New GTEx Gene Expression Track for hg19 and hg38

[Christopher Lee]

We are excited to announce the release of a new gene expression track based on data from the NIH Genotype-Tissue Expression (GTEx) project. This track displays tissue-specific gene expression based on RNA-seq in 53 tissues from 570 donors obtained from the GTEx 'midpoint' data release (V6, October 2015). The original data for this track can be found at the GTEX Portal hosted by the Broad Institute.

This track also features a new gene expression display method that extends the traditional Genome Browser display — a horizontal bar graph. Every gene is annotated by a graph with colored bars, each of which corresponds to a specific tissue assayed by the GTEx project. Within a graph, the bar color indicates the tissue type, using GTEx conventions, and the bar height depicts the median expression level (in RPKM). To quickly view the tissue and expression level represented by a bar in the tracks display, mouse over the bar in the graph. The complete tissue color legend is shown on the track configuration page, and can also be popped up for viewing alongside the track. Below the bar graph, a line is shown indicating the gene extent that was used to generate the annotation, colored by gene class using GENCODE conventions (e.g. blue for protein-coding, green for non-coding).

User alert: In the figure above, do you notice how the bar graph sometimes extends past the righthand end of the associated gene annotation? This is because all bar graphs in the display are the same width (and have the same tissue ordering) to facilitate comparison. For example, in this figure (which may also be viewed here), the three tall bars in the TCAP annotation indicate this entire gene is highly expressed in cardiac and skeletal muscle. It does not indicate that there are expression peaks in the intergenic region beyond the gene!

Similar to other Genome Browser track displays, clicking on the graph will bring up a description page that shows a detailed box and whiskers plot of the RPKM data, as well as a description of the methods used to generate the data.

Credit goes to Kate Rosenbloom and Christopher Lee for the implementation and testing of this feature.