. 2009 Jan 26;10:49. doi: 10.1186/1471-2164-10-49

Table 4.

Comparison of significant infection responses identified with SAS Proc Mixed LMMA or the non-parametric Wilcoxon signed rank test.

Data Pre-Processing^a	Genotype	Infection Response^b	Wilcoxon^c		LMMA^d		Number Matches^g	Percentage matches^h

			Significant^e	% sig^f	Significant	% sig
GC-RMA	V71-370	Upper vs. Mock	23,016	81.2	20,418	72.0	20,406	99.94
		Lower vs. Mock	23,739	83.7	21,772	76.8	21,749	99.89
	Sloan	Upper vs. Mock	24,369	86	22,758	80.3	22,712	99.8
		Lower vs. Mock	24,872	87.7	23,556	83.1	23,513	99.82
	VPRIL9	Upper vs. Mock	24,851	87.7	23,543	83.0	23,482	99.74
		Lower vs. Mock	25,260	89.1	24,180	85.3	24,087	99.62

MAS5	V71-370	Upper vs. Mock	22,300	78.7	19,486	68.7	19,478	99.96
		Lower vs. Mock	23,406	82.6	21,772	74.8	21,189	99.96
	Sloan	Upper vs. Mock	23,984	84.6	22,758	78.5	22,208	99.82
		Lower vs. Mock	24,578	86.7	23,556	81.7	23,153	99.94
	VPRIL9	Upper vs. Mock	24,714	87.2	23,543	82.5	23,366	99.91
		Lower vs. Mock	25,144	88.7	24,180	85.4	24,165	99.81

^aThe data used for the analyses were either preprocessed using GC-RMA or using MAS5 algorithm as described in the Methods. ^bInfection responses relative to mock-inoculated tissue were evaluated for the upper infection court (Upper vs. Mock) and the lower infection court (Lower vs. Mock). ^cWilcoxon signed ranks test implemented by the wilcox.test function of R package stats version 2.6.0. ^dLinear mixed model analysis performed in SAS Proc Mixed. Model: y = Genotype + Treatment + Time + Genotype × Treatment + Block + Block × Genotype + Block × Treatment + Block × Time + Block × Genotype × Treatment + Error, where y refers to the log2 scale median polish summarized gene expression values, the main factors (Genotype, Treatment and Time) and the interaction Genotype × Treatment were fixed factors while all the remaining terms were random factors. ^eTotal number of significant genes with TST-FDR adjusted p ≤ 0.01. ^fPercentage of the total number of detectable genes that are significant. ^gNumber of significant genes found with both Wilcoxon and LMMA. ^hThe number of matches found with both methods, as a percentage of the total number of significant genes found by Wilcoxon or LMMA, whichever was smaller.