Mapping the Universe - Mapping Scientific Frontiers: The Quest for Knowledge Visualization - page 81

Fig. 2.34 Musculoskeletal points (©1996 Terry Oleson, UCLA School of Medicine. http://www.
americanwholehealth.com/images/earms.gif )

2.4.3 Genomic Maps

Due to the publicity of the Human Genome project, genomic maps, gene expression
visualization, and bioinformatics have become the buzzwords in mass media.
Traditionally the common practice of analyzing expression data is done in a single
dimension. Single-dimensional analysis places genes in a total ordering, limiting the
ability to see important relationships.
Kim et al. ( 2001 ) visualize the C. elegans expression data in three dimensions.
Groups of related genes in this three-dimensional approach appear as mountains,
and the entire transcriptome appears as a mountain range. Distances in this synthetic
geography are related to gene similarity, and mountain heights are related to the
density of observed genes in a similar location. Expression visualization allows us
to hypothesize potential gene-gene relationships that can be experimentally tested.
To find out which genes are co-expressed, Kim et al. first assembled a gene
expression matrix in which each row represents a different gene (17,817 genes) and
each column corresponds to a different microarray experiment (553 experiments).
The matrix contains the relative expression level for each gene in each experiment
(expressed as log2 of the mornalized Cy3/Cy5 rations). They calculated the Pearson