Genetic Network Identification using Convex Programming

Proceedings 8th International Conference on Systems Biology, paper F15, Oct. 2007.

• gene_net_id.pdf

Genes in living cells regulate various cellular biochemical processes through genetic regulatory networks. In such a network, the genes produce proteins that act as transcription factors for other genes or themselves. The use of RNA microarrays has made it possible to have an expression profile for a large number of genes when exposed to different conditions. One of the most important problems in systems biology is to use these data to identify the interaction pattern between genes in a regulatory network, especially in a large scale network. In the literature, this is sometimes called reverse engineering the genetic network. Genetic network identification has important potential applications, for example in drugs discovery where a systems wide understanding of the regulatory network is crucial for identifying the targeted pathways.

In this paper we propose a method for identifying genetic regulatory networks using genetic perturbation data. In a genetic perturbation experiment, small perturbations are applied to a genetic network in an equilibrium state and the resulting changes in expression activity are measured. We aim at identifying the smallest model, corresponding to the sparsest network, that explains the data, while conforming to known a priori structural information about the network, if any. A priori biological knowledge is typically qualitative, encoding whether one gene affects another gene or not, or whether the effect is positive or negative. We solve the combinatorially hard problem of finding the sparsest model using a technique from convex optimization, and demonstrate that our method performs better than other existing methods.