We are interested in how life emerges from collections of molecules, cells, and organisms; asking how bacterial cells grow, divide, and move; how collections of cells form patterns; and how and why discrete animal behaviors are generated by the brain. To answer these questions we combine physics-style experiments, advanced data analysis techniques, and theoretical modeling, and use tools from optics, computer vision, machine learning, and more.
Our group focuses primarily on two scales, the cellular and the organismal. On the cellular scale we ask how bacteria deal with a physical world, how they grow into the correct shape, how they sense their environments, and how they move in large groups to generate beautiful dynamical patterns. On the organismal scale, we study the behaviors of animals, asking in the worm, fly, fish, and mouse how behavioral patterns are performed and how they are controlled by the brain. Examples of current projects:
The proteins responsible for producing cell shape are typically several orders of magnitude small than the cell itself. We use a variety of different shaped cells (rod-shaped Escherichia coli; crescent-shaped Caulobacter crescentus; and corkscrew-shaped Spiroplasmas, Vibrio cholerae, and Helicobacter pylori) to study how nanometer-sized cell wall synthases produce a micron-sized cell wall. For this project, we have employed novel combinations of 3D imaging, imaging analysis, optical tweezers, and atomic force microscopy to discover that the biophysical interactions between specific cell-shape-determining polymeric proteins and the local geometry of the cell envelope contribute in crucial ways to the resultant cell shape. For our most recent work on these systems, see Morgenstein 2015, Nguyen “Methods in Molecular Biology” 2016, Ouzounov 2016, and Bartlett 2017.
Myxococcus xanthus is a fascinating social bacterium that forms large, structured groups containing up to a million cells in order to effectively hunt prey and form protective structures. To form these multi-cellular structures, these cells coordinate their motion along surfaces. We draw conclusions about the origins of this activity by studying the mechano-chemical properties of cell propulsion at the molecular scale (Wartel 2013, Balagam 2014, Czerwinski 2014), the production of large coherent forces by groups of cells (Sabass 2017), and the motion and biochemical state of thousands of individual cells at the population level by simultaneously tracking thousands of closely-packed cells (Thutupalli 2015).
Despite the recent explosion in our ability to characterize the basic molecular, cellular and genetic components of organisms, our understanding of their behaviors has advanced at a much slower pace. Underlying this discrepancy has been our inability to define what behaviors are and how they relate to an organism’s interactions with the world. We have developed technology to map the behavioral landscape of an animal and find the essential building blocks of behavior using state-of-the-art imaging and machine learning techniques. Using this platform we are currently investigating the temporal pattern of behaviors (Berman 2014, 2016), how sensory inputs are translated into behavioral outputs by the brain and neural-muscular systems (Nguyen PNAS 2016, 2017), and how interactions between individuals are used during social behaviors in multiple model neural and genetic model systems (Wang 2016, Klibaite 2017). We have recently received NIH funding for two exciting new projects on closed-loop behavioral interactions between courting individuals (with Murthy, Bialek, and Pillow), and neural mechanisms in the cerebellum involved in flexible and social behavior (with Wang and Pillow).
Josh also dabbles in the physics of brass musical instruments, with particular interest in the design of historical valveless trumpets and horns. Baroque and early brass instruments lacked valves and were longer than their modern counterparts, using the very high harmonics to approximate the notes of a major scale. These designs are notoriously out of tune and even the best players with the strongest embouchures have trouble making them play well to the ears of modern audiences. However, there is some indication that historic instruments might play more in tune compared to modern reproductions. I have been using computer algorithms to alter the design of these historic horns to show that they can be made to play in tune to exacting precision with minor alterations to the bore profile. Our next steps are to explore whether historic instrument makers took advantage of these ideas.