To provide a more complete representation of the therapeutic space, we recently expanded our entities to include drugs from the ChEMBL database (13). ChEMBL curates and aggregates bioactive molecules with drug-like properties, as well as records from different public resources including Drugs@FDA, ClinicalTrials.gov and DailyMed, among others. The new drug index consists of all parent molecules with known pharmacological action or disease indication to a total of 6515 entries belonging to seven different modalities. Among the most relevant drug information, users can find curated mechanisms of action, approved or experimental indications, small molecule representations, synonyms and trade names. Moreover, we expanded the drug annotation with a statistical analysis on post-marketing significant adverse drug reactions (ADRs) from FDA Adverse Event Reporting System (FAERS) (14). From the >12 million publicly available reports in FAERS, we filtered the most reliable entries following similar published approaches (15). The significant drug-ADR pairs were then evaluated using a Likelihood Ratio Test (LRT) and critical values inferred using a Montecarlo simulation (16). The significance of a given drug-ADR is implicitly corrected by how often a drug is found in a report and how often an event is reported across drugs (Figure 2).
