Student Posters Repository

Yeasts, a type of fungi, play a critical role in both ecological systems and industrial processes, including biofuel production. Despite their importance, many genes responsible for yeast metabolism remain unidentified, and the functions of numerous yeast genes are unknown. Our research aims to link these unknown genes to the metabolic processes they control by leveraging a combination of computational techniques and experimental methods to analyze nearly 1100 yeast genomes.

Fungi are pivotal in the natural breakdown of plant biomass and the conversion of resulting sugars into biofuels and valuable bioproducts, a process predominantly carried out by yeasts from the Saccharomycotina subphylum. While Saccharomycotina yeasts collectively possess a remarkable ability to metabolize a wide array of sugars, many of these metabolic capabilities are poorly understood. However, phenotypic data on sugar metabolism and other traits under 75 conditions are available for a diverse array of yeasts, many of which now have corresponding genomic data.

In this study, we apply machine learning methods to analyze gene orthologs across approximately 1100 yeast genomes with the goal of linking genotypes to metabolic phenotypes. Our approach accurately predicts numerous sugar metabolic traits across the Saccharomycotina subphylum and ranks genes potentially related to each trait. Among the high-ranking genes, we identify many experimentally validated Saccharomycotina genes associated with sugar metabolic pathways. Additionally, we uncover genes of unknown function that have not previously been associated with sugar metabolism, making them ideal candidates for further experimental investigation.

Wastewater treatment plants (WWTPs) are an important reservoir of antibiotic-resistant bacteria and likely contribute to the propagation of antibiotic resistance genes (ARGs). While ARGs have been quantified in such matrices, the bacterial host range of ARGs is determined in silico. Pore-C, a novel 3C long-read sequencing technique, relies on spatial cross-linking of genomic material within the same cell. Applying this technique for the first time in the WWTPs matrix expands the knowledge of the host range of ARGs as it is determined with in vitro association techniques and bioinformatics tools.

Indoor air quality (IAQ) in educational buildings significantly affects student health, comfort, and learning outcomes. Many existing K-12 schools face challenges in maintaining optimal IAQ due to outdated infrastructure and inadequate ventilation systems. This research aims to develop a comprehensive retrofit guide to enhance IAQ in these buildings using advanced Computational Fluid Dynamics (CFD) simulations and High-Performance Computing (HPC). 

Data were collected from several representative classroom environments, capturing key parameters affecting IAQ like temperature and air velocity. CFD simulations were conducted on HPC  clusters and Azure Cloud using an octree-based parallelized in-house dendrite code. These simulations employed PDE-based multiphysics models and Large Eddy Simulations (LES) with Variational Multiscale Methods to accurately model airflow patterns. 

The simulations revealed significant insights into the current IAQ, showing areas in the classrooms with sub-standard comfort. Future research will extend these simulations to additional classroom typologies and develop detailed retrofit guidelines for different climate zones. Collaborations with K-12 schools will facilitate the implementation and testing of these guidelines, ensuring their practicality and effectiveness.

Galaxy clusters, the largest gravitationally bound structures in the universe, offer valuable insights into the evolution of large-scale structures and the fundamental properties of the Universe, such as dark matter and dark energy. Accurate mass calibration of these clusters is crucial for utilizing them as cosmological probes, but significant observational challenges often complicate this. In this work, we fit the mass of galaxy clusters while incorporating systematic effects into cosmological simulations, addressing a common shortfall in such simulations. We consider data from the mini Uchuu and Cardinal simulations, binning the data by richness and redshift. By stacking the lensing profiles of these clusters, we derive the mean mass profiles, concentration, and additional parameters accounting for systematic effects, such as boost factor parameters. This approach mitigates the high statistical uncertainty inherent in individual cluster measurements and allows for a direct comparison between the simulation results and observational data. Our analysis considers various systematic effects, including shear and photometric redshift uncertainties, cluster miscentering, dilution of the source sample, and projection effects. We demonstrate that by systematically incorporating these effects into simulations, we can accurately retrieve the true parameters from observational data, enhancing the fidelity and applicability of simulations in cosmological research. Furthermore, our exploration of small and large radial scales reveals the differential impact of systematic effects, with the small-scale fits exhibiting better constraining power and agreement with the data. This is because the projection effect is a predominantly large-scale effect. The insights gained from this study contribute to our enhanced understanding of the systematic effects and how to factor them into existing analytic models and simulations, paving the way for more accurate analyses of the latest data releases from astronomical surveys.

Much like the turbulence we experience in an airplane, ocean turbulence is characterized by vigorous, random motion of water. This is important for the modulation of currents and material properties, influencing spatial scales that vary from a few meters to hundreds of kilometers. These variations are important for understanding and anticipating climatic changes and ecosystem resilience. In the ocean, irregular bottom topography is particularly efficient at generating near-bottom turbulence, although what controls the strength and nature of the turbulence is still not thoroughly understood. My research uses a method called Large Eddy Simulations to understand the interplay between turbulence, topography, and earth's rotation. 

Kilometer-scale modeling of global atmosphere dynamics enables fine-grained weather forecasting and decreases the risk of disastrous weather and climate activity. Therefore, building a kilometer-scale global forecast model is a persistent pursuit in the meteorology domain. Active international eorts have been made in past decades to improve the spatial resolution of numerical weather models. Nonetheless, developing the higher resolution numerical model remains a long-standing challenge due to the substantial consumption of computational resources. Recent advances in data-driven global weather forecasting models utilize reanalysis data for model training and have demonstrated comparable or even higher forecasting skills than numerical models. However, they are all limited by the resolution of reanalysis data and incapable of generating higher resolution forecasts.

Low-Energy Electron Diffraction (LEED) is a commonly employed technique for surface structure characterization. It can provide quick, qualitative insights into surface periodicity and ordering. However, simple LEED analysis overlooks a large amount of quantitative information that is contained in the diffracted electron beams. This additional information is extracted by LEED I(V) via analysis of beam intensities as a function of the electron acceleration voltage. 

LEED I(V) can be a powerful tool for surface structure investigation, as the curves are sensitive surface atom positions at the picometer scale. Comparison of measured and theoretical intensities gives a strong test for surface structure models. Yet, LEED I(V) remains an underutilized technique because current software solutions are lacking in usability to a degree that makes them inaccessible for routine applications for all but few specialized groups.

To overcome this issue, we present the Vienna Package for TensErLEED (ViPErLEED) which aims to provide a unique all-in-one package for LEED I(V). On the experimental side, ViPErLEED includes electronics to easily upgrade existing LEED setups without extensive modifications and a spot-tracking tool for I(V) curve extraction. On the theory side, we introduce a Python package for calculation and optimization of theoretical LEED-I(V) spectra. The ViPErLEED software is developed based on, and as an extension to, the established TensErLEED package. It further uses standard file formats, enables automated symmetry detection, and can set up calculations with just a handful of parameters.