Identification and analysis of promiscuity cliffs formed by bioactive compounds and experimental implications

Abstract

Multi-target activities of small molecules must be distinguished from apparent promiscuity resulting from assay artifacts. The molecular origins of specific multi-target activities are currently poorly understood. Compounds from the medicinal chemistry literature with available high-confidence activity data were systematically searched for ‘promiscuity cliffs’, defined as pairs of structural analogs with large differences between the number of targets they are active against. During the search, compounds with detectable aggregator properties, pan-assay interference characteristics, or other possible chemical liabilities were eliminated. A large number of promiscuity cliffs remained, many of which were centered on a limited number of highly promiscuous compounds, as revealed by network representations. The analysis of promiscuity cliffs often suggested follow-up experiments to further explore the molecular basis of promiscuity and assess the influence of data sparseness. Therefore, promiscuity cliffs identified herein are made freely available to support follow-up investigations.