In silico chemical testing is a useful tool to prioritize chemical risk assessment and can help making the animal and resource intensive traditional toxicological testing more efficient (see also Tox 21). In pharmaceutical research, in silico testing is a recognized tool to predict a chemical’s ligand affinity. In silico methods are thus also likely useful in food safety evaluations, in particular for predicting the ability of chemicals to interact with endocrine pathways. So far, a major limitation for the use of in silico testing methods in food safety has been the lack of a dataset of 3D small molecules relevant for food. Tiziana Ginex and colleagues from the University of Parma and Modena now developed a freely available database of 3D small molecules discussed in “FADB: a Food Additive Molecular Database for in silico screening in food toxicology” published on February 11, 2014 in Food Additives & Contaminants: Part A. FADB is a 3D version of the Everything Added to Food in the United States (EAFUS) database of 3969 substances authorized by the U.S. Food and Drug Administration (FDA). While food additives are defined as substances intentionally added to food to perform a function, they may also be used as indirect food additives (substances used in food contact materials) if they comply with the restrictions laid out in the respective rule in the U.S.. Accordingly, the database is also relevant for food contact materials. The authors excluded substances that could not be identified unambiguously by their Chemical Abstract Service Registry Number, reducing the original 3969 EAFUS substances to 2742. Ginex and colleagues further removed inorganic components of simple salts and small inorganic salts from the database, and split effective multi-molecular substances into individual 3D files identified by CAS RN plus a letter. For the majority of molecules, information was retrieved from PubChem, whereas the remaining substances were manually edited and added to the database. 3D chemical structures were energetically minimized to arrive at chemically correct and stable confirmations.

The database includes CAS RN, common name, synonym, E code, IUPAC Name, molecular formula, molecular weight, exact mass, LogP, Topological Polar Surface Area (TPSA), Molar Refractivity (MR), functional uses (from JECFA (Joint FAO/WHO Expert Committee on Food Additives)), SMILES and InCHI Code and 3D chemical structures in SD and Mol2 file formats of the final 2742substances. For most substances information on stereochemistry, regioisomeric forms,  and protonation states, which are particularly relevant for docking evaluations, are also available. The authors consider the use of virtual screening an effective approach for the identification of xenoestrogens and have identified the food additives propyl gallate and 4-hexylresorcinol as ERα ligands using a preliminary version of FADB. Different to existing databases, FADB focuses on substances relevant to food and combines 3D models of molecules with other relevant information, such as CAS numbers and functional uses. While the database is not exhaustive based on its small size, it allows for a high degree of efficiency and addresses problems of redundancy. Further, final users may apply modifications geared to their needs. The exclusion of ambiguous mixtures optimizes efficiency, but may also be a drawback as some substances used in food may not be covered. Further, EAFUS does not include all food additives, for example those are considered GRAS and have not been explicitly authorized by the FDA.


Ginex, T. et al. (2014). “FADB: a Food Additive Molecular Database for in silico screening in food toxicology.Food Additives and Contaminants: Part A (published online February 11, 2014).