This dissertation presents novel methods and applications for face image editing using Generative Adversarial Networks (GANs). GANs have significantly advanced the field of face image editing, and this work delves into the challenges and possibilities of this technology, with a focus on improving segmentationguided image editing, GAN embedding for real face editing, domain adaptation, and hairstyle transfer. The research introduces new techniques, including the Semantic Region-Adaptive Normalization (SEAN) block for GANs conditioned on segmentation masks, an improved GAN embedding algorithm, a single-shot domain adaptation method, and the Barbershop approach for hair transfer. The dissertation also proposes pose-invariant hairstyle transfer techniques for images with varying poses. These techniques are evaluated on multiple datasets, and results demonstrate significant improvements over existing state-of-the-art techniques. By providing a comprehensive understanding of GAN-based face image editing, this work contributes to the ongoing evolution of face image editing and GAN applications, paving the way for future research and development.

Aerial base stations, which can be realized using unmanned aerial vehicles (UAVs) equipped with base station functionalities, have emerged as a promising solution to provide flexible and rapid network access in scenarios where conventional cellular networks face challenges with high traffic loads or insufficient coverage, especially in rural areas. Nevertheless, the increasing congestion of the radio frequency (RF) spectrum adversely affects the reliability and quality of service (QoS) of RF systems due to the elevated interference levels, which motivates the exploration of alternative electromagnetic spectrum bands. Recently, visible light communication (VLC), which uses light emitting diode (LED)--based luminaires to provide simultaneous illumination and communication, has been considered in a growing number of use cases, benefiting from the increasing deployment of LED-based lighting in a wide range of applications. This thesis investigates the optimal deployment of UAVs equipped with VLC technology to provide concurrent wireless service to multiple mobile users. We adopt a swarm intelligence approach and employ the particle swarm optimization (PSO) algorithm to design the trajectory of the UAVs; this enhances the data rate and fairness among the users while reducing the UAV power consumption due to their mobility. Following this, we also explore the potential of using VLC-enabled UAVs in precision agriculture systems, which can serve as dual-purpose devices that provide illumination and communication using LED-based grow lights. This can enhance the plants’ growth and support the wireless connectivity among various user nodes in the system, such as sensors, robots, and internet of things (IoT) devices. The advantages include achieving efficient resource utilization as grow lights simultaneously provide data communication and support plant growth and vegetation needs, especially in areas with limited sunlight. In addition, we highlight the impact of optimizing the UAV locations using simulations.

The increased deployment for renewable energy sources to mitigate the climate crisis has accelerated the need to develop efficient energy storage devices. Batteries are at the top of the list of the most in-demand devices in the current decade. Nowadays, research is in full swing to develop a battery that meets the needs of today’s renewable energy systems, which are intermittent by nature. Within the framework of improving the performance of batteries, there are parameters in the composition of the battery that play an important role in its performance: electrode materials, electrolytes, separators, and other factors. The key to battery development is the manufacture of electrode materials with optimal properties. Two-dimensional (2D) materials have led to advances in this field, firstly, using graphite as the anode in lithium-ion batteries (LIBs). However, when using the standard graphite as the anode for sodium-ion batteries (NIBs), the large ionic size and energetic instability of Na+ limit intercalation, resulting in a low storage capacity. Therefore, other 2D materials with large interlayer spacing need to be identified for use as electrodes. In this dissertation, our approach is focus on optimizing anode electrode materials by in situ conversion of 2D materials to obtain hybrid materials. These hybrids materials will synergistically improve the performance of LIBs and NIBs by combining the advantages of individual 2D materials. Starting with converted Ti0.87O2 nanosheets to the TiO2/TiS2 hybrid nanosheets. Then, taking advantage of the properties of MXene, we developed hybrid electrodes based on MXenes by converted V2CTx MXene into V2S3@C@V2S3 heterostructures. Finally, we boosted the redox kinetics and cycling stability of Mo2CTx MXene by using a laser scribing process to construct a multiple-scale Mo2CTx/Mo2C-carbon (LS-Mo2CTx) hybrid material.

Early detection of illness is essential in preventing symptoms from escalating and infectious diseases from spreading. Electrochemical biosensors are a promis- ing tool in healthcare detection. Previously, the collaboration between the Arold and Inal labs has led to the design of organic electrochemical transistors (OECT) capable of rapidly detecting coronavirus in saliva by using nanobody constructs as biorecognition units. In this project, I aimed to prove the versatility of nanobody- functionalized OECT biosensors in detecting other relevant viruses, specifically, Zika and Dengue. Both viruses pose a risk to multiple populations around the world, including the Kingdom of Saudi Arabia. I designed and produced nanobod- ies that are reported to bind to the NS1 glycoprotein, which is released by Zika and Dengue into the blood of the patient. Then, I confirmed the binding of the nanobodies to their associated targets. I also developed a robotic liquid handling script to automate the biosensing operations. Ultimately, this project aims to support the design of a multiplex OECT biosensor for blood-borne pathogens.

Isoprene is a 5-carbon volatile chemical monomer used in the production of various industrial products. Its current sourcing is primarily through petrochemical-based processes, but there is growing interest in developing more sustainable and bio-based production methods. This has led to the exploration of genetic engineering techniques to enable isoprene biosynthesis in photosynthetic microorganisms that convert carbon dioxide into their biomass and concomitantly, isoprene. This work focused on the use of the polyextremophilic red microalga, Cyanidioschyzon merolae strain 10D, as a candidate for heterologous isoprene synthesis. Through synthetic biology-enabled custom designed plasmid transformation and foreign transgene expression, a heterologous isoprene synthase (IspS) from sweet potato was successfully expressed from the nuclear genome of C. merolae and the translated protein was targeted to the algal plastid. In this work, transformant C. merolae strains were grown in a co-cultivation with related Galdieria sulphuraria, a mixotrophic Cyanidiophyceae, here fed with glucose to deliver respiratory CO2 to C. merolae when grown together in closed gas-chromatography vials. Through this strategy, the heterologous generation of isoprene could be quantified from transformants. It was determined that IspS-YFP fusions produced a higher isoprene yield compared to IspS-chloramphenicol selection marker fusions. The best isoprene-producing transformant was used to monitor the isoprene yield over 9-days of cultivation, reaching 231 mg/L culture on d9. This work represents a first-of-its-kind genetic engineering in red microalgae. The successful production of isoprene from this microalga could pave the way for the development of new and sustainable industrial applications for this commodity chemical.

Salt-affected soil is a prominent challenge in agriculture. Nowadays, more than 800 million hectares of land (about 6% of the world’s total land area) are induced with high salt concentrations, and thus, are unsuitable for growing typically salt-sensitive crop plants. The ongoing salinization of arable land exacerbates this limitation. To address this issue, the development of salinity-tolerant crop plants has gained considerable interest, with a protein identified by Prof. Mark Tester's group, named "SALTY2," offering promising potential. SALTY2 is overexpressed in response to NaCl treatment on Salicornia plants conferring salinity tolerance, and following the function of the SALTY2 protein from Salicornia and analogous proteins in Arabidopsis, yeast and in vitro, a universal mechanism in evolution is suggested. During my thesis, we analyzed the biophysical properties of SALTY2, and based on spectroscopic methods we confirmed it is an intrinsic disordered protein (IDP), which is consistent with previous studies claiming that IDPs play a vital role in stress response pathways. We have identified and characterized the loss-of-function "RG/RGG" to "KG/KGG" type mutation and a deltaSTM1 N-terminal mutation, and investigated the interaction of SALTY2 and other cellular components, including short fragment RNA, and 80S ribosome. Together with state-of-the-art high-resolution NMR and Cryo-EM methods we validated the direct interaction of SALTY2 with plant ribosomes, and 25nts random RNA, and determined the 3D structure of ribosome with the potential binding site of the SALTY2 protein. Combining biophysical, structural and functional analyses of the wild-type and loss-of-function mutants of SALTY2, we proposed a potential mechanism by which the IDP protein SALTY2 confers salinity tolerance in plants. These findings offer a deeper understanding of the molecular basis of salinity tolerance in plants via IDPs and contribute to the ongoing efforts to develop salinity-tolerant crop plants.

Ocean warming is leading to increased occurrence of coral mass bleaching events, threatening the persistence of these ecosystems and the communities that rely on them. While reef recovery is possible, conservation approaches based purely on transplantation/coral-gardening will not suffice to maintain these ecosystems over the projected environmental changes. Assisted evolution approaches aim to boost acclimatization and adaptation processes. A potential approach could be to harness the naturally occurring mechanism of environmental memory that has been observed in corals and other organisms, where an organism remembers a priming stress event to allow a faster/stronger response when the stress re-occurs. In this thesis I aimed to investigate whether this mechanism exists and how it is regulated on a molecular level in the sea anemone Aiptasia. Aiptasia were primed to heat stress by exposing them to 32 °C water for several years, or for one week. After a recovery period of one week at 25 °C, a naïve and the primed treatments were exposed to lethal thermal stress at 34 °C for three days. Primed treatments performed better than the naïve treatment in survival, photosynthetic efficiency and symbiont density for two days, after which the priming advantage was lost. The difference between the primed treatments indicated that the priming dose may affect priming success. There were clear indications of an epigenetic transcriptional memory mechanism on a transcriptional level. I observed a pronounced difference between control and heat-stressed treatments, indicating that transcription returned to near baseline expression after cessation of the priming exposure. The functional categories of differentially expressed genes in heat stress relative to control were similar between naïve and primed treatments, with the main difference observed in a stronger up- and downregulation of stress response genes in the long-term primed treatment. I optimized a chromatin immunoprecipitation protocol for use with Aiptasia by adjusting fixation, sonication and immunoprecipitation conditions. The enrichment of H3K4me2/me3 and poised RNA Pol II in the promoters of stress response genes will be investigated next to elucidate the mechanism of the observed epigenetic transcriptional memory in Aiptasia, and to ultimately inform conservation strategies for coral reefs globally.

Targeted protein degradation refers to a technique that selectively eliminates specific proteins within cells. This can be achieved by recruiting E3 ligases to specific proteins, which label the target with ubiquitin and lead to their eventual degradation by the proteasome. Cdc48/p97, a highly conserved AAA+ ATPase, is an essential player in the ubiquitin-proteasome system. It works as an unfoldase and segregase, unfolding ubiquitinated proteins and driving their removal from protein complexes, chromatin, or membranes to be sent for proteasomal degradation. Cdc48/p97 cooperates with many UBX domain-containing adaptor proteins, which use their UBX domain to bind to Cdc48/p97 and enable it to perform various functions. To expand on previous targeted protein degradation techniques, we designed nanobody-UBX degraders (NUDs) that recruit Cdc48/p97 to unfold target proteins without requiring ubiquitination. By establishing in vitro assays, we have uncovered insights into the mechanisms of substrate unfolding by Cdc48 in conjunction with NUDs and its endogenous adaptors Ufd1-Npl4 and PUX proteins. Our findings suggest the potential for a novel targeted protein degradation strategy that leverages the unfolding mechanism of Cdc48/p97 and its interactions with adaptors.

For over 50 years, bone marrow transplants have used CD34 to select stem cells. Recent research suggests that the most primitive hematopoietic stem cells (HSCs), long-term HSCs (LT-HSCs), are found in the CD34-negative portion of murine and human bone marrow cells. LT-HSCs are rare and cannot be isolated directly, making them difficult to study. During a bone marrow transplant, these stem cells must find their way to the bone marrow niche and engraft to become blood cells. Several cell adhesion molecules on the stem cell engage with their ligands on the endothelial cells lining the bone marrow vasculature to control this migration. Human LT-HSCs cells do not migrate and engraft well when infused in vivo, which may be due to a lack of adhesion molecules. Thus, the goal of this study was to determine whether this population of HSCs lacked adhesion systems (proteins and carbohydrate modifications) and, if so, to improve their migration and engraft ability by modifying key mechanistic steps in the adhesion cascade. Therefore, we investigated how distinct hematopoietic stem cell populations migrate to the bone marrow using adhesion mechanisms. This study represents the first direct analysis of adhesion molecules expression in LT-HSC and will potentially shed light on methods to optimally use these very valuable cells in the clinical bone marrow and cord blood transplants worldwide.

During the last decade, the average mobile wireless data usage per person has tremendously increased. An even faster growth of the traffic demand is expected for the incoming years, due to several factors such as the increasing global population, the spread of the Internet of things (IoT), and the development of advanced technologies that require a higher amount of data. While mobile communication technologies have rapidly evolved to meet this need in the most usual situations, it is expected that the sixth generation (6G) of mobile connectivity will be the first one paying considerable attention to under-connected environments such as low-income, remote, or disaster-struck regions. Many specialized researchers and entrepreneurs are trying to design and implement alternative network architectures specifically meant for enhancing the performances of the current telecommunication (telecom) infrastructure. In particular, the use of aerial base stations (ABSs) has received considerable attention due to the main advantages of easy deployability and low-cost that are typical of unmanned aerial vehicles (UAVs), which are available in several fashions depending on the application; moreover, UAVs are also eligible to carry reflective intelligent surfaces (RISs), which represent a promising technology that allows to reflect signals towards specific directions. Another possibility that we have investigated consists in integrating the transceivers inside or atop existing rural wind turbine (WT) towers, in order to increase the coverage radius while avoiding the cost of building a separate telecom infrastructure. A powerful mathematical tool for evaluating the performance metrics of either terrestrial, aerial, or vertical heterogeneous wireless networks is stochastic geometry (SG), since it can be used to model the locations of the base stations (BSs) according to tractable spatial distributions (with either a fixed or a random cardinality) in order to imitate the typical deployments of the nodes made in realistic scenarios; in particular, in this work we focus on rural and post-disaster situations. SG makes use of point processes to model networks' topologies. The developed spatial models, in turn, allow us to analyze the quality of service (QoS) experienced by the typical user served by the proposed networks. To this extent, we creatively and efficiently studied our inhomogeneous systems by making use of what we call the indicator method, meaning that we do not subdivide the ground plane in multiple homogeneous sub-regions, but we use indicator functions to provide general expressions that are valid over the entire ground plane. To prove the effectiveness of the novel architectures, insightful comparisons with the conventional ones are presented.