Pratyush Tiwary, PhD
“From atoms to mechanisms: with a little help from statistical physics and artificial intelligence”
The ability to rapidly learn from high-dimensional data to make reliable predictions about the future of a given system is crucial in many contexts. This could be a fly avoiding predators, or the retina processing terabytes of data almost instantaneously to guide complex human actions. In this work we draw parallels between such tasks, and the efficient sampling of complex molecules with hundreds of thousands of atoms. Such sampling is critical for predictive computer simulations in condensed matter physics and biophysics, including but not limited to problems such as crystal nucleation and drug unbinding. For this we use and reformulate ideas from statistical physics and AI such as Maximum Caliber and the Predictive Information Bottleneck (PIB) framework for the sampling of biomolecular structure and dynamics, especially when plagued with rare events [1-2]. We demonstrate our methods on different test-pieces, where we calculate the dissociation pathway and timescales slower than milliseconds. These include ligand dissociation and sequence dependent conformational plasticity in proteins and RNA.
1. Tiwary and Berne, PNAS 2016
2. Wang, Ribeiro and Tiwary, Nature Commun. 2019
Ethan Garner, PhD
“From atoms to mechanisms: with a little help from statistical physics and artificial intelligence”
It is not known how cells grow in reproducible shapes with defined dimensions. We study rod shape formation in bacteria, as only a handful of proteins are required to build these shapes. Bacterial shape is determined by the cell wall, a single cross-linked macromolecule, and to elongate in these defined shapes, new material must be added into this structure in a spatially controlled manner. Two distinct enzymatic systems mediate rod-shaped growth: The Rod complex moves around the cell circumference, while class A penicillin-binding proteins (aPBPs) do not.
In order for biological systems to construct long-range shape, small proteins must be able to sense and respond to the much larger cellular geometry. In bacteria, this is accomplished by MreB, an actin homolog that polymerizes into inwardly curved filaments that bind to the membrane. These filaments are pulled around the cell by the activity of the associated cell wall synthesis enzymes. Using single-molecule tracking of these and other proteins, we work to understand how cells grow in rod shapes. We find that a key feature of MreB is that it acts as short-axis sensor, pointing along the greatest inward membrane curvature, constraining filaments and associated enzymes to move and insert material around the rod width. The wod width is not determined by properties within MreB filaments, rather it depends on the balance between the two systems: the Rod system reduces diameter, while non-MreB associated enzymes increase it. As Rod complex levels increase and cells get thinner, we see an increased density of moving MreB filaments and more directionally moving enzymes. This increased circumferential synthesis increases the amount of oriented material within the cell wall, which increases its structural anisotropy: in response to internal turgor pressure, cells stretch less across their width (reinforcing rod-shape) instead, stretching more along their length, causing rod-shaped growth.
Thus, the orienting of MreB filaments to the greatest curvature, coupled with cell-wall reinforcing synthesis, creates a geometric and mechanical feedback loop that reinforces existing rod shape. This local feedback also allows the creation of rod shape from spheres, as oriented MreB motion arises in small bulges on the sphere surface attract oriented MreB motion, causing that bulge to rapidly elongate into an emerging rod
Pratyush Tiwary, PhD
“The Role of Chance in the Survival of the Fittest”
The spreading of evolutionary novelties across populations is the central element of biological adaptation. Models of evolutionary spread have ignored, until quite recently, the randomness inherent in the reproduction process. But having excellent genes is not sufficient to be successful in evolution – one also needs luck to avoid accidents or to be at the right place at the right time. Using microbial evolution experiments and simulations, I elucidate the role of chance in evolutionary processes and show that deterministic models indeed fail to predict the dynamics of evolution as it is observed in microbial evolution experiments. I present novel approaches, combining modeling and experiments, that explain the observed patterns of genetic diversity, spatial spread and adaptation at a cellular scale, where self-driven jamming can impede the expansion, and at a global scale, where the dynamics is sped up by long-range dispersal.
