[Seminar]

18 Jan 2022, Tue, 2pm-3pm, KI bldg. 5^th fl. Room B501

Zoom: https://kaist.zoom.us/j/89586032430

회의 ID: 895 860 324 30

Data-driven interrogation of biological dynamics:
from subcellular interactions to neuronal networks in vivo

YoungJu Jo

PhD Candidate in Applied Physics, Deisseroth Laboratory, Stanford University

Biological systems are nonlinear dynamical systems consisting of heterogeneous entities. Understanding the logic of the complex spatiotemporal dynamics in such systems, robustly implementing specific biological functions, may require new approaches beyond the traditional hypothesis-driven experimental designs. Here we present a data-driven approach, enabled by high-throughput experimental and computational technologies, across multiple scales. We first discuss a computational imaging technique for simultaneously visualizing multiple aspects of subcellular dynamics [1, 2], its potential combination with molecular optogenetics to study the cell signaling networks, and the remaining challenges in these systems. Then we turn to neuronal networks in behaving animals where high-dimensional neural population activity could be reliably measured and perturbed over extended time. Synergizing with recent technical advances, we propose and experimentally demonstrate a unified deep learning framework to identify the underlying neural dynamical systems, reverse-engineer the neural computation implemented by the dynamics, and design spatiotemporally patterned optogenetic stimulation for naturalistic manipulation of animal behavior [3]. Application of this framework to the mouse habenular circuitry reveals cell-type-specific reward history coding implemented by line attractor dynamics [4].

References:

1. Jo*, Park* et al. Science Advances 3(8), e1700606, 2017.

2. Jo*, Cho*, Park* et al. Nature Cell Biology 23, 1329–1337, 2021.

3. Jo et al. in preparation.

4. Sylwestrak*, Vesuna*, Jo* et al. in revision.

문의: 박용근 교수 (내선:2514)