Table 3 

Summary of statistics measuring stability of gene expression 

GENE 
MEAN^{a} 
F^{b} 
MSEANOVA^{c} 
CV (%)^{d} 
SLOPE^{e} 
INTERCEPT 
R^{2} 
MSREG^{f} 
STABILITY INDEX^{g} 


UBQ 
15.8 
1.42 
0.53 
3.4 
0.11 
15.3 
0.56 
0.49 
0.37 
TUA 
28.9 
1.33 
1.56 
5.4 
0.10 
28.7 
0.28 
1.65 
0.54 
18S 
14.0 
4.72** 
0.58 
4.1 
0.20 
13.2 
0.52 
0.38 
0.83 
ACT2 
17.3 
1.29 
1.25 
7.2 
0.16 
16.6 
0.57 
1.14 
1.16 
UBQL 
19.8 
1.97 
1.57 
7.9 
0.17 
19.1 
0.35 
1.62 
1.35 
EF1β 
25.1 
3.76** 
1.87 
7.5 
0.28 
23.8 
0.6 
1.41 
2.09 
TUB 
17.7 
4.45* 
2.7 
15.3 
0.37 
16 
0.63 
1.92 
5.64 
ACT11 
22.8 
1.01 
5.71 
25.0 
0.24 
21.7 
0.37 
5.29 
6.01 
EIF4BL 
20.6 
11.44** 
3.22 
15.6 
0.52 
18.2 
0.68 
1.79 
8.13 
CYP 
19.2 
7.35** 
5.14 
26.8 
0.67 
16.2 
0.84 
2.66 
17.94 


^{a}Data based on analysis of C_{T }values. Genes are ordered, top to bottom, from those tending to show the highest stability to those showing the lowest, based on the stability index. ^{b}Approximate Ftests of variance among tissue samples tested. *, P < 0.05; **, P < 0.01. Degrees of freedom for numerator were 7 and for denominator were 40, except for 18S RNA where they were 7 and 16, respectively. ^{c}MSEANOVA represents variance among experiments and RTPCR reactions within experiments. 18S amplification was only included in one experiment; thus, MSEANOVA for 18S only represents within experiment variance. ^{d}Coefficient of variation (MSE divided by mean multiplied by 100). ^{e}Slope of regression of gene means (over experiments and samples within experiments) against overall means for the different samples. Intercepts and coefficient of determination (R^{2}) are also given for the estimated regression lines. ^{f}Mean square of deviation of means from estimated regression line (MSreg), which estimates the degree to which genes deviate from the linear model in their level of mean expression for a particular tissue sample. ^{g}Stability index is the product of CV and slope (multiplication of columns 4 and 5). Genes whose expression shows the lowest random variation within tissue samples due to variation among experiments or PCR reactions (MSEANOVA), and whose expression depends least in a predictable way on tissue sample (slope), are preferred as controls. 

Brunner et al. BMC Plant Biology 2004 4:14 doi:10.1186/14712229414 