TABLE 1.

Information of top 20 genes ranked according to AUC for single malignant cell distinguishing and genes included in the prediction model

Rank	Gene symbol	AUC	Sensitivity	Specificity	|log2FC|	Secreted or membrane‐bound proteins	Included in the regression model	Coefficients
1	KRT18	0.927	0.903	0.865	1.855	No	Yes	0.8449
2	KRT8	0.924	0.839	0.923	1.851	No	No	–
3	NAPSA	0.921	0.899	0.881	1.994	No	Yes	1.2584
4	MUC1	0.916	0.907	0.861	1.729	Yes	No	–
5	WFDC2	0.912	0.897	0.850	1.898	No	No	–
6	SFTPB	0.901	0.888	0.854	1.989	No	No	–
7	HOPX	0.900	0.879	0.827	1.693	No	No	–
8	KRT7	0.892	0.867	0.865	1.468	No	Yes	1.5488
9	EPCAM	0.885	0.885	0.837	1.485	Yes	No	–
10	KRT19	0.884	0.839	0.867	1.413	No	No	–
11	SFTA2	0.878	0.840	0.878	1.218	No	No	–
12	ZFP36L1	0.877	0.812	0.828	1.232	No	No	–
13	ERRFI1	0.874	0.842	0.831	1.671	No	No	–
14	C8orf4	0.873	0.840	0.833	2.120	No	No	–
15	ELF3	0.872	0.878	0.827	1.561	No	No	–
16	NKX2‐1	0.865	0.924	0.790	1.109	Yes	No	–
17	RNASE1	0.865	0.830	0.824	1.169	No	No	–
18	GPRC5A	0.864	0.813	0.781	1.025	Yes	Yes	0.6344
19	CLDN4	0.862	0.879	0.807	1.247	Yes	No	–
20	SLC34A2	0.861	0.875	0.822	1.461	Yes	No	–
…
34	IRX2	0.835	0.950	0.701	1.734	No	Yes	0.7221
…
64	CAPN8	0.806	0.949	0.652	0.942	No	Yes	0.4969
…						No
215	SPINK13	0.725	0.992	0.456	1.010	Yes	Yes	2.3251

Abbreviations: AUC, area under the curve; FC, fold change.