Pancreatic ductal adenocarcinoma (PDAC) is the 3rd leading cause of cancer-related death in the United States. PDAC is highly metastatic with the liver being the most common site of metastasis. Notably, even after surgical resection of early stage of PDAC tumors, recurrence often occurs in the liver instead of the primary site that was surgically treated, suggesting the existence of micrometastasis in the liver of early-stage diseases. There is currently no approach to treat or prevent liver metastasis besides chemotherapy that is highly toxic and lacks efficacy. Therefore, new therapeutic targets need to be investigated to make a difference in patients’ outcomes. My thesis research project is to investigate small ubiquitin-like modification (SUMOylation) as a target to prevent and treat PDAC liver metastasis. SUMOylation is a reversible post-translational modification involving the attachment of a SUMO protein covalently to a target protein. Previous studies have shown that SUMOylation is required for KRas and c-Myc-dependent oncogenesis that are main drivers of PDAC progression. In addition, SUMOylation has been established as a target for activation of anti-tumor immunity. My findings suggest that SUMOylation inhibition can block PDAC liver metastasis. The mechanisms could be through both immune-mediated and tumor-specific effects.

Density functional theory (DFT) is an effective computational model, which enables calculations of properties and dynamical evolution under external fields for quantum many-body systems from first principles. On the other hand, there has been a burgeoning interest in addressing the ``inverse" problem: can we design a control field to steer a quantum system toward a desired configuration? Quantum Optimal Control (QOC) has risen to prominence as a potent framework in this regard. A lot of progress has been made, especially for the finite-dimensional quantum spin system. However, the optimal control of interacting many-body systems is a relatively young research field.

The first part of this dissertation presents a computational scheme that integrates the exact nonlocal exchange operator into ground-state calculations for multi-shell nanowires with various cross-sectional shapes, employing the finite element method. This method is applied to several core-shell nanowires, underscoring the crucial role of the nonlocal exchange operator. We demonstrate its significant influence on electronic properties, such as electron occupancy numbers, energy eigenvalues, energy separations, and electron localization patterns.

The latter half of this work delineates a computational methodology for applying QOC to interacting many-body systems within arbitrary geometric domains within the DFT context. Employing the Lagrangian multiplier method, we derive the gradient expression for the loss functional. A propagator integration method (Green's function) is implemented to evolve wavefunctions forward and backward, incorporating the WKB approximation to accommodate spatially varying effective electron mass. This optimization problem is iteratively solved to determine the optimal control field. Our approach is validated through a test example and subsequently applied to two complex systems, demonstrating its reliability and efficacy. These applications also allow us to investigate the effects of varying propagation times on control strategies and explore the feasibility of manipulating entire systems using localized control potentials.

Since the development of cereblon (CRBN) as proteolysis targeting chimeric (PROTAC), targeted protein degradation (TPD) has become the new paradigm in therapeutic design. Currently TPD has a limited number of E3 engagers, the small diversity leading to potential common off-target effects. Dr. Yuan Chen’s research group has developed ligands for the UBR E3 ligases with an affinity of Kd<1μM. This project focused on elucidating the structure of these ligands bound to the UBR-box domain of human UBR2 protein via x-ray crystallography. Three structures were resolved. UBR2 crystal was grown in 0.1M Bis-Tris pH 5.5 28% PEG 4000 for 7 days. UBR2-AF27 crystal was grown in 0.1M HEPES pH 7.0 26% PEG 4000 for 3 days. UBR2-AF23 crystal was grown in 0.1M HEPES pH 7.0 27% PEG 4000 for 3 days. Diffraction data sets were collected remotely at Stanford

Synchrotron Radiation Light source on a Dectris CCD PILATUS 6M detector on beamline BL12-2(λ = 0.97946Å), with beam size of 50.0 x 15.0 μm and energy at 12658.000eV with transmission of 10.602%. The resulting ligand bound structures deviate from docking simulations. While the Arg recognizing binding pocket behaved as expected, the resulting structures show insight on how bulky hydrophobic groups were bound to the UBR-box of UBR2. Gly144, Gly145, Gly146, and Gly147 (the 4G pocket) display flexibility, stabilize the ligand via hydrogen bonding and house the bulky hydrophobic residue as opposed to the previously reported hydrophobic pocket.

According to M Firth, PMY Fung, OM Rui (2006), the stock market is pretty sensitive to the top management turnover. Therefore, it is important to monitor the top management turnover in a very short time. This thesis attempts to use statistical and machine learning techniques to identify the Chief Executive Officer in a specific company. The natural language processing techniques we used may be able to show who is the Chief Executive Officer from millions of words in financial reports within few minutes. The whole process comes with four sectors: data collection, manipulation, named entity recognition and relationship analysis. We demonstrate this method with Tesla’s proxy statements (code: DEF 14A) from the U.S. Securities and Exchange Commission. The output shows that we can use both the named entity recognition algorithm with word2vec algorithm to detect the relationship between a job title and a human name.

Advances in high-speed DAC implemented in deeply scaled-CMOS processes open up the possibility of direct IF and RF waveform synthesis in a next-generation digitized transmitter architecture. A state of the art CMOS current-steering DAC is capable of providing wide bandwidth and high dynamic range for direct RF waveform generation. This enables the mostly-digital implementation of a direct waveform synthesis transmitter for TV band cognitive radio. RF waveforms with various modulation schemes can be synthesized in the digital domain and directly converted by a high-speed DAC. The transmission characteristics can be dynamically adapted to time-varying spectral environment. This transmitter architecture brings the benefits of reconfigurability and frequency agility. However, conventional waveform synthesis requires a >2GS/s DAC to fulfill the Nyquist requirement for covering the whole TV band (54 ~ 862MHz). Nevertheless, an alternative method is to convert the signals at a sampling rate below the Nyquist requirement and extract the image spectrum in higher Nyquist zones by bandpass filters.

This work demonstrates a proof-of-concept design of sub-Nyquist rate conversion and wideband direct synthesis using image spectrum. By utilizing the proposed multi-mode reconstruction, the DAC can shape the spectral envelope for enhancing the image spectrum located in the target channels. The desired transmission waveforms can be extracted from the second or third Nyquist zone by a bandpass filter. A circuit prototype demonstrating the proposed concept has been designed, fabricated, and measured in a general-purpose 65nm CMOS process.

This work presents the implementation of a 600MS/s 10-bit multi-mode sub-Nyquist rate DAC that enables wideband direct waveform synthesis for TV band cognitive radio transmitters. Measurement results show SFDR >55dB across the first three Nyquist zones and a low power consumption of 30mW. The IM3 is < -60dBc in the first and second Nyquist zones and < -55dBc in the third Nyquist zone.

Many studies to date are centered on how SUMOylation promotes cancer cells proliferation, but emerging information suggests that SUMOylation also regulates anti-tumor immune response and the tumor microenvironment. Here, with a clinical-stage small molecular SUMOylation inhibitor, TAK-981, I investigated the effects of SUMOylation inhibition in immune-competent mouse models of head and neck cancers. TAK-981 treatment of the mouse significantly prolonged survival and induced completed regression in some cases. The “cured” mice are resistant to rechallenge with the same tumor cells, indicating the formation of immunological memory. Single-cell transcriptomic analysis showed that treatment with the SUMOylation inhibitor decreased terminal exhausted T cells and increased T helper cells in the tumor microenvironment. Besides immune cells, I observed that the SUMOylation inhibitor synergized with IFN- in upregulating major histocompatibility complex (MHC) class I, which is essential for presenting neoantigen on cancer cells. Based on these findings, I performed combination therapy of the SUMOylation inhibitor with an immune checkpoint inhibitor (anti-PD1) that showed improved efficacy than the monotherapy groups