Thursday, March 1, 2007
Presented by the Systems Biology Discussion Group
Organizers: Andrea Califano, Columbia University; Anirvan Sengupta, Rutgers University
Genetics of Natural Variation in Human Gene Expression
Vivian G. Cheung, MD
University of Pennsylvania
The focus of our study is the genetic basis of variation in expression levels of genes in human cells.
The expression levels of genes differ among cell types and are responsible for the distinctive characteristics of cells. Recent studies have shown that gene expression levels differ not only among cell types within an individual but also among individuals. Previously, we showed that there is a genetic component to individual variation in human gene expression. The variability is greater among unrelated individuals than among related individuals. Our study now focuses on identifying the genetic determinants of this variation. Results have allowed us to identify a network of cis- and trans-acting transcriptional regulators. In this presentation, I will describe our genetic approach to finding transcriptional regulators and the resulting findings.
Gene Networks to Map Multigenic Inheritance
T. Conrad Gilliam, PhD
University of Chicago
Genetic susceptibility to complex (common) disorders arises in large part from the fateful combination of heritable mutations distributed among multiple genes. Identification of the culprit gene combinations presents a problem of 'high dimensionality', i.e., the number of potential solutions (gene combinations) vastly exceeds the number of experimental observations. We limit the gene combination 'search space' to a dramatically reduced subset of molecular interactions retrievable from the electronic literature by automated text mining, and describe a prototype molecular interaction network-based linkage model developed by Andrey Rzhetsky and coworkers (Columbia University), and its application to the analysis of heritable neuropsychiatric disease.
Coauthor: Andrey Rzhetsky, Columbia University
Reverse Engineering Biological Networks to Identify and Validate the Key Drivers of Physiological Phenotypes
Eric E. Schadt, PhD
Merck & Co., Inc.
There has been intense focus the last ten years on developing and enhancing statistical methods for the analysis of DNA variation and molecular profiling data, where the methods developed are largely specific to one type of data. However, a number of recent studies have well demonstrated the power of integrating genotypic, molecular profiling, and clinical data to elucidate common human diseases and drug response, resulting in the identification of a number of genes and gene networks associated with these complex traits. Underlying the novel integrative genomics approaches employed in these studies are a number of fairly simple statistical procedures that simultaneously consider orthogonal data from a number of sources. I describe several approaches that take a more holistic view of biological systems (compared to the classic reductionist view), characterize the extent of discovery resulting from their application, and discuss ways the underlying integrative genomics approaches could be advanced to achieve even greater power in such studies. In addition, given that many complex phenotypes like common human diseases are likely emergent properties of biological networks that are themselves defined by a complex network of genetic and environmental perturbations, I discuss the need to move beyond the concept of single gene perturbation experiments as a way to validate these networks and the complex phenotypes they induce. I demonstrate how single and multifactorial gene perturbation experiments play a critical role in validating the predictive power of reverse engineered networks. More generally, I discuss the important role independent sets of experiments over varie