Table S5.

Dependence of calculated precision/enrichment factor of CoMOAdrug and CoVLS for the top 20 drugs on the cutoff of the number of known drugs associated with a disease.

Drug # cutoff	# of evaluated indications	MEDICASCY^a	CoMOAdrug	CoVLS
20	1029	0.273/3.10	0.270/4.53	0.305/5.03
30	891	0.309/3.48	0.304/4.61	0.346/5.29
40	790	0.335/3.62	0.332/4.52	0.376/5.13
50	710	0.364/3.94	0.360/4.60	0.408/5.27
60	675	0.377/4.10	0.374/4.67	0.425/5.37
70	652	0.386/4.20	0.383/4.71	0.434/5.38
80	602	0.406/4.44	0.407/4.89	0.464/5.66
90	458	0.487/5.43	0.516/5.96	0.575/6.59
100	423	0.509/5.63	0.546/6.18	0.612/6.90
110	404	0.521/5.78	0.563/6.31	0.634/7.10
120	384	0.536/5.93	0.586/6.51	0.659/7.32
130	370	0.549/6.05	0.601/6.65	0.678/7.49
140	364	0.555/6.12	0.609/6.72	0.685/7.56
150	356	0.561/6.17	0.617/6.79	0.696/7.66
160	347	0.568/6.23	0.628/6.90	0.708/7.78
170	336	0.580/6.35	0.642/7.05	0.726/7.97
180	228	0.572/6.05	0.623/6.61	0.715/7.61
190	106	0.558/5.57	0.591/5.91	0.673/6.78
200	51	0.537/5.04	0.573/5.38	0.620/5.90

^aMEDICASCY results are included for comparison. Bold number is the maximal calculated precision (1^st number) of each method. For a given testing drug, any drugs with Tanimoto Coefficient > 0.8 to it are excluded from training. Enrichment factor (2^nd number) is defined as the following ratio: (number of predicted true positives within the top 20)/(number of true positives by randomly selecting 20 drugs from the 2,059 drugs).