Description:

Title: Higher-Order Interactions among Drugs, Genes, and Species

Speaker: Van M. Savage, UCLA; Santa Fe Institute

Abstract: This talk will discuss a general framework for expanding beyond the pairwise interactions that are often measured and displayed for networks. This general framework will describe how to measure emergent interactions among more than two components. A key part of the talk will be to explain how to develop measures at these higher levels that disentangle emergent interactions from interactions that arise purely due to combinations of pairwise parts. Another key part is to explain methods for rescaling that normalize for the strength of lower-order effects so that the magnitude of these interaction measures can be easily interpreted in terms of information about interactions. These general principles and methods can be applied to derive measures based on any starting definition of what is an interaction (i.e., covariance, mutual information, ANOVA, etc.) This framework will be applied to experimental data on antibiotic interactions to show that higher-order interactions are common and can be seen to be shifting to become more antagonistic. Moreover, I will briefly discuss how these types of interactions could affect the shapes and roughness of fitness landscapes such that some drug combinations could be used to slow the evolution of drug resistance by bacteria. Ideas will be discussed for how these findings might generally to genetic epistasis, predator-prey interactions, social systems and more.

Host: Adilson Motter

Keywords: Physics, Astronomy, Complex Systems