Journal of Cheminformatics

Table 5 Results on the test set for the ML models trained and validated on the combined PC-3/DU-145 dataset (PCC = 95)

From: Development of machine learning classifiers to predict compound activity on prostate cancer cell lines

Cell line	PC-3						DU-145
Algorithm	SVM		RF		KNN		SVM		RF		KNN
Activity threshold method	GAP 20	FIX 50	GAP 20	FIX 50	GAP 20	FIX 50	GAP 20	FIX 50	GAP 20	FIX 50	GAP 20	FIX 50
N compounds	461	587	461	587	461	587	485	586	485	586	485	586
N active	286	351	286	351	286	351	286	336	286	336	286	336
N inactive	175	236	175	236	175	236	199	250	199	250	199	250
Accuracy	0.65	0.65	0.67	0.65	0.64	0.63	0.66	0.66	0.67	0.65	0.64	0.63
Precision	0.85	0.82	0.81	0.74	0.80	0.73	0.83	0.79	0.78	0.72	0.74	0.71
Recall	0.54	0.53	0.61	0.63	0.57	0.58	0.54	0.54	0.61	0.63	0.58	0.59
F1	0.66	0.64	0.69	0.68	0.67	0.65	0.65	0.64	0.69	0.67	0.65	0.65
MCC	0.38	0.36	0.36	0.30	0.33	0.27	0.38	0.36	0.36	0.29	0.29	0.27
ROCAUC	0.69	0.68	0.68	0.65	0.67	0.64	0.69	0.68	0.68	0.65	0.65	0.64
TNR	0.84	0.82	0.76	0.68	0.76	0.69	0.84	0.81	0.75	0.66	0.71	0.68
FPR	0.16	0.18	0.24	0.32	0.24	0.31	0.16	0.19	0.25	0.34	0.29	0.32
FNR	0.46	0.47	0.39	0.37	0.43	0.42	0.46	0.46	0.39	0.37	0.42	0.41
TPR	0.54	0.53	0.61	0.63	0.57	0.58	0.54	0.54	0.61	0.63	0.58	0.59

Back to article page

ISSN: 1758-2946

Contact us

Submission enquiries: journalsubmissions@springernature.com