Related Experiment Videos
Functional and evolutionary inference in gene networks: does topology matter?
Mark L Siegal1, Daniel E L Promislow, Aviv Bergman
1Department of Biology, New York University, New York, NY 10003, USA. mark.siegal@nyu.edu
Genetica
|August 10, 2006
Summary
Biological network topology alone is insufficient for predicting function. Evolutionary simulations show that detailed gene location and expression data are crucial for accurate functional and evolutionary predictions.
Area of Science:
- Systems Biology
- Network Science
- Evolutionary Biology
Background:
- Biological networks, such as gene regulatory and protein-protein interaction networks, are often characterized as 'scale-free' with power-law distributions of connections.
- It is hypothesized that highly connected nodes (hubs) in these networks have a greater impact on network function and evolutionary outcomes.
- However, empirical evidence shows weak correlations between node connectivity and functional impact (e.g., lethality upon knockout) or gene expression responses.
Purpose of the Study:
- To investigate the utility of simple network topology measures for predicting biological properties.
- To explore the role of gene location and dynamic expression data in functional and evolutionary predictions.
- To challenge the 'top-down' inference approach based solely on network structure.
Main Methods:
- Evolutionary simulations of gene-regulatory networks.
- Analysis of relationships between gene connectivity, equilibrium expression levels, and fitness effects.
- Examination of correlations between connectivity and genetic variation (polymorphism and divergence).
Main Results:
- The fitness effect of gene knockout is dependent on both connectivity and equilibrium expression level.
- Correlation between connectivity and genetic variation is weak, but some nodes consistently show low or high polymorphism across independent evolutionary runs.
- Low polymorphism coupled with high divergence in certain genes can arise from neutral coevolution, not necessarily positive selection.
Conclusions:
- Simple measures of biological network topology have limited predictive power for functional and evolutionary properties.
- Accurate predictions require integrating detailed network position information with dynamic gene expression data.
- The 'top-down' inference approach from network topology alone is insufficient for understanding complex biological systems.