A Multidimensional Matrix for Systems Biology Research and Its Application to Interaction Networks

A multidimensional matrix containing 76 parameters from
21 transcriptomics,
proteomics, interactomics, phenotypic and sequence-based data sets,
in which each data set covered most of the Saccharomyces cerevisiae proteome, was compiled
for systems biology research. The maximal information coefficient
(MIC) was used to measure correlations between every pair of parameters.
Out of 2850 possible comparisons, 340 pairs of variables (12%) showed
statistically significant MIC scores. There were 321 relationships
that were expected; these included relationships within physicochemical
parameters of proteins, between abundance levels of genes/proteins
and expression noise, and between different types of intracellular
networks. We found 19 potentially novel relationships between different
types of “-omics” data. The strongest of these involved
genetic interaction networks, which were correlated with pleiotropy
and cell-to-cell variability in protein expression. Protein disorder
also showed a number of significant relationships with protein abundance,
signaling and regulatory networks. Significant cross-talk was seen
between the signaling and kinase interaction networks. Investigation
of this revealed densely connected kinase clusters and significant
signaling between them, along with signaling centers that act as integrators
or broadcasters of intracellular information. These centers may allow
for redundancy and a means of dampening noise in networks under a
variety of genetic or environmental perturbations.