Target prediction by cascaded self-organizing maps for ligand de-orphaning and side-effect investigation
Journal of Cheminformatics volume 6, Article number: P47 (2014)
Computational chemogenomics approaches have emerged as a means to predict modulations of biomolecules by ligands. We implemented a method for the prediction of the macromolecular targets of small molecules combining state-of-the-art approaches that compare physicochemical properties and pharmacophoric features of query molecules with known drugs. Investigating similarity from multiple vantage points has been shown to increase the prediction accuracy in a retrospective evaluation. The method has been applied in multiple projects to “de-orphan” molecules with unknown main target and investigate potential side-effects of drug candidates. In a first application, the method identified a molecular scaffold as a potentially privileged structure of druglike compounds for chemoresistant tumor therapy . In a second project, the tool revealed the potential of up to 5% of known bioactive substances to have unrecognized epigenetic effects by modulating histone deacetylase (HDAC) activity – thereby stressing the importance of probing for epigenetic effects in long-term drug toxicity studies .
Reutlinger M, Koch CP, Reker D, Todoroff N, Schneider P, Rodrigues T, Schneider G: Mol Inf. 2013, 32: 133-138. 10.1002/minf.201200141.
Lötsch J, Schneider G, Reker D, Parnham MJ, Schneider P, Geisslinger G, Doehring A: Trends Mol Med. 2013,
About this article
Cite this article
Reker, D., Rodrigues, T., Schneider, P. et al. Target prediction by cascaded self-organizing maps for ligand de-orphaning and side-effect investigation. J Cheminform 6 (Suppl 1), P47 (2014). https://doi.org/10.1186/1758-2946-6-S1-P47