Dimensionality of big data sets explored by Cluj descriptors

Abstract

Dimensionality of a relatively big data set (95 compounds) observed for toxicity (mutagenicity) was explored in order to compute QSAR models. Distinct molecular descriptors were used. Dimensionality of data, using PCA, correlation plots and clustering, was evaluated. Analyzing data dimensionality allowed model optimization. Docking studies and PCA were used in order to expand data dimensionality. Pearson correlation coefficient (r) values, obtained for both perceptive and predictive models, were satisfactory.