Table 3. Comparison of our pipeline to existing drug-target prediction methods.

	Structures available	Genomic Rank	Structural Re-rank	HTD	PHM
All hits (n = 63)	69	22	24	56	23
Hits w/ known target structures (n = 53)	71	23	13	50	12
New predictions (n = 7)	73	28	31	n/a	n/a

The average ranking of the highest ranked known target is listed for all 63 validation ‘hits’, for the subset of 53 validation hits with known target structures, and for our seven predicted interactions. ‘Structures available’ indicates the average number of top-100 potential targets with available crystal structures for the compound set. Rankings are compared between the initial random-forest genomic ranking, the structural re-ranking of the top 100 RF predicted targets, the HTDocking server (HTD), and the PharmMapper server (PHM).