Table 1.

Motif associated with histological grades or prognosis identified based on independent datasets

	aMotif ID	bReproducibility	cP value for training data	dP value for test data
	JSP$NF_Y(20)	20	3.1 × 10-10	0.000158
	V$NRF1_Q6(10)	15	3.09 × 10-14	6.02 × 10-7
motifs associated with histological grades based on the data by Sotiriou et al.	V$ELK1_02(20)	12	9.25 × 10-26	1.41 × 10-6
	DME$CTTCCGSYN(5)	9	5.71 × 10-14	6.82 × 10-5
	V$E2F1_Q4_01(5)	7	5.71 × 10-15	0.002372
	JSP$NF_Y(10)	15	2.46 × 10-14	0.011049
	DME$RMSYSSARGCGC(5)	11	4.02 × 10-5	0.063412
	V$ELK1_02(10)	10	2.03 × 10-16	2.08 × 10-7
motifs associated with prognosis based on the data by Sotiriou et al.	DME$YYYGSGCMYGCG(5)	8	1.65 × 10-9	0.008054
	V$E2F1_Q4_01(10)	8	1.05 × 10-17	2.37 × 10-5
	V$IRF_Q6_01(10)	7	2.06 × 10-8	0.000152
	DME$NMSTTCYKSYR(5)	6	0.000669	0.084446
	V$NRF1_Q6(20)	6	9.02 × 10-22	1.31 × 10-6
	JSP$NF_Y(20)	22	5.93 × 10-8	0.01116
motifs associated with histological grades based on the data by Pawitan et al.	V$E2F1_Q4_01(5)	10	6.56 × 10-7	0.049423
	DME$RCRKGCGCAVN(5)	6	5.71 × 10-8	0.060899
	V$E2F1_Q4_01(15)	6	9.59 × 10-6	0.017285
	V$ELK1_02(20)	16	1.26 × 10-27	6.13 × 10-12
	V$NRF1_Q6(15)	11	9.2 × 10-23	3.89 × 10-7
motifs associated with prognosis based on the data by Pawitan et al.	V$NRF1_Q6(20)	11	4.31 × 10-22	2.49 × 10-7
	V$ELK1_02(15)	9	1.63 × 10-25	6.41 × 10-11
	DME$RCGCHKGCGY(5)	6	3.23 × 10-20	4.8 × 10-6

^aIDs starting from "V$", "JSP$", and "DME$" Motifs denote motifs from the TRANSFAC database, the JASPAR database, and our DME analysis, respectively, followed by values of the threshold parameter for motif searches in parentheses.

^bThe number of appearances of sequence feature in 30 searches with bootstrap resampling.

^cdP values calculated by Wilcoxon rank sum tests for training and test data, respectively.