RePairDB: An interventional clinical trial drug-disease dataset for evaluating repurposing outcomes.

Description

RePairDB (Repurposing Pairs Database) is a dataset of drug-disease pairs drawn from registered interventional clinical trials. It holds 168,742 pairs covering 8,310 drugs and 4,772 disease terms, assembled from 156,388 trials registered on ClinicalTrials.gov, in one dataset of 56 documented columns. Each row describes a single drug-disease combination: how many trials tested it and the highest phase it reached, whether it is an approved indication and which regulator or drug label records that, whether trials of it were terminated or suspended for lack of efficacy or on safety grounds, the first and last dates it was tried, whether the disease is rare, and the ClinicalTrials.gov identifier of every trial behind the row. Each row also carries ChEMBL, DrugBank, InChIKey, MONDO and MeSH identifiers, so it can be joined to other resources. The dataset is meant for drug repurposing research and for work on clinical trial outcomes. It offers a reference set for testing computational repurposing predictions against combinations that have a real clinical history, a supply of negative examples that were genuinely given to patients rather than assumed inactive, and a basis for studying why trials stop at the level of the drug-disease pair rather than the individual trial. Rare-disease pairs are flagged, as are the pairs supported by trials in which the drug was the treatment under test rather than a comparator, so either can be selected directly. It is built from public releases of AACT (ClinicalTrials.gov), the Open Targets Platform, MONDO and MeSH. The deposit contains the data file, the Python code that reproduces it from those sources, and a user guide documenting every column, definition and rebuilding guide.

Institutions

• Khulna University

  Khulna Division, Khulna

Categories

Licence