Breakdown of the benchmark dataset according to the 14 main ATC classes.

aSee Eqs.2–3 for the definition about the number of virtual drugs, and its relation with the number of structural different drugs.
bOf the 3,883 structural different drugs, 3,295 belong to one class, 370 to two classes, 110 to three classes, 37 to four classes, 27 to five classes, and 44 to six classes. See Supporting Information S1 for the detailed drug codes listed in each of 14 ATC-classes.