Table 4. CCLE Drug sensitivity prediction results in the form of correlation coefficients between experimental and predicted sensitivities.

	Correlation Co-efficients
Drug Name	Elastic Net [7]	CRF-400	CRF-20,000
17AAG	0.43	0.4116	0.4397
AEW541	0.33	0.4037	0.3934
AZD0530 (Saracatinib)	0.18	0.2855	0.2747
AZD6244 (Selumetinib)	0.58	0.516	0.5909
Erlotinib	0.3	0.4034	0.4333
Irinotecan	0.68	0.6214	0.6776
L685458	0.47	0.5351	0.5423
Lapatinib	0.45	0.5488	0.5263
LBW242	0.08	0.184	0.1400
Nilotinib	0.76	0.5458	0.5476
Nutlin3	0.1	0.2892	0.3096
Paclitaxel	0.6	0.5453	0.5531
Panobinostat	0.65	0.616	0.6503
PD0325901	0.64	0.5837	0.6471
PD0332991	0.58	0.5077	0.5141
PF2341066 (Crizotinib)	0.36	0.5121	0.5055
PHA665752	0.27	0.3393	0.3437
PLX4720	0.55	0.4459	0.4768
RAF265	0.35	0.4378	0.4394
Sorafenib	0.27	0.4099	0.4685
TAE684	0.35	0.4073	0.4453
TKI258 (dovitinib)	0.3	0.4463	0.4611
Topotecan	0.58	0.5619	0.6226
ZD6474 (Vandetanib)	0.24	0.3331	0.3494
Average	0.421	0.454	0.473

Elastic Net denotes the approach applied in [7] for predicting sensitivity using 10-fold cross validation from CCLE database. CRF-400 denotes our proposed combined Random Forest approach using gene expression and SNP6 data of only 400 genes. CRF-20000 denotes our proposed combined Random Forest approach using 18,988 gene expression and 21,217 SNP6 features.