These sites represent those with moderate and low levels of error variance, relative to additive genetic variance, (see Table 4).
For the purpose of this study, we will refer to this variance as drift-variance (see Materials and methods).
Given the relationship between 500 K and 100 K performance (see below), we would expect to see similar estimates of microarray variance for the 500 K microarray.
Direct comparison of batch-corrected and non-batch-corrected data (see Additional file 9) indicates that this correction dramatically improved correlation of several genes in the CD14 dataset while the median gene correlation decreased slightly, possibly due to a reduction in covariance that would be predicted if batch correction reduced microarray variance in comparison to noise levels.
Of the many useful tests used to detect regulated genes from a small number of microarray replicates, we see the intensity-based variance estimation and Z-statistic described in this study to be a good combination of simplicity, robustness, precision, and accuracy.
For additional information on microarray processing, see SI text.
High experimental variance associated with the microarrays and the dyes, notably, is not specific to this work but is a general property of microarrays experiments (see, for instance, [36]), making it important to consider these sources of variance in the ANOVA model.
For microarray data analysis, see the Supplemental Experimental Procedures.
Statistical analyses of the microarray were stepwise (see Table 2).
For SBP, 48.8% of the total variance was due to variance between patients, whereas 51.2% of the total variance was seen within patients.
