NB/CNS Seminar- William Bialek | Tuesday, January 16, 2024 at 12 pm

NB/CNS Seminar

Date: Tuesday, January 16, 2024

Reception: 11:30 am

Time: 12 pm

Location: Broad 100

Speaker: William Bialek, Professor

Princeton University

Faculty Host: Varun Wadia (Yuki Oka)

Title: Toward a theory of genetic networks

Abstract:

If we try to make a moderately realistic model of genetic regulatory networks we are led into a dense forest of parameters. A theory of these networks would cut through this complexity, deriving rather than fitting essential networks behaviors and the properties of underlying mechanisms. I'll described our efforts in this direction, using the optimization of information flow as a principle. All the work is motivated by the "gap gene" network involved in the early steps of patterning the fruit fly embryo. We'll see evidence that information transmission is optimized by matching the distribution of maternal inputs to the signal and noise properties of the network, and by having readout mechanisms that make the most reliable possible inferences from the gene expression levels. Finally we tackle the problem of optimizing the network itself, maximizing the transmission of positional information with a limited number of molecules. The class of models over which we optimize has 50+ parameters, and we find optima whose behaviors are very close to those of the real network. Indeed, one cannot do better by fitting. This approach provides a novel framework within which to answer classical questions about chance and necessity.