One core visualization tool developed at Molquant is the Bathymetry plot, a two dimensional matrix cluster diagram to visualize up to thousands of gene expression correlations simultaneously.  Each queried gene is represented by and plotted as a network comprising the top "n" correlations to the queried gene. Each gene in the plot is correlated with every other gene and correlation values are represented as color: brighter color = higher correlation.  Each network is also clustered by correlation  similarity such that correlating networks typically group together (not always, depending on the biology and clustering algorithm used).   With these plots, inferring biological function relies on the widely used concept that genes involved in common biological processes exhibit correlated expression (guilt by association).  We have carefully curated, selected, and edited data from large publicly available datasets to form the basis for generating correlations.

Glycolytic pathway genes exhibit correlated expression in 1200 samples

Assembly of a Bathymetry plot


Using biologically-seeded correlation and a large amount of data, genome-wide correlation networks are established for selected genes or biological processes.  To visualize the relationships among any set of networks, correlations among genes of each network are plotted in a two dimensional matrix as a "bathymetry" plot  using color and height to represent the degree of correlation among the networks.    Shown here are relationships among eight Parkinson's Disease-associated genes plotting both gene::gene relationships and linkage to Parkinson's Disease-associated biological processes.  These data indicate that Parkinson's associated genes LRRK2, PARK2, PINK1 and newly reported NOVA2 (M.Lin et al., ASHG, 10/13) form a tight network, and includes linkage to lysosomal function.  MAPT and SYNJ1 exhibit weak linkage to the LRRK2/PARK2/PINK1/NOVA2 group.  SNCA exhibits linkage to Axon Motor function, whereas PARK7 is tightly associated with mitochondria associated genes.

Small Scale High Resolution.  12 Networks comprising 120 genes plotted.

Large Scale Bathymetry Plot.  351 networks comprising 3510 genes plotted.

A list of 351 genes associated with Intellectual Disability were compiled from Reviews/OMIM.  Networks were generated for all 351 genes and 10 core genes from each network were plotted.  Network clusters enriched for Mitochondrial Respiration genes, Chromatin Modification and Neuronal Synapse genes are indicated.


To visualize gene expression and gene signatures across diverse datasets, Gene Expression Plots display expression heatmaps ordered by data subtype (e.g. tumor type) along with quantitative summed score of included genes.

The plot below shows a 23 gene signature specific to ovarian cancer, demonstrating selectivity among more than 9000 tumors representing 30 tumor types.  The mean signal for ovarian cancer is more than 2.5 fold higher than the mean for the next highest tumor type, endometrial cancer.