Biology Reference
In-Depth Information
data from almost 2000 published microarrays from human tumors to
establish a compendium of modules combining genes with similar behav-
ior across arrays. This cancer module map allowed them to characterize
clinical conditions (like tumor stage and type) in terms of a profile of acti-
vated and deactivated modules. For instance, they found that a growth
inhibitory module, consisting primarily of growth suppressors, was coor-
dinately suppressed in a subset of acute leukemia arrays, suggesting a pos-
sible mechanism for the uncontrolled proliferation in these tumors.
These and other results 42-45 illustrate the value of analyzing the com-
plex processes underlying biological conditions such as human diseases in
terms of transcription modules. Yet, while a modular characterization of
such processes provides a powerful tool to elucidate aspects of the nor-
mal and defective regulatory mechanisms, it is only one step towards the
goal of obtaining detailed mechanistic models of processes pertaining to
disease. Since cellular processes are regulated at all stages leading from
DNA to functional proteins, the integration of information on regulatory
sequence as well as posttranscriptional regulation is crucial in this
endeavor.
1.6. Data Integration
For many organisms, different types of high-throughput data are rap-
idly accumulating, including protein-protein interaction data, 46 tran-
scription factor binding information, 47 genomic and promoter
sequence and gene ontologies, 48 and protein localization studies. 49
Yet, these data are often noisy and incomplete and cannot be used to
infer coregulation directly.
Coregulation of genes is generally expected to reflect involvement
in related cellular functions or pathways, and to result from shared
promoter-binding sites for a common set of transcription factors or
other forms of coordinated regulation. Therefore, several authors
employed the above-mentioned types of additional data to provide a
starting point and focus for coexpression analysis methods. 7,33,37,38,41,50
For example, target-regulator network analyses can be simplified if the
set of potential regulators does not include the entire genome but can
be restricted to a smaller number of candidates. 37,38
Combining gene
Search WWH ::




Custom Search