Title: Bayesian analysis of complex biological and social systems
Speaker: Edo Aroldi

Abstract: Modern technology has transformed the concept of data in the biological, social and computational sciences. Data collections in general have grown large and heterogeneous, both in terms of the units of analysis and the measurements on such units. Measurements are typically collected over time, as underlying non-observable mechanisms unfold. This increase in size and depth of data about complex systems, however, has hardly translated into a richer understanding of fundamental mechanisms and principles that govern them. This problem is paramount in biology, where large-scale and high-throughput measurements of genes, proteins or enzymes promise key insights into the internal organization of organisms and the development of diseases like cancer.

In this talk, I will introduce mechanistic models and inference algorithms for the analysis of complex biological and social systems with the goals of testing substantive hypotheses, make predictions, and drive further experimentation. I shall then discuss alternative specifications and extensions that address issues fundamental to this kind of problems: data integration, dynamics, and scalability.

The proposed models allow us to properly ground the analysis in the context of accepted theories and empirical observations. Posterior inference enables us to reveal social structure, to explore dynamics theories of social failure in isolated communities, and to relate protein interactions to function. In particular, while analyzing aspects of cellular growth and environmental stress response, we identify a small set of genes that instantiate growth-specific programs of gene expression in yeast, thus enabling predictions about the expression of such genes under new experimental conditions, and estimates of the ``effective growth rate'' of any cellular culture. The effective growth rate of a cellular culture is a novel biological concept, and it is useful in interpreting the system-level connections among growth rate, metabolism, environmental stress response, and the cell division cycle. 

Bio:
Edo Airoldi is a postdoctoral fellow at Princeton University, affiliated with the Department of Computer Science and the Lewis-Sigler Institute for Integrative Genomics. His research interests include statistical methodology, machine learning, mechanistic models of complex systems and random graph dynamics, with application to the biological and social sciences.