Table 3. Correlation coefficient, mean absolute pairwise deviation, and repeatability statistics, σ for clones across arrays.

	Correlation coefficient		Mean absolute pairwise deviation		Repeatability statistic, σ
	All	Filtered	All	Filtered	All	Filtered
Mopo	0.646 (0.334)	0.676 (0.347)	0.598 (0.425)	0.510 (0.389)	0.588 (0.355)	0.470 (0.310)
Mopo-clin	0.527 (0.398)	0.634 (0.469)	0.929 (0.629)	0.565 (0.343)	0.855 (0.507)	0.539 (0.359)
Lymphoma	0.714 (0.269)	0.777 (0.397)	0.519 (0.317)	0.366 (0.222)	0.518 (0.294)	0.335 (0.200)
NCI60	0.429 (0.285)	0.515 (0.686)	1.159 (0.277)	0.558 (0.635)	1.101 (0.223)	0.463 (0.538)
Prostate	0.293 (0.342)	NA	0.615 (0.313)	NA	0.580 (0.313)	NA
DNR	0.592 (0.628)	0.593 (0.627)	0.900 (1.038)	0.899 (1.038)	0.901 (0.746)	0.889 (0.758)

The numbers are average values with standard errors in parentheses. For correlation coefficients, a clone average, across correlation coefficients from all pairs for the clone, was first calculated before a non-weighted average and standard error across clones was calculated for each data set. The mean absolute pairwise deviation was calculated as the average across all combinations of same-clone spot-pair and array within each data set. The ‘Repeatability statistic, ’ column shows the average for each data set of estimates of the standard deviations in the ANOVA model calculated for each repeatedly spotted clone (high quality corresponds with a small value for the repeatability coefficient). For each statistic, the value as calculated from all available data points (excluding spots that were below background or manually flagged) as well as the value calculated from the ‘standard’ filter-criterion of SB ratio ≥1.4, are given in columns ‘All’ and ‘Filtered’, respectively. The repeatability coefficient (as defined in the Materials and Methods) can be obtained from the estimated standard deviations by multiplying the repeatability statistics by 2.83. The corresponding standard error can also be found by scaling with the same factor.