test points according to a nearest-neighbor criterion. An ensemble of these two-dimensional convex-hull classifiers is formed by trimming the (m)C(2) possible classifiers derived from the m predictors to a set of classifiers comprised of only unique predictor variables. Because only two-dimensional spaces are required to be populated by training-set samples, the "curse of dimensionality" is not an issue. At the same time, the power of ensemble voting is exploited by combining the classifications of the unique two-dimensional classifiers to reach a final classification.

RESULTS

The algorithm is illustrated by application to three publicly available biomedical data sets with genomic predictors and is shown to have prediction accuracy that is competitive with a number of published classification procedures.

CONCLUSION

Because of its superior performance in terms of sensitivity and negative predictive value compared to its competitors, the convex-hull ensemble classifier demonstrates good potential for medical screening, where often the major emphasis is placed on having reliable negative predictions.

Using quantitative structure-activity relationships (QSAR) to predict toxic endpoints for polycyclic aromatic hydrocarbons (PAH)

Quantitative structure-activity relationships (QSAR) offer a reliable, cost-effective alternative to the time, money, and animal lives necessary to determine chemical toxicity by traditional methods. Additionally, humans are exposed to tens of thousands of chemicals in their lifetimes, necessitating the determination of chemical toxicity and screening for those posing the greatest risk to human health. This study developed models to predict toxic endpoints for three bioassays specific to several stages of carcinogenesis. The ethoxyresorufin O-deethylase assay (EROD), the Salmonella/microsome assay, and a gap junction intercellular communication (GJIC) assay were chosen for their ability to measure toxic endpoints specific to activation-, induction-, and promotion-related effects of polycyclic aromatic hydrocarbons (PAH). Shape-electronic, spatial, information content, and topological descriptors proved to be important descriptors in predicting the toxicity of PAH in these bioassays. Bioassay-based toxic equivalency factors (TEF(B)) were developed for several PAH using the quantitative structure-toxicity relationships (QSTR) developed. Predicting toxicity for a specific PAH compound, such as a bioassay-based potential potency (PP(B)) or a TEF(B), is possible by combining the predicted behavior from the QSTR models. These toxicity estimates may then be incorporated into a risk assessment for compounds that lack toxicity data. Accurate toxicity predictions are made by examining each type of endpoint important to the process of carcinogenicity, and a clearer understanding between composition and toxicity can be obtained.