Gene Networks and Transcription Regulation

No gene is an island. All genes interact with other genes and gene products to produce the observable traits that determine fitness and allow natural selection to operate. Recent empirical advances have allowed the large scale exploration of genetic interactions through protocols such as microarrays, high throughput genetic interaction approaches, and the yeast two-hybrid assay. At the same time ecologists have been producing datasets with interactions between large numbers of species. These kinds of datasets can be characterized mathematically as networks because they involve interactions of varying strengths between a group of similar ``nodes''.

I have analyzed the application of the network approach to both ecological and evolutionary problems and explored how network analysis techniques taken from social sciences and physics can be applied to these biological problems. One of the major challenges in this area is to understand how genetic networks come to have the patterns that we observe. Will natural selection tend to cause gene networks to have power law degree distributions? Will robustness evolve as a general feature of these networks? I have developed analytical models to determine the opportunity for canalizing selection to create robust gene networks. The conclusion of these studies is that genetic robustness is unlikely to evolve through direct selection unless a large number of genes can be buffered by a single gene. Environmental robustness, on the other hand, can easily evolve even if it is controlled by a single gene.

In order to understand how patterns arise in genetic networks we must understand where new genes come from and are how they are inserted into the existing genetic network. In particular, the adaptive landscape plays a critical role in determining the likelihood of gene duplication and the time required for duplication or subfunctionalization. The longstanding explanation for gene duplication is that redundancy between duplicate gene pairs allows the divergence of gene function. This restrictive model is a slow, partially neutral process. I have approached this problem by asking how natural selection can lead to gene duplication and have shown that genetic divergence is selected for under largely similar conditions regardless of how many gene copies there are. My work has shown that adaptive forces can lead to the rapid formation of gene families and the evolution of novel genetic function. I have extended my framework to include both changes in coding and regulatory regions and found that a form of stochastic tunneling can allow genetic variation necessary for adaptive duplication to accumulate even when the intermediate steps are not positively selected.

Affiliated Researchers