A new topological property as a measure of robustness of regulatory networks

Abstract

The growth of molecular biology has advanced such that we can measure the expression of thousands of genes. However, just measuring expression of multiple individual genes is insufficient to describe a systems issue such as complex diseases. To relate gene expression to physiological states (disease) and variables in an organism's environment we must utilize gene expression networks. These networks enable more intelligent identification of molecular subtypes of diseases and molecular targets for treatment. However, the reconstruction of gene expression networks is not accomplished easily. Constructing reliable gene expression networks with current methods requires large data sets or discard much of the data to reduce false positives. Although False Discovery Rate (FDR) is the most popular multiple hypothesis correction procedure for network inferences it is a conservative procedure and makes an unfitting assumption of independence for gene networks. Consequently, it tends to have a high false negative rate (i.e. low power) and requires huge sample sizes in order to be certain about a reconstructed network. In this project we will investigate a new method, that does not make any independence assumption, based on the co-regulation property of a gene expression network, we have discovered, (within one class of samples two "up" or two "down" regulated genes are positively correlated and two "up"/"down" genes are negative correlated). While the FDR method has been useful and its drawbacks tolerable; a new proposed method that better fits gene networks and could yield more usable results is highly warranted. (AU)