Hu · F1000Research 2014 · Retrospective database mining and chemoinformatics analysis · n=518 approved drugs

Monitoring drug promiscuity over time.

Cited 19 times in the scientific literature.

Level 5 - mechanism / opinion, no new human data

Chemoinformatics and database analysis of compound activity records; graded Level 5 by design analogy to non-clinical/bench data.

PubMed 25352982 · doi:10.12688/f1000research.5250.2 · record verified 2026-08-26

What was done

The authors assessed drug promiscuity over time by systematically collecting and mining activity records for 518 diverse approved drugs across time intervals from 2000 to 2014. They compared promiscuity rates derived from high-confidence activity data against estimates obtained when data selection criteria were progressively relaxed.

What was found

Using high-confidence activity data, average drug promiscuity increased from 1.5 to 3.2 targets per drug between 2000 and 2014. Reducing the stringency of data selection criteria produced substantial increases in calculated promiscuity, rising from ~6 targets per drug in 2000 to more than 28 targets per drug. Promiscuity rates derived from raw activity records differed significantly from the number of formally reported drug targets.

Why it matters

This study shows that real drug promiscuity rates are lower than commonly assumed, and that polypharmacology metrics heavily depend on the data-filtering criteria applied during computational mining.

Limits

The analysis relies entirely on historical database records, which reflect publication and testing biases toward heavily studied compounds. The abstract does not specify the specific databases mined, the exact quantitative thresholds defining high-confidence activity, or whether identified targets represent clinically meaningful in vivo interactions.

Cited by