Supplemental Table 1 -
Genes that show significant differences in their expression across MN cell types. Supplemental Table S1 is a 10 page table. The data in supplement 1 represent the 813 genes that are differentially expressed among cell types. The data are displayed in spreadsheet format with the probe sets listed alphabetically, according to class as defined in Figure 5. Each row represents one probe set, except for probe set 99056_at, whose name is too long to fit on one line. The information in the columns is as indicated. Statistically significant fold changes are shown in orange ink. Some genes are represented by multiple probe sets (highlighted with violet), and in 18 cases, each of the 2 probe sets for a single gene fell into a different class. When this occurred, the probe sets were moved to the same class if the distribution of signal intensities across cell types was the same for the 2 probe sets. These probe sets are indicated with asterisks (4930553; cyba; mgs; mmp9; nde1; zpf275; atp5c1; Ncam1; ndr4; rex3). While multiple probe sets for a given gene usually produced consistent results, the probe sets for 8 genes produced dissimilar signal distributions and were not moved to the same class. These sixteen probe sets (2 for each of the 8 genes) are indicated by red ink. These discrepancies could represent an alternately spliced and differentially expressed transcript. On the other hand, in many of these cases (5/8) one probe set in the pair reported a strong signal intensity, while the other probe set reported a very low, and perhaps inaccurate, signal.
In some cases, the genes in the list do not appear to be expressed in the LCM population. For example, the transglutaminase 3 gene (class V) is thought to be specifically expressed in the epidermis. The signal intensity associated with the probe set representing this gene is always at or below 20, and it is called absent. Therefore, the inclusion of this gene in the data set most likely represents a false positive (type II) error or some basal expression level. Because it is impossible to establish a criterion for measuring background noise that can be applied to all probe sets on an array, no genes were omitted from the list. Thus, we expect that we have excluded false negative (type I) errors from our data set at the expense of including false positive (type II) errors.
The abbreviations are:
G, golgi; N, nucleus; ER, endoplasmic reticulum; EC, extracellular; M, membrane; IC, intracellular; CySk, cytoskeleton; lys, lysosome; mit, mitochondria; mic, microsome
