|
  |
|
|
Krasnow Institute > Monday Seminars > Abstracts COMPUTATION IN GENETIC NETWORKS Roland Somogyi Advances in biological signal transduction and gene regulation have given us insight into the fundamental molecular processes of living organisms. But how can such complex behavior be sensibly coordinated into networks which form the real basis of the output that we associate with organizational structures as cells and multicellular organisms? One approach to this question is hardware oriented, meaning that by the careful study of all of the molecular structures and all of their possible interactions according to lock-and-key principles we can take data on these reduced components and calculate up the organism in a high-powered computer. Alternatively, we may approach this problem from a software perspective. To this end the paradigm of genetic networks may provide most useful, particularly in the decompression of genetic sequence information underlying ontogeny. Simply stated, development may be understood in terms of patterns of coordinate gene expression governing proliferation and differentiation. A simple modeling language that accounts for the fundamental features of the global behavior of genetic networks can be found in Boolean networks. Genes are modeled as binary elements which can receive and project outputs to one another in a combinatorial fashion determined by a fixed set of Boolean (logical) rules. Each state, i.e. set of values for all elements, deterministically leads to one resultant state, forming a trajectory. Inadvertently, a state must be reached which has been occupied before. Such a repeating sequence of states is referred to as an attractor. Trajectories serve as an analogy to the patterns of developmental gene expression, while attractors resemble the final dynamic patterns of e.g. differentiated cell types. Moreover, many different trajectories may lead to the same attractor, demonstrating stability. Integrating these analogies from computation with experimentally obtained trajectories of gene expression may eventually enable us to extract genetic network logic diagrams. Such hypotheses may be tested and updated, and applied to targeted repair of "damaged" networks as found in cancer and degenerative disorders.
The Krasnow Institute for Advanced Study |
|
