Abstract

I present work in my group using Bayesian network models for exploring interactions within complex biological systems (e.g., genetic regulatory networks, neuronal networks, ecosystems) and biology-adjacent systems (e.g., sociobiological/health systems). These systems consist of dense webs of interactions among many elements. For example, genes can regulate the expression level of other genes as well as react to the environment; neurons activate or inhibit other neurons plus can be modulated by sensory input; organisms predate upon and compete with each other while responding to abiotic factors; patient behaviour is influenced by a host of socioeconomic factors and past experiences. However, often these interactions are unknown and must be learnt from observational data. Here, I present our work in developing algorithms for this task, known as the network inference problem; particularly, I cover advances we’ve made to Bayesian networks and their machine learning approaches for characteristics of specific systems, and tools we’ve developed for helping to apply these models. I provide examples of how we use inferred networks to enhance knowledge discovery across these many types of complex systems.