Tools, Applications, and Case Studies (q-RA and q-RASAR)

Java-based tools for quantitative read-across (Quantitative Read-Across v 4.2.1) and q-RASAR (RASAR v 3.0.2) have been developed and made available from the DTC Laboratory websites. The application of q-RA has been done for several nanotoxicity and ecotoxicity endpoints while q-RASAR has been successfully applied for the predictions of several endpoints of biological activity, toxicity, and materials properties. In most cases, q-RA and q-RASAR-based predictions showed superior quality results than the corresponding QSAR-derived predictions. The RASAR descriptors are generated for the query compounds not from the chemical structures of those compounds (unlike QSAR), but from their close congeners with a similarity consideration. Thus, the prediction aspect is included in the learning process, and the “prediction-inspired” modeling can give better quality predictions with the same quantum of chemical information compared to conventional descriptor-based QSAR modeling approaches. Thus, in most of the examples of modeling biological activity, toxicity, and materials property modeling using the q-RASAR technique presented in this chapter, the q-RASAR models show better quality of predictions compared to the corresponding QSAR models.