The reverse engineering of cellular genetic circuits from gene expression data is an important and scientifically challenging problem. When applied to system more complex than a yeast cell, however, most of the approaches for reverse engineering are limited by their reliance on a significant number of ad-hoc perturbation, such as gene knockout. We propose to address these issues and to attempt to reconstruct the genetic circuitry of human B-cells by coupling three distinct components: 1) an information theoretic method for the inference of cellular networks from microarray profile data for a large number of distinct cellular and molecular phenotypes 2) A synthetic network simulation framework for the assessment of the performance of the reverse engineering under a variety of constraints and conditions, such as noise and network complexity. 3) An adaptive learning method that will iteratively apply optimal perturbations to the biological system, which will allow refining the cellular network model over time. Furthermore, we propose to biologically validate this approach by elucidating the cellular networks of human B lymphocytes, from a variety of normal, tumor-derived, and experimentally manipulated B cell phenotypes, which are significant both from an oncological and immunological perspective. This will be accomplished by analyzing an existing set of over 340 high-quality microarray expression profiles. Finally, we will create a software platform to reverse engineer any biological system for which adequate microarray data is available. The platform will also allow to design and perform virtual, in-silico gene perturbation experiments.
Showing the most recent 10 out of 18 publications
