Relating Anatomical Therapeutic Indications by the Ensemble Similarity of Drug Sets

The
anatomical therapeutic chemical (ATC) system is a world standard to
define drug indications. Despite its broad applications in pharmaceutical
and biomedical research, only a few studies that examine the relationships
among ATC classes have been published. Here we present a similarity-based
approach, named the indication similarity ensemble approach (iSEA),
that innovatively correlates ATC classes by their drug set similarity.
Our study demonstrated that iSEA was capable of relating ATC classes,
and these relationships could accurately assign the right indications
for approved drugs and make reasonable predictions about possible
clinical indications for unclassified drugs, which would provide valuable
information for drug repositioning. Additionally, on the basis of
iSEA, we constructed the first ATC relationship network to reflect
correlations among ATCs from a network view, which would further render
novel insight to understand the intrinsic relationships in the ATC
system.