In this project the PIs will explore phenomena that span the tree of life: from metabolism in bacteria, through the determination of cell fate in embryonic development, to coding and computation of sensory information in the brain. They have identified broad theoretical problems which cut across the traditional biological divisions of organism and system: Do living organisms operate near the limits set by the laws of physics as they gather and process information? Can the detailed microscopic model of an organism, its wiring diagram be understood from the finite set of observations that can be made on how it behaves? How do organisms set the parameters that govern their function (i.e. how do they learn from experience)? These questions will be given a mathematical form, which will guide a search for answers in terms of general principles, in the tradition of physics, that will apply across disparate biological domains. The participants in the project will assemble into subgroups to attack instances of these problems. The individual projects will have unusual scope: as an example, the question whether the complex statistics of biological behavior can be captured in a learnable mathematical model will be asked in very similar terms both of spiking retinal neurons, and of the antibody sequence repertoire of individual zebrafish. These questions will be answered in the light of accurate data and the work will involve a close partnership with many experimental groups in fields ranging from bacteriology to human perceptual psychology. The product of these interactions will be the design of novel experiments and the creation of novel data analysis methods in order to address clearly formulated mathematical questions of broad significance. An important component of this project is the training of a new generation of physicists for whom the development of a theoretical understanding of biological systems is a central part of their discipline. The graduate students and postdoctoral scholars who pass through the group will learn by example how to pursue that goal in a way consistent with the intellectual rigor and traditions of physics.
For generations, our exploration of the biological world has been largely qualitative. As more and more of life becomes accessible to quantitative experiment, we see that life’s mechanisms are described by a vast array of numbers, at all scales: the rates of thousands of chemical reactions, the number of copies of each protein in a cell, the strengths of connections between neurons in the brain, and more. In the core of physics, we are used to the idea that the numbers characterizing the world fit together into a larger mathematical framework: the laws of physics. Could this also be true of the phenomena of life? This project has taken small steps toward this long-term goal by finding specific instances where important biological phenomena can be understood, not just qualitatively, but in quantitative detail, through the use of physical and mathematical principles. In this outcomes report we briefly describe three such attempts. Bacteria come in a bewildering variety of shapes. A rigid scaffold over which the cell membrane is draped, like the sheathing on a skyscraper’s frame, determines the shape of a bacterium. The scaffold consists of special proteins that are produced inside the cell and stick together to make polymer "girders" that spontaneously incorporate themselves into the scaffold at random places. Remarkably, this process allows the bacterial scaffold direct where new growth occurs while maintaining the cell’s shape. On the skyscraper analogy, it is as if we could add new floors anywhere in the middle of a finished forty-story building while not disturbing the offices on the original floors. We showed that this bacterial engineering feat is accounted for by energy. Combining the energy of binding of the scaffold proteins to the cell membrane, the energy it takes to bend these proteins, and the energy it takes to bend the membrane itself, we find that the existing shape of the cell membrane determines the energetically favorable orientation for adding a new scaffold protein. For a cylindrical shape, the most important "girder" protein prefers to be inserted along the circumference of the cylinder, insuring that as the cylinder grows in length, it remains straight (as observed). Scaffold proteins with different energy parameters might prefer to grow in a corkscrew pattern. As physicists would hope (and expect), simple energy considerations account for bacterial shapes. The immune cells (T-cells and B-cells) in our blood help fight infection by binding to invading viruses and bacteria, marking them for destruction by scavenger cells. Roughly speaking, each immune cell carries a surface receptor that recognizes only one unique pathogen; random "editing" of the genome each time a new immune cell is born is the source of this diversity. With DNA sequencing technology, we can obtain the unique DNA codes for the surface receptors of millions of immune cells in a diagnostic blood sample. We found a way to work back from these DNA sequences to infer the rules of the random genome editing process that generates them in the first place. This allows us to accurately estimate how many distinct surface receptor genes could in principle be produced: the answer is hundreds of trillions of distinct genes, vastly more than the tens of millions of distinct surface receptors circulating in your body at any instant (so that each of us has our own unique immune cell population). With this understanding of the statistical properties of immune cell genes as a foundation, we can attack fundamental questions about how immunity works, and perhaps understand why it malfunctions in autoimmune diseases. Finally, we have tried to understand how biological wholes can be more than the sum of their parts. Examples abound: a consensus flight direction emerges from interactions among birds in a flock, a coherent thought or percept emerges from interactions among neurons in the brain, and the fate of cells in the developing embryo emerges from interactions among genes. We believe that we have found a unified mathematical framework for describing all of these phenomena. In each case, we can start with just a handful of experimental observations, and build a model that predicts, quantitatively, the results of a wide range of experiments. These models are variations on the models used by physicists to describe the emergence of collective behavior in inanimate systems (e.g., the emergence of a rigid solid from the interactions among molecules). Through this mapping we can see that biological systems are poised near special places, or "critical points", in the "phase diagram" of possible systems. These critical points are special because they allow for optimal transmission of information among different parts of the system. Working with experimental colleagues, we have found direct signatures of this optimization in real data. We believe that future work will find many similar examples of successful mathematical description of biological systems, encompassing even broader realms of biological phenomena.
