Reverse Engineering Gene Networks
From Dibernardo
One of the main challenges in the era of post-genomic research is to develop methods to extract information from the vast amount of data generated by high-throughput techniques. Tools such as microarrays probing expression of all known genes are now a standard technique worldwide, whereas new approaches based on sequencing are fast emerging.
This genome-wide data yields information not only at the single gene level, but also at the ‘systems’ level, i.e. how genes, proteins and metabolites interact with each other to perform a specific function. In order to ‘read’ such information new methods coming from quantitative sciences such as physics and engineering, have to be used. The identification of gene regulatory networks is of major importance in order to understand the working mechanisms of the cell in patho-physiological conditions.
Our reserch aims at developing and applying experimental protocols and computational algorithms to infer gene regulatory networks. Once the biological process is formalized as a network, the outcome is a mathematical model of the biological process. This model can be explored to generate novel hypotheses on the functioning of the biological process under study, to be then tested experimentally. Mathematical models of gene-networks may also aid in optimizing the administration of novel drugs.
The algorithms we have developed use measurements of mRNA concentration of the genes that are part of the network. These measurements, usually performed using gene expression microarrays, are repeated in a number of experiments that varies with the size of the network. In each experiment one or more of the genes of the network are perturbed (i.e. overexpression, drug treatment, knockdown, etc.). From these measurements it is then possible to reconstruct the causal relationships among the different genes, and therefore the regulatory network.
We have shown that it is possible to use reverse engineering methods to infer the regulatory gene network and drug mode of action from gene expression alone on a genome wide scale. We have applied our reverse-engineering approach to elucidate the transcriptional network regulated by the transcription factor p63, whose mutations are causative of human malformaton syndromes, in primary keratinocytes. We are now extending our reverse-engineering approach to infer a ‘consensus’ regulatory network in human and mouse species for all the known genes by using the available microarray data in public databases. The inferred model will be used both to understand gene regulations of specific disease genes, and also as a predictive model able to identify the dysregualted pathways in diseases starting from a single gene expression profile from a patient.
