Table 5.

Performance of Random Forest (RF) models averaged across 100 training runs, using different features as inputs

Input features	AUC	ACA $+$ (Class 1)		Scl $70+$ (Class 2)
		TNR$(\%)$	FPR$(\%)$	TPR$(\%)$	FNR$(\%)$
All 531 proteins	0.73	84	16	62	38
PCA	0.85	83	17	88	12
MDS	0.49	63	37	34	66
ISOMAP	0.54	58	42	49	51
LLE	0.56	54	46	58	42
t-SNE	0.48	37	63	59	41
UMAP	0.66	77	23	54	46
RF-MDG	0.77	68	32	86	14
RF-RFE	0.93	100	0	85	15
SVM-RFE	0.84	93	7	75	25
DCLDL	1.00	100	0	100	0