Table 2. Breakdown of the 3,883 drug compounds in the benchmark dataset $\mathbb{S}$ according to the 14 ATC classes (cf. Eq.1).

Subset	Name	Number of drugs
${\mathbb{S}}_1$	Alimentary tract and metabolism	540
${\mathbb{S}}_2$	Blood and blood forming organs	133
${\mathbb{S}}_3$	Cardiovascular system	591
${\mathbb{S}}_4$	Dermatologicals	421
${\mathbb{S}}_5$	Genito-urinary system and sex hormones	248
${\mathbb{S}}_6$	Systemic hormonal preparations, excluding sex hormones and insulins	126
${\mathbb{S}}_7$	Antiinfectives for systemic use	521
${\mathbb{S}}_8$	Antineoplastic and immunomodulating agents	232
${\mathbb{S}}_9$	Musculo-skeletal system	208
${\mathbb{S}}_{10}$	Nervous system	737
${\mathbb{S}}_{11}$	Antiparasitic products, insecticides and repellents	127
${\mathbb{S}}_{12}$	Respiratory system	427
${\mathbb{S}}_{13}$	Sensory organs	390
${\mathbb{S}}_{14}$	Various	211
Number of total virtual drugs		4,912a
Number of total structural different drugs		3,883b

^a The number of virtual drugs is counted as follows: for a structurally same drug, its contribution to the total number of virtual drugs is 2 if it occurs in two different ATC classes; that is 3 if it occurs in three different ATC classes; and so forth.

^b Of 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.