Mass Spectrometry in Systems Biology - Systems Biology: Networks, Models, and Applications - page 26

hypotheses involves prediction of how a system would respond to a
certain perturbation. Experiments can then be performed by applying
this very stimulus, monitoring responses, integrating data from differ-
ent omics approaches, and comparing the outcome with the initial
prediction in order to refine the hypothesis [4].
Eventually, computational modeling should monitor the system
response to perturbation and output. Experiments should identify
components that participate in biological process, quantify their
amounts, and describe their interactions through perturbation,
response, and output. In others words, we need to define each particu-
lar chemical structure (primary sequence for proteins), its quantity, and
what portion of that chemical structure transmits information through
interaction. High-throughput genomics and large-scale proteomics
and metabolomics experiments are capable of performing these tasks
and their data could be further processed to calculate the level of cor-
relation of trends and changes in quantitaties across different data sets.
We summarize here some examples of correlative studies between
mRNA and protein expression levels. Figure 1.6 illustrates the current
state on integration of omics data from the prospective mass spectrom-
etry. Two approaches characterize correlative studies of mRNA and
protein expression levels. First and most stringent is the derivation of
a correlation coefficient that is based on statistics of two global meas-
urements: quantitative proteomics (provided by stable isotope analogs
followed by mass spectrometry measurements) and microarray or
SAGE experiments (an interesting discussion on integration of high
throughput omics data is presented in ref. [87]). The second approach
uses an RNA/protein concordance test based on detection of quanti-
ties of these two bioanalytes by mass spectrometry and microarrays.
Eventually, in both cases, one uses a defined metric (for definition and
applications of metrics see ref. [88]) to cluster mRNA/protein data in
pathways, complexes, and so on, with the goal of extracting specific
features of the system under analysis.
In one mRNA/protein correlative study, Ideker et al. [89] measured
the perturbation applied to wild-type yeast and mutants by the
absence of galactose. Quantitative proteomics (ICAT, MS) and microar-
ray experiments showed a weak but significant correlation ( r 0.61) of

mRNA/protein expression levels for 289 genes. Correlation of the
observed gene products was superimposed to data describing protein-
protein interactions in yeast [90]. In general, gene products known to
interact from two-hybrid experiments showed increased correlation as
compared to the whole data set.
Washburn et al. [43] compared yeast grown in rich and minimal
media (enriched in 15 N). Quantitative MudPIT and oligonucleotide
arrays provided data for mRNA/protein expression levels plotted on
a log-log graph. Analysis of the graph showed a weak correlation

=