Morphogenesis of follicular epithelium in Drosophila melanogaster: function of von Hippel-Lindau tumour suppressor gene

During my PhD I have been involved in several projects regarding the morphogenesis of the follicular epithelium, such as the analysis of the pathways that correlate follicular epithelium patterning and eggshell genes expression. Moreover, I used the follicular epithelium as a model system to analyze the function of the Drosophila homolog of the human von Hippel-Lindau (d-VHL) during oogenesis, in order to gain insight into the role of h-VHL for the pathogenesis of VHL disease. h-VHL is implicated in a variety of processes and there is now a greater appreciation of HIF-independent h-VHL functions that are relevant to tumour development, including maintenance and organization of the primary cilium, maintenance of the differentiated phenotype in renal cells and regulation of epithelial-mesenchymal transition. However, the function of h-VHL gene during development has not been fully understood. It was previously shown that d-VHL down-regulates the motility of tubular epithelial cells (tracheal cells) during embryogenesis. Epithelial morphogenesis is important for organogenesis and pivotal for carcinogenesis, but mechanisms that control it are poorly understood. The Drosophila follicular epithelium is a genetically tractable model to understand these mechanisms in vivo. Therefore, to examine whether d-VHL has a role in epithelial morphogenesis and maintenance, I performed genetic and molecular analyses by using in vivo and in vitro approaches. From my analysis, I determined that d-VHL binds to and stabilizes microtubules. Loss of d-VHL depolymerizes the microtubule network during oogenesis, leading to a possible deregulation in the subcellular trafficking transport of polarity markers from Golgi apparatus to the different domains in which follicle cells are divided. The analysis carried out has allowed to establish a significant role of d-VHL in the maintenance of the follicular epithelium integrity.

Unsupervised reinforcement learning via state entropy maximization

Reinforcement Learning (RL) provides a powerful framework to address sequential decision-making problems in which the transition dynamics is unknown or too complex to be represented. The RL approach is based on speculating what is the best decision to make given sample estimates obtained from previous interactions, a recipe that led to several breakthroughs in various domains, ranging from game playing to robotics. Despite their success, current RL methods hardly generalize from one task to another, and achieving the kind of generalization obtained through unsupervised pre-training in non-sequential problems seems unthinkable. Unsupervised RL has recently emerged as a way to improve generalization of RL methods. Just as its non-sequential counterpart, the unsupervised RL framework comprises two phases: An unsupervised pre-training phase, in which the agent interacts with the environment without external feedback, and a supervised fine-tuning phase, in which the agent aims to efficiently solve a task in the same environment by exploiting the knowledge acquired during pre-training. In this thesis, we study unsupervised RL via state entropy maximization, in which the agent makes use of the unsupervised interactions to pre-train a policy that maximizes the entropy of its induced state distribution. First, we provide a theoretical characterization of the learning problem by considering a convex RL formulation that subsumes state entropy maximization. Our analysis shows that maximizing the state entropy in finite trials is inherently harder than RL. Then, we study the state entropy maximization problem from an optimization perspective. Especially, we show that the primal formulation of the corresponding optimization problem can be (approximately) addressed through tractable linear programs. Finally, we provide the first practical methodologies for state entropy maximization in complex domains, both when the pre-training takes place in a single environment as well as multiple environments.

Correlations and Quantum Dynamics of 1D Fermionic Models: New Results for the Kitaev Chain with Long-Range Pairing

In the first part of the thesis, we propose an exactly-solvable one-dimensional model for fermions with long-range p-wave pairing decaying with distance as a power law. We studied the phase diagram by analyzing the critical lines, the decay of correlation functions and the scaling of the von Neumann entropy with the system size. We found two gapped regimes, where correlation functions decay (i) exponentially at short range and algebraically at long range, (ii) purely algebraically. In the latter the entanglement entropy is found to diverge logarithmically. Most interestingly, along the critical lines, long-range pairing breaks also the conformal symmetry. This can be detected via the dynamics of entanglement following a quench. In the second part of the thesis we studied the evolution in time of the entanglement entropy for the Ising model in a transverse field varying linearly in time with different velocities. We found different regimes: an adiabatic one (small velocities) when the system evolves according the instantaneous ground state; a sudden quench (large velocities) when the system is essentially frozen to its initial state; and an intermediate one, where the entropy starts growing linearly but then displays oscillations (also as a function of the velocity). Finally, we discussed the Kibble-Zurek mechanism for the transition between the paramagnetic and the ordered phase.