Table 2.

Highest and lowest correlation among spot intensities between technical replicates in a 2D gel experiment; highest and lowest Kappa coefficients between technical replicates

Sample	No of Replicates	Highest Correlation Coefficient (R-sq%)	Lowest Correlation Coefficient (R-sq%)	Highest Kappa 95% CI	Lowest Kappa 95% CI
6 (GSE)	3	0.774 (60%)	0.547 (30%)	0.4962 (-0.1039,1.000)	0.1476 (-0.1374,0.4326)
7 (GSE)	4	0.907 (82%)	0.589 (35%)	0.4437 (0.1004,0.7870)	-0.0767 (-0.114,-0.0350)
8 (GSE)	4	0.932 (87%)	0.617 (38%)	0.4445 (0.1027,0.7864)	0.0099 (-0.0855,0.1052)
9 (GSE)	2	0.747 (56%)	0.747 (56%)	0.1299 (-0.1384,0.3982)	0.1299 (-0.1384,0.3982)
10 (GSE)	2	0.837 (70%)	0.837 (70%)	0.1231 (-0.0486,0.2948)	0.1231 (-0.0486,0.2948)
22 (CONT)	4	0.805 (65%)	0.467 (22%)	0.6538 (0.3705, 0.9372)	0.2599 (-0.0566,0.5765)
23 (CONT)	4	0.845 (71%)	0.632 (40%)	0.3269 (0.0304,0.6235)	0.0322 (-0.0296,0.0941)
24 (CONT)	2	0.711 (50%)	0.711 (50%)	0.0743 (-0.1284,0.2770)	0.0743 (-0.1284,0.2770)
25 (CONT)	3	0.837 (70%)	0.524 (27%)	0.2843 (0.0073,0.5613)	0.2384 (-0.0629,0.5397)
26 (CONT)	2	0.578 (33%)	0.578 (33%)	0.1946 (0.0531,0.3361)	0.1946 (0.0531,0.3361)

The Pearson correlation coefficient is a measure of the linear relationship between two variables. R square, the square of Pearson's correlation is a measure of how much variability in one variable is explained by the variability in the other. Since technical replicates are expected to be identical, the r-squares are expected to be very high, at least 0.95. The table demonstrates the degree of variability between technical replicates after normalization. The Kappa coefficients with the 95% confidence intervals confirm the same thing. Ten out of sixteen confidence intervals span zero, indicating no agreement between technical replicates of the same sample in those cases.