Chemical Mixture Fingerprints and Applications
Chemical mixtures have important applications in pharmaceutical, cosmetics, flavor and other industries. Composition and stoichiometry of a mixture are required to have complete and correct representation of chemical mixture. We propose a mixture descriptor to encode both chemical mater present in a mixture and its stoichiometry. The proposed chemical mixture descriptor implementation can handle mixtures with arbitrary number of components and support additive and non-additive used defined models. Ranking and clustering of pharmaceutical formulations, bioassays, and scientific publications as potential applications of chemical mixtures descriptors are presented. Computation of mixture fingerprints and similarity coefficients is implemented on top of ChemAxon JChem library chosen for it comprehensive and robust API, which allowed for quick prototyping, deployment, and integration of the algorithm in existing workflows.