Incorporating genetic similarity of auxiliary samples into eGene identification under the transfer learning framework

Table 3 Number of significant genes identified by TLegene for all the ten TCGA cancers

Cancer	TLegene				G₁	G₂	Linear regression (%)	PancanQTL (%)
Cancer	oScore	aScore	fScore	HMP	G₁	G₂	Linear regression (%)	PancanQTL (%)
ACC	2	2	2	2	2	2	2 (100)	2 (100)
BRCA	72	73	85	83	68	87	74 (85.1)	80 (92.0)
COAD	39	40	35	39	34	41	34 (82.9)	36 (87.8)
LIHC	13	13	9	12	9	13	8 (61.5)	13 (100)
LUAD	68	67	65	69	60	71	54 (76.1)	69 (97.2)
LUSC	39	40	40	40	37	42	32 (76.2)	40 (95.2)
OV	9	9	10	10	9	10	7 (70.0)	10 (100)
PAAD	42	43	40	41	36	47	38 (80.9)	41 (87.2)
STAD	21	22	24	24	21	24	20 (83.3)	22 (91.7)
UCEC	5	5	4	5	3	6	5 (83.3)	4 (66.7)
Total	310	314	314	325	279	343	274 (79.9)	317 (92.4)

G₁: the number of eGenes simultaneously identified by all the four TLegene tests; G₂: the number of eGenes identified by any of the four tests; the last two columns denote the number (proportion) of eGenes replicated by the traditional eGene identification method using linear regression or PancanQTL

ISSN: 1479-5876