P-MTAB-25831 - P-MTAB-25831
Microarray data resulting from the Cy3 and Cy5 channels of each slide were always treated as single channel data, rather than two-channel data, facilitating more flexible analysis of experimental data. The data was analyzed as if derived from single colour arrays by combining input from individual arrays from any one slide. At least triplicate slides were used at each time-point investigated. In this study median net fluorescence intensities were initially converted to log2 scale. The log2 values were subjected to data normalization and transformation to diminish contributions from non-biological and biological variation, making signal intensities comparable across arrays, and achieving approximate normality and constant variance for subsequent statistical modelling. Typically the microarray data were subjected to z-score normalization, often followed by median subtract normalization or scaling between 0 and 1.Cluster analysis (k-means-, hierarchical clustering and self-organizing maps) are used to decrease the dimensionality of microarray data enabling isolation of genes showing similar time-course expression profiles during tooth development. Genes with a time-course of expression significantly similar to that of a pre-selected gene(s) throughout the time-course can also be found using the Profile Search feature in Spotfire v.9 Microarray Analysis Software. The expression pattern of the pre-selected gene(s) is called the master profile. The Profile Search tool calculates the similarity to this selected gene profile for all genes in the microarray data.