Table 3.

Comparison of preprocessing data sets by GC-RMA or MAS5 algorithms on the significance of fixed factors as revealed by LMMA analyses

FDR (TST)a	Fixed factorb	Number significant genesc		Number matchesd	Percentage matchese
		GC-RMA	MAS5
0.05	Genotype	27,156	26,651	25,896	97.2
	Treatment	27,732	27,585	27,099	98.2
	Genotype × Treatment	26,500	25,932	24,977	96.3
	Time	9,960	9,038	6,961	77
0.01	Genotype	25,923	25,162	24,024	95.5
	Treatment	27,139	26,864	26,079	97.1
	Genotype × Treatment	24,669	23,643	22,327	94.4
	Time	4,453	3,655	2,941	80.5
0.001	Genotype	24,191	23,161	21,780	94
	Treatment	26,351	25,962	24,873	95.8
	Genotype × Treatment	22,381	20,924	19,507	93.2
	Time	2,025	1,551	1,348	86.9
0.0001	Genotype	22,611	21,270	19,800	93.1
	Treatment	25,560	25,126	23,817	94.8
	Genotype × Treatment	20,293	18,691	17,256	92.3
	Time	1,173	857	768	89.6

^aFalse discovery rate (FDR) was controlled using the two-stage linear step-up method (TST). ^bThe fixed factors used in the LMMA model included genotype (V71-370, Sloan and VPRIL9), treatment (Upper, Lower, and Mock), time (9 am, 10:30 am, and 12 pm), and the genotype × treatment interaction. ^cNumber of significant genes for a factor was determined by LMMA analysis with FDR control. ^dThe number of genes called significant following GC-RMA and following MAS5 pre-processing. ^eThe number of matches as a percentage of the minimum of the number of significant genes called following GC-RMA or MAS5 preprocessing.