ED04-03

With continually falling cost of genome-scale assays, the reuse of using specimens from previously sampled individuals for new studies is becoming significantly faster and more affordable. The quest of utilization of gene expression profiling to increase the accuracy of diagnosis and prognosis, or to improving our capability to choose among therapies has generated a body of thought-provoking observations reported in prominent medical journals such as New England Journal of Medicine. The challenge remains that, in many cases, that the gene expression sets (signatures) of related studies are mostly non-overlapping and vastly uncharacterized, raising questions about their biological significance and pathophysiological consequences with clinical implications. Investigation of these expression profiles is complex as regulatory networks consisting of thousands of genes and gene products interact across a wide range of biological scales spanning the transcription factor and microRNA regulation of gene expression, to protein interactions via direct binding, to protein interactions achieved via phosphorylation events in a signaling pathway, and all the way to the manifestation of pathological processes. The magnitude of questions arising from these profiled genes calls for a new level of sophistication in the undertaking of biological and clinical validations. Indeed, traditional biological characterizations of gene functions and clinical trials are rate-limiting and costly. In addition, interpretation of uncharacterized gene expression arrays signatures for targeted therapy, particularly equivocal signatures, is poised to become a salient topic in clinical education and practice. Thus, an efficient and progressive environment for the comprehensive understanding of biomedical mechanisms underpinning enigmatic molecular signatures would be a powerful resource for the translational research community and for clinicians.
 Challenges for translational scientists and clinicians in understanding Clinical Signatures in Expression Arrays (CSEAs) and other genome-scale assays:
 •Understanding the clinical implications of CSEAs is challenging, as the number of published articles are increasing geometrically and there is no group focusing on their interpretation and synthesis for the clinical domain.
 •Determination of the pathophysiological consequences of specific GSEAs is perplexing as their genes are in majority uncharacterized for the clinical condition studied.
 •Current approaches to reduce the complexity in CSEAs paradoxically consist of reducing their targeting. Indeed, CSEAs consist of a long lists of genes in a dataset, for which a stated proportion of genes need to be dysregulated in an individual patient for this signature to be considered positive in a leading to conclude, for example, that an individual is at higher risk for lung metastases. Shorter signatures are derived from these large lists of dysregulated genes by selecting a specific gene among co-regulated ones, usually based on a statistical rationale.
 •Lists of relevant genes are generally organized in a table according to an arbitrarily (baised) grouping around biological functions and processes.
 We demonstrate current and novel systems biology approaches to understanding the molecular underpinnings of molecular signatures. Current approaches to analyze CSEAs consist essentially in inserting the CSEAs gene set in one of twenty or so software that will calculate the “statistically significant enrichment” of the proportion of CSEAs genes across different types of classifications such as molecular functions, biological processes, cellular localization, known pathways, and chromosomal location. This approach has remained unchanged for the last seven years and suffers from a number of limitations that will also be addressed. Novel approaches based on rigorous analyzes of protein interactions will also be addressed.

First AACR International Conference on the Science of Cancer Health Disparities-- Nov 27-30, 2007; Atlanta, GA