Molecular similarity in computational toxicology

Abstract

The concept of chemical similarity has many applications in several fields of cheminformatics. One common use of chemical similarity measurements, based on the principle that similar molecules have similar properties, is in the context of the read-across approach, where estimates of a specific endpoint for a chemical are obtained starting from experimental data available from highly similar compounds. This chapter reports an implementation of chemical similarity and the analysis of multiple combinations of binary fingerprints and similarity metrics in the context of the read-across technique. This analysis demonstrates that the classical similarity measurements can be improved with a generalizable model of similarity. The approach presented here has been implemented in two open-source software tools for computational toxicology (CAESAR and VEGA).