Modern medicine and an increasingly complex environment contribute to exposure of humans to a large number of chemical compounds, that can potentially be toxic. Although widely used, compound testing in animals has important limitations. In vitro testing provides a promising alternative. However, because of the relative inaccessibility and fragmentation of available data, the in vitro approach largely underperforms its potential. The aim of this study is to investigate how available public

more »

... e resources (tools and databases) support accessing and distribution of in vitro compound data. We examined 19 public online resources, mapped their features, and evaluated their usability with a set of four model compounds (aspirin, rosiglitazone, valproic acid, and tamoxifen). By investigating compound names and identifiers, we observed extensive variation and inconsistencies in available resources: the synonyms were different, compounds' structural identifiers (InChI, InChIKey, SMILES and IUPAC systematic name) underperformed in omics databases, identification of compound related metadata (e.g. concentrations used in the experiments) from omics experiments was complex and none of the available resources clearly distinguished between in vivo and in vitro data. In addition, we estimated accessibility of selected public resources using computational queries. Only a few public resources provided access to compound-related data using semantic web technology. The general quality of experiment annotations created further difficulties in identifying data of interest. Therefore, we identified several standardized ontologies with potential to provide an increased accuracy for extensive data retrieval of in vitro compound data. Furthermore, using the examples of our model compounds, we provide recommendations on the use of ontologies by suggesting specific ontology terms to annotate in vitro experimental data when being published.