Description

This column shows microarray gene expression data from the Affymetrix All Exon dataset. An associated Genome Browser track is available on the Genome Browser. Please note that while the data in the Genome Browser is displayed at the exon level, the data in the Gene Sorter is displayed at the gene level.

By default, the column shows the ratio of expression of a gene in a specific set of tissues to the expression of the gene overall. A gene that is more highly expressed in a tissue is colored red, and a less expressed gene is shown in green. The values are colored on a logarithmic scale. This coloring is standard, but is the opposite of what an inexperienced user might expect: in this case, red means go and green means stop! Black indicates that a gene is neither over- nor under-expressed in the tissue. Uncolored boxes (white on most browsers) represent missing data.

Column Configuration

Several column configuration options are available on the Configuration page, which can be accessed via the configure button in the top menu.

Methods

RNA (from a commercial source) from 11 tissues were hybridized to Affymetrix All Exon 1.0 ST arrays. For each tissue, 3 replicate experiments were done for a total of 33 arrays. The arrays' raw signal intensity was normalized with a quantile normalization method, then run through the PLIER algorithm. The normalized data were then converted to log-ratios, which are displayed as green for negative log-ratios (under-expression), and red for positive (over-expression).

When calculating expression ratios, the overall expression level in the denominator was calculated by first taking the median of replicates for each tissue and then taking the median of these medians.

To associate particular genes with expression data, the following algorithm was used: for each gene, the overlapping exons from the microarray dataset were collected and the median intensity of these was assigned to the gene. If the gene contained exons not overlapping an exon from the microarray, those exons were ignored. If the microarray dataset contained exons not overlapping a gene, those were ignored as well. Once the overall expression level of each gene was determined, this was then log-ratio'd. In addition, the median was taken for replicate arrays per tissue for both the absolute and log-ratio varieties.

Credits

The data for this track was provided and analyzed by Affymetrix.

Links