Electronic Northern

The term electronic Northern is used for the calculation of gene expression levels in different tissues, and is based on the number of ESTs for a given gene that have been identified in a set of cDNA libraries.
We connected the EST-data of the UniGene dataset with the tissue-definitions and library-mappings from the eVOC-ontology project (Kelso et al., Genome Res 13: 1222-1230, 2003). eVOC ontologies define a set of controlled terms to biological entities (tissues, cell types), which are hierarchically structured. In addition, mappings of these entities to dbEST libraries are provided.
The respective output shows the relative gene expression of one cDNA (mapped to the respective UniGene cluster) in each anatomical system, based on the mean expression of this cluster in all tissues of such particular anatomical system. The level of the relative gene expression is colour-coded and normalized for each tissue (see calculation). The datasets are updated when new versions become available.

Calculation

The first level of the ontology ‘Anatomical System’ was used as definitions for the tissues. First the libraries belonging to the respective tissues were determined using the eVOC annotation of the dbEST libraries. For each tissue, all libraries assigned to a particular tissue or to the respective tissues further downstream in the ontology-hierarchy were pooled.
Then the frequency of a UniGene cluster’s occurrence f_ugc in each tissue t was calculated:
img height=55 src="../../images/graphics/Enorthern1.gif" width=98 v:shapes="_x0000_i1025">
where N_ugc,t is the number of clones of the UniGene cluster in the tissue and N_t is the total number of clones in the tissue.
Next, the frequencies of all tissues were added up and divided by the number of tissues considered. This yielded the average frequency of occurrence for one UniGene cluster:

The relative gene expression of one UniGene cluster was then calculated by dividing the frequency of one tissue by the average frequency:

Thus ge_rel represents a value for the occurrence of a UniGene cluster in a tissue relative to the overall occurrence.

Display

Relative gene expression values are displayed as coloured boxes showing an abbreviation of the underlying tissue. Values < 1 (relative ‘under-expression’) are displayed in blue and values > 1 are displayed in red (relative ‘over-expression’). Darker colours represent a higher degree of under- or over-expression. The box has a white background when no EST from that particular tissue was identified in the respective UniGene cluster.