Integrated Calendar

Saare - Circa-SCOPE: high-throughput live single -cell imaging method for analysis of circadian clock resetting.
Moshe- "Oligomeric states and much more using Mass Photometry”

Understanding the role of depth is a fundamental endeavor in explaining the practical success of deep learning. In this talk, we will focus on the problem of approximating the maximum function over $d$ inputs using deep ReLU networks and with respect to the uniform distribution over a hypercube. We will show that while approximating the maximum using depth 2 networks to arbitrary accuracy requires arbitrary width, this can be done to arbitrary accuracy using a depth 3 network with a fixed width of $d^2$. Additionally, we will also show that this upper bound is tight, namely that width $Omega(d^2)$ is also necessary when approximating using depth 3. Moreover, using this efficient depth 3 construction, we will show that greater depths result in a lesser width requirement, where width $mathcal{O}(d)$ suffices when we allow depth $mathcal{O}(log(log(d)))$. Lastly, we will show a size lower bound of $d$ neurons for approximating the maximum using any depth. These results establish a partial depth hierarchy for approximating a naturally occuring function which helps explain the benefits of depth over width.

Protein secretion typically involves the conventional pathway, where secretory cargoes containing a signal peptide are transported into the endoplasmic reticulum (ER) through the SRP-SEC61 system and then released via ER-Golgi trafficking. However, our understanding of protein secretion has recently undergone a revolution with the discovery of multiple secretory proteins that lack a signal peptide. These include interleukin-1beta, TGF2, and Tau, which are secreted through unconventional protein secretion (UcPS) involving vesicle trafficking as a major pathway.
The mechanism by which UcPS cargoes enter into the vesicle has been unclear due to the absence of a signal peptide. In our previous work, we identified a membrane protein called TMED10, localized in the ER-Golgi intermediate compartment (ERGIC), as a potential translocator that regulates the entry of UcPS cargoes into the ERGIC, thus initiating vesicle trafficking (Cell, 2020). In this study, we discovered that not only TMED10 but also other TMED family proteins serve as cargo translocators in UcPS. These TMED proteins individually and cooperatively regulate the translocation of different sets of UcPS cargoes into secretory vesicles, leading to diversified regulation of UcPS under various conditions.
The ERGIC functions as a crucial station for translocation due to its unique lipid composition. Sphingomyelin stimulates translocation, while cholesterol antagonizes this effect. TMED10 forms a channel with two open states, and a high current state is associated with translocation activity. Together, our findings suggest that TMEDs represent a novel class of protein channels involved in the distinct translocation and release of numerous UcPS cargoes. This sheds new light on the intricate processes underlying protein secretion.