back to people
|
 |
Biological
systems are constantly under environmental and genetic changes, yet their
phenotypes tend to remain invariant. It has been argued that robustness to
changes arises as a buffering of the developmental process. Gene regulatory
networks often exhibit robustness and are found at the core of the
developmental process. Our main interest is to decipher the molecular mechanisms
granting robustness to biological systems. Therefore, we
adopt an in silico approach to
modeling gene regulatory networks that integrates two levels: local sequence
information and network architecture. Using a set of 10 TF-DNA complexes from the PDB as templates we modeled TF binding to all possible nucleotide sequences of
8bp in length by generating all 4^8 models of TF-DNA complexes. A distance
dependent statistical pair-potential is used to determine the free energy of
each modeled structure. For each TF we obtained free energy estimates for all
possible interacting DNA binding sites. This allows an explicit sequence-level
representation of upstream regulatory regions that determines the architecture
of a gene regulatory network model. In the model, mutations at the DNA level
cause changes on two levels: (a) at the local sequence level in individual
motifs, e.g. by changing the binding affinity, and (b) at the network
architecture level by creating and destroying motifs or binding sites, which
results in dynamically rewiring network connections. The regulatory network
model is used to determine gene expression dynamics while at a higher,
population dynamics, level the networks undergo cycles of reproduction,
mutation and selection. We use this multilevel hierarchical model to
investigate the molecular mechanisms underlying the evolution of robustness in
gene regulatory networks.