HER2 expression, progression tempo, risk of recurrence, and patterns of dissemination during metastatic recurrence, much of which affects the need for and response to systemic therapies. Differences in breast Ginsenoside-F4 cancer biology and prognosis are demonstrably reflected in underlying differences in gene expression; indeed, variability in transcriptomic profiles were first observed and summarized into five well-defined intrinsic molecular tumor subtypes in Perou’s landmark study in 2000, a classification largely recapitulated in the recent much larger TCGA study incorporating protein expression, DNA methylation, copy number aberrations, and microRNA expression. Other studies have produced different but related molecular definitions of breast cancer heterogeneity, expanding the catalog of breast cancer to perhaps ten molecular subtypes. This study is an effort to further functionally characterize breast cancer heterogeneity through the concept of modules; we Echinatin hypothesize that such modular decomposition could yield clinically actionable components useful in achieving the goals of personalized oncology. Many definitions for biological modules have been proposed over the years; what unifies these definitions is that they attempt to simplify complex systems with large webs of interacting components into a smaller set of functionally integrated themes. The canonical ��hallmarks of cancer’, while primarily describing fundamental processes of carcinogenesis, can also be viewed as an informal attempt to impose or extract a modular structure on the complexity of cancer dynamics. According to this paradigm, the hallmarks of cancer include sustaining proliferative signaling, evading growth suppressors, resisting cell death, enabling replicative immortality, inducing angiogenesis, and activating invasion and metastasis.Tothesesix,arecent extension has added the reprogramming of energy metabolism and evading immune destruction, with emphasis placed on the interplay between malignant and hijacked ‘normal’ cells in the tumor microenvironment. Thegrowingnumberofbreastcancerrelatedgenome-widegeneexpression profiling datasets provides an opportunity to perform a comprehensive search for common patterns of gene co-expression using a formal, computable approach to distinguish different gene programs in breast cancer. Such co-expression modules can be viewed as an empirically derived catalog of coherent gene groups that might act together, and may have been selected for, as a unit to perform a function important to the cancer. Thus, the activity of modules within a tumor may be useful in understanding how that cancer developed, its likelihood of distant recurrence without systemic treatment,and potential vulner abilities that may be targeted by therapeutics. A prior study comparingthegenomicandtranscriptomicprofilesofnormalandmalignant breast identified 16 modules, one enriched for proliferation and two for immune response. Other studies have interrogated transcriptomic profiles for associations between co-expressed gene clusters and grade, aberrant chromosomal regions, and tumorigenesis, among others. Though not explicitly termed ��modules’, many other breast cancer gene expression studies implicitly rely on and address gene expression modularity by identifying cohesive gene expression clusters observable in unsupervised hierarchical clustering, followed by gene set enrichment to assign pathway-activation patterns that may be associated with a phenotype.