Table 2.

The top 20 weighted genes selected from different machine-learnings.

LASSO		PCA		GBM		RF		NN		SVM
Genes	Weight	Genes	Weight	Genes	Weight	Genes	Weight	Genes	Weight	Genes	Weight
S100P	0.52	C4BPA	1.85	OLFM4	1	OLFM4	3.89	TUBB2A	− 2.39	OLFM4	8.87
RARRES3	0.42	RIPK2	1.85	HLA-DMA	0.23	C4BPB	3.7	TIMP1	2.25	C4BPB	3.37
IFITM3	− 0.31	PYY	1.85	C4BPB	0.2	ISG20	1.63	CCL19	− 2.25	NMI	2.13
CD19	0.29	REG3A	1.85	NMI	0.18	DMBT1	2.43	DEFA6	− 2.01	HLA-DMA	1.96
CHAD	− 0.28	DUSP10	1.85	CLDN8	0.13	CXCL1	1.08	CD55	1.87	VNN1	1.78
NMI	0.24	CNTNAP2	1.84	VNN1	0.12	CLDN8	2.46	CXCL9	1.77	DEFA5	1.78
PLA2G2A	− 0.24	ATP2C2	1.84	HYOU1	0.11	LCN2	0.69	IFITM1	1.7	S100P	1.77
C4BPB	0.19	LRRN2	1.84	DEFA5	0.1	PRDX1	2.67	PCBP1	1.65	PRDX1	1.65
HYOU1	0.19	CHI3L2	1.83	PRDX1	0.1	GNA15	1.01	AQP8	1.64	CLDN8	1.55
VNN1	0.18	TRIM22	1.83	NPTX2	0.08	S100P	2.44	FTL	1.48	REG3A	1.38
NPTX2	0.18	ALOX5	1.83	S100P	0.08	IFITM1	2.03	ASS1	1.4	IRF9	1.34
DMBT1	0.17	OAZ1	1.83	RARRES3	0.08	NMI	3.47	HSPA5	1.34	HYOU1	1.32
OLFM4	0.15	ZNF189	1.82	CXCL1	0.07	RARRES3	1.96	ADM	− 1.34	CXCL1	1.2
CSF2RB	0.15	STAT3	1.82	DEFA6	0.05	MAP2K1	0.93	C4BPB	1.33	NPTX2	1.14
COL6A3	− 0.12	ZNF143	1.82	REG3A	0.05	LYN	1.54	ISG20	1.31	CD55	1.1
PCK1	− 0.11	GPR161	1.82	CHAD	0.05	STAT3	1.35	SDCBP	1.25	RARRES3	0.94
SERPINA3	− 0.08	SWAP70	1.82	VOPP1	0.04	TIMP1	1.23	REG1B	− 1.19	ISG20	0.86
CLDN8	− 0.05	ME1	1.82	CD19	0.04	CD55	1.45	TRIM22	− 1.17	CD19	0.86
COL4A2	0.04	BIRC3	1.82	PCK1	0.04	HLA-DMA	2.11	SERPINA3	1.09	HLA-DRA	0.85
SPINK4	− 0.04	ADRA2A	1.81	HLA-DRA	0.04	S100A8	0.64	CTSK	1.07	SELL	0.81

LASSO, Least Absolute Shrinkage and Selection Operator; PCA, principal component analysis; GBM, Gradient boosting machine; RF, Random forest; NN, Neural network, SVM, Support Vector Machine.

Different MLS process different weights, and negative weights in LASSO and NN that we sort the weighted genes with absolute value.