2021 University of Chicago Undergraduate Research Symposium: Session 1

Linkage disequilibrium (LD) refers to the non-random association between alleles on chromosomes. Numerous factors affect LD, including recombination rates, population demographic changes, and natural selection. Differences in LD data between populations can be leveraged to determine the particular alleles responsible for a given phenotype, allowing us to improve our understanding of disease mechanisms and potentially improving our ability to treat these diseases. At present, a universal tool for accessing LD data across global populations through an API remains absent. With this project, we aimed to fill that gap by developing a tool to access LD data and compare it across populations with ease. In the process of developing this framework, the EmeraLD package and Flask API were used extensively. We also provide an exploration of strategies for storing LD data. The estimated storage requirements for chromosome 22 without the use of such strategies were roughly 5GB, necessitating such efforts. We utilized banded matrices, and adaptively banded matrices, in an attempt to improve efficiency through reducing the storage required by the data. Through performance-testing these strategies, we found that it was not viable to satisfy this aim without sacrificing an unacceptable level of LD data that we determined to be relevant. Through the development of a public facing API, we can enable broader access to LD data without requiring substantial data storage costs.

Achyutha Menon

Prof. John Novembre, Human Genetics; Arjun Biddanda, Human Genetics

College Summer Research Fellow

Biological & Health Sciences, Computing Science

COVID-19 is an infectious disease that can cause mild to severe respiratory symptoms, such as trouble breathing and pneumonia, with severe cases leading to death. An efficient and automatic method for analyzing COVID-19 presentation in the lungs on thoracic CT scans is desirable for informing treatment decisions and patient management. Accordingly, the purpose of this study is to create a deep learning model that can simultaneously produce segmentations of the lungs and tissue indicative of COVID-19 disease on CT scans in order to extract quantitative information, such as the percentage of COVID-19-infected tissue in the lungs, and to use this information to predict clinically relevant information. Thoracic CT scan images from 41 patients each with multiple temporally separate CT acquisitions (total of 221 scans) were manually segmented, serving as “truth” for training, validation, and testing data using the MATLAB Image Labeler program. A multi-headed U-Net convolutional neural network model architecture was used for the automatic segmentation with two up-sampling paths, one for lung segmentation and one for COVID-19 lung involvement segmentation. The U-Net model is evaluated through the Dice coefficient, which assesses the overlap between the model segmentations of the lungs and COVID-19 involvement compared to manual segmentations. With the ability to produce lung and COVID involvement segmentations, a correlational analysis will be performed to determine the association between the COVID-19-infected tissue in the lungs and clinical information, such as the patient’s duration of hospitalization. Accurate prediction of a patient’s length of hospital stay from their thoracic CT scan upon admission to the hospital has broad reaching applications, including the hospital’s ability to better plan patient care and schedule hospital staff and rooms, as well as economic applications for the hospital, insurance, and the patient’s family.

Beatrice Katsnelson; Elise Katsnelson

Prof. Maryellen Giger, Radiology; Jordan Fuhrman, Committee on Medical Physics, Giger Lab

College Research Fellow (Global Health)

Biological & Health Sciences

Many types of cells need to adopt polarized morphologies to function. In metazoans, polarity often emerges from asymmetric distributions of proteins in the highly conserved partitioning-defective (PAR) network. The PAR network consists of two sets of proteins which locally compete for binding to the cell membrane and are typically enriched in complementary sub-cellular domains. The protein LGL-1 is a core member of the PAR network and plays a key role in maintaining polarity in multiple contexts, but how it does so remains unclear. In the C. elegans zygote, PAR proteins segregate along the anterior/posterior axis. LGL-1 localizes to the posterior, and is necessary for maintaining PAR asymmetries in sensitized genetic backgrounds. I will use this cell as a model to evaluate two hypothesized mechanisms for how LGL-1 maintains cell polarity: (1) LGL-1 sequesters the anteriorly localized heterodimer PAR-6/aPKC in the cytoplasm, and (2) LGL-1 binds to PAR-6/aPKC on the cell membrane and dissociates it into the cytoplasm. To resolve this question, I have represented these hypotheses as mathematical models, which I have constrained with empirical measurements. I will use these models to determine whether or not these hypotheses can plausibly explain a set of experimental observations where LGL-1 is required for stable polarity in the absence of other posterior PAR proteins. Ultimately, I hope to quantify the extent to which each hypothesis contributes to polarity maintenance in the C. elegans embryo.

Eric Chen

Prof. Edwin Munro, Molecular Genetics and Cell Biology

Liew Family Research Fellow

Biological & Health Sciences, Computing Science

Prostate cancer (PCa) is the second leading cause of cancer-related mortality in the United States. The standard of care for advanced prostate cancer patients is androgen deprivation therapy, but disease progression to metastatic castration-resistance prostate cancer (mCRPC) remains inevitable. In addition to genomic alterations in the androgen receptor (AR) pathway, PTEN loss-of function (LOF) occurs in 50-75% of mCRPC patients and correlates with disease progression and poor clinical outcomes. PTEN LOF leads to hyperactivation of the PI3K pathway to promote growth, proliferation, and angiogenesis. It has also been shown to induce an immunosuppressive tumor microenvironment (TME) with low T-cell infiltration. Prior studies have shown that the androgen/AR signaling axis can suppress the expression of a B7 superfamily member, B7-H3 (CD276), and that B7H3 expression correlates with high Gleason score and progression towards metastatic/hormone refractory PCa. Furthermore, activation of PI3K/AKT/mTOR signaling upregulates B7-H3 in non-small cell lung cancer. We hypothesized that PTEN loss-of-function coupled with AR suppression enhances the expression of B7-H3, leading to an adaptive resistance mechanism that results in an immunosuppressive, pro-tumorigenic tumor microenvironment. In vitro results have confirmed that loss of PTEN via CRISPR/CAS9 deletion or transient knockdown in murine and human PCa models increases the expression of B7-H3. Mechanistically, preliminary findings suggest that B7-H3 is regulated by PTEN in a PI3K/AKT/mTOR-independent manner. Experiments are in progress to evaluate the immunosuppressive function of B7-H3 in vivo using MYC-driven murine PCa models, in order to better utilize novel immuno-oncology strategies to treat PTEN-deficient PCa.

Farah Doughan

Prof. Akash Patnaik, Medicine; Dr. Jordan Shafran, Medicine, Patnaik Lab

Biological & Health Sciences

Air Traffic Control (ATC) facilities are high-stress environments that require rapid decision making in dynamic situations; errors can be life-threatening and are unacceptable. This project is a critical component of NextGen, an FAA project to modernize the nation’s airspace, focusing specifically on improving the safety and effectiveness of alarms, alerts, and warnings (collectively called signals) for air traffic controllers. My aim was to identify both the scope and nature of any signal problems by modeling data collected from NASA’s Aviation Safety Reporting System (ASRS) database. Through simple odds ratio calculations, I found that the automated alarms were much more prone to signal response error than the air traffic controllers themselves, and using logistic regression, I found that signal errors may be associated with human error. The data suggest, then, that signals may contribute to human error. Moreover, I intend to use techniques such as multinomial logistic regression to identify which alarms are most prone to signal error. Altogether, this information will help my research team to both identify the problem and suggest enhancements to alarms, alerts, and warnings, as part of a handbook that the FAA will use to design signals for new equipment.

Grace Richards

Prof. Keith Ruskin, Anesthesia and Critical Care

College Research Fellow

Statistics

Actin cytoskeleton networks are necessary for a diverse number of cellular processes, such as migration, polarization, trafficking, vesicle transport, and division. Assembly and organization of these networks are mediated by a variety of actin-binding proteins. TIRF microscopy is a valuable strategy to directly visualize how individual actin filaments in these networks are formed and maintained; however, current analysis of this microscopy data is time-consuming and requires manual tracking. To improve analysis efficiency, we worked on developing an algorithm that 1) tracks user-specified filaments with high fidelity, 2) allows manual adjustments to correct any potential mistakes, 3) handles events such as filament annealing, breaking, and crossing over, and 4) analyzes the data to output filament elongation rates. With further work, we plan to increase filament detection sensitivity and robustness against noise, optimize computing speed, and develop modules that will output different measurements and analyses. This algorithm will greatly reduce the amount of time it takes to analyze actin filaments in microscopy movies and will be useful in my own analysis as I move forward with biochemical characterization of the actin-binding protein Diaphanous, which is a member of the formin class of actin-binding proteins. These proteins typically work in tandem with another actin-binding protein, profilin, to increase the rate at which individual actin filaments elongate. However, preliminary data has suggested that Diaphanous is capable of mediating increased rates of actin filament elongation in the absence of profilin. Through in vitro pyrene assays and TIRF microscopy, I will characterize the biochemical properties of Diaphanous and determine whether it does enhance filament elongation rates in the absence of profilin. This property in the formin would suggest a novel mechanism by which the protein mediates actin filament elongation and deepens our understanding of how Diaphanous and actin interact to form cellular networks.

Hannah Ye

Prof. David Kovar, Molecular Genetics and Cell Biology

College Research Fellow

Biological & Health Sciences

Tobacco smoking poses health risks, and, consequently, governments have taken a role in efforts to reduce tobacco usage. Governments have focused on increasing taxes on tobacco products, requiring smoke-free laws in public establishments, and sponsoring mass media campaigns to reach this goal. The objective of this project was to review the effectiveness of these policies and provide recommendations for future policies. In this qualitative narrative review, articles examining the success of initiatives to reduce smoking were analyzed. These articles were found in public health and economics journals located using the search engine Google Scholar. Methodologies referenced in these articles included time-series analyses and population-based studies. Metrics, including the smoking rate and cigarette usage, were used to evaluate the success of these policies. The findings of the current study indicate that cigarette taxation is regressive and leads smokers with lower incomes to ingest nicotine more intensely. Advertisements evoking strong emotions are most successful in convincing viewers to quit. Finally, smoking bans within public establishments play a critical role in reducing the likelihood that individuals begin smoking and incentivizing current smokers to quit, particularly among persons with low income. Recommendations based on the current study’s findings include (a) taxes on tobacco products should be increased; however, the revenue should be directed to low-income populations adversely impacted by smoking, (b) governments should invest increased resources in anti-tobacco media campaigns, especially advertisements featuring strong emotions and graphic images, and (c) governments should expand the scope of public facilities where smoking bans apply. Tobacco control research is valuable as it helps inform policymakers at the local, state, and federal levels on the most effective strategies to minimize the consumption and societal effects of tobacco products. An extension of the current research will focus on the role of school curricula in curbing smoking among young adults.

Jacob Gillis

Prof. Marcia Tan, Public Health Sciences

Biological & Health Sciences, Public Policy

Genomic analysis of formalin-fixed paraffin-embedded (FFPE) tissue samples offers a window into disease evolution and the genetic basis of observed epidemiological patterns over time. Key limitations in the analysis of FFPE DNA include inadequate DNA extraction methods and formalin-induced DNA damage. The aim of this project was twofold: (1) to optimize DNA extraction from archival FFPE tissue samples spanning the last ~100 years and (2) to characterize FFPE-specific DNA damage patterns using published genomic datasets. For the first aim, systematically testing and modifying 4 commercial FFPE-DNA extraction protocols using freshly prepared FFPE blocks revealed that the Covaris truXTRAC FFPE DNA kit with a modified incubation step yields the largest amount of DNA (mean = 4881 ng). Subsequent DNA extractions from archival FFPE tissue samples dating back to 1927 revealed that the novel extraction method recovers enough high-quality DNA for next-generation sequencing from samples that are over 30 years old. For the second aim, comparatively analyzing data from 26 pairs of FFPE and fresh-frozen (FF) tissue showed that the FFPE-derived DNA is significantly more fragmented than FF-derived DNA (FFPE mean = 141 bp, FF mean = 148 bp; p = 2.10E-08). Additionally, an average of 30.5% of the sequencing reads in FFPE-derived DNA were enriched in adenine immediately 5’ upstream of the read start, but only 23.3% of the sequencing reads in FF-derived DNA (p = 1.22E-20), suggesting that fragmentation in FFPE DNA predominantly occurs at an adenine base. In contrast, FF DNA showed a significant increase in the frequency of cytosine at the same position (FF mean = 28.3, FFPE mean = 24.8%; p = 7.61E-16). In summary, my research proposes an optimized method for DNA extraction and establishes preliminary DNA damage patterns resulting from formalin fixation. These insights are useful to researchers generating sequencing data from historical FFPE tissue.

Jovan Stanisavic

Prof. Maanasa Raghavan, Human Genetics

College Research Fellow

Biological & Health Sciences

Infectious disease has impacted the world in a myriad of ways. In agricultural settings, infection alone accounts for $220 billion in damage and majorly affects 20 – 40 percent of global agricultural efforts. While the molecular basis of single infection in many diseases is well-established in many agricultural and model species, efforts to provide lasting resistance to disease have failed. One important reason for this is that plants interact with complex microbial communities, whose interactions can greatly influence both pathogen evolution (including resistance) and disease outcomes. In particular, coinfection of multiple pathogen strains is common in nature, and has been shown to drastically alter pathogen performance. Although the outcome of a given host-pathogen interaction often differs drastically between single and co-infections, the genetic basis of these differences remains poorly understood. The model plant Arabidopsis thaliana, and one of its dominant bacterial pathogens in nature, Pseudomonas viridiflava,provide a well-characterized, practical study system to characterize how coinfection shapes the genetic architecture of pathogen virulence. Using a diverse set of 224 viridiflava strainswith sequenced genomes, isolated from Arabidopsis, and transformed with luciferase, we have developed a high-throughput luminescence-based assay which quantifies in-planta bacterial growth, a common measure for pathogenicity. This system has provided insight into the differential outcomes associated with coinfection. From this data, we conduct GWAS using both phylogeny-based and kinship-based approaches to discover loci associated with coinfection effects. P.viridiflava strains have revealed genetic variants strongly associated with coinfecting strains but absent in singular infection, suggesting evolution of traits particular to coinfection settings. We infer aspects of the genetic underpinnings of coinfection in a dominant Arabidopsis pathogen, and provide further evidence of the importance of microbial interactions in pathogen evolution and plant disease.

Peter Laurin

Prof. Joy Bergelson, Ecology and Evolution

Biological & Health Sciences

Donor-specific transplantation tolerance can be achieved in an experimental model of heterotopic heart transplantation where, at the time of transplantation, recipients receive an injection of donor splenocytes (DST) + an antibody that blocks CD40/CD40L costimulation (anti-CD154). Using this model, we have previously shown that endogenous CD4+ conventional T cell (Tconv) populations specific for highly expressed constitutive graft alloantigen fail to expand their high avidity clones, in contrast to the same T cell populations in rejecting mice. This low avidity population-level state is fixed long term and resistant to alloantigen rechallenge and occurs even in mice treated with DST+anti-CD154 in the absence of a graft. We have also reported that infections with Listeria monocytogenes (Lm) after tolerance induction can precipitate T cell-dependent transplant rejection. Our goal is to investigate if Lm infection at the maintenance phase of tolerance can expand high avidity Tconv clones, perhaps explaining Lm-dependent loss of tolerance. We found that infection with Lm engineered to express the same constitutive alloantigen shared with the DST could expand high avidity alloreactive Tconv clones in mice previously treated with DST+anti-CD154 in the absence, but not in the presence of a DST-matched allograft. In transplanted recipients, Lm-mediated loss of tolerance did not alter the allospecific Treg:Tconv ratio, nor the expression of the anergy markers CD73 and FR4. This suggests that antigen persistence in the graft results in stronger cell-intrinsic and cell-extrinsic control of alloreactive Tconvs than transient alloantigen expression in the form of DST without a graft. Our results also raise the hypothesis that rejection after Lm infection may be driven by T cells specific for alloantigens upregulated during inflammation, rather than those constitutively expressed at high levels.

Peter Wang

Prof. Maria-Luisa Alegre, Medicine, Rheumatology

Dean's Fund for Undergraduate Research Awardee

Biological & Health Sciences

Existing nasal swab tests for SARS-CoV2 can diagnose patients infected with COVID-19, however, tests that can predict severe symptom development are still needed. The wide range of symptom intensities is a major challenge in the on-going COVID-19 pandemic, even looking into the far future. In response to viral infection, many features of cellular physiology change as cells mount a defense and viruses hijack the cells’ translation machinery for their own replication. One set of molecules involved in these changes are transfer RNAs (tRNAs). These abundant non-coding RNA molecules decode mRNA and deliver amino acids for protein translation. During viral infection, tRNAs can be cleaved and undergo chemical modifications and changes in expression. Recent work in the Pan lab has studied how cellular tRNAs respond to SARS-CoV2 infection using patient nasal swabs and identified several possible biomarkers that predict symptom severity at the time of diagnosis. However, these data were generated with Next-generation sequencing. Such measurements are slow and expensive, and are not suitable for rapid scalable testing. Here, I develop computational tools to identify potential qPCR-based targets from tRNA-sequencing data. In contrast to mRNA, qPCR targets for tRNA are not trivial, as one must consider strong secondary structure, abundant modifications, and incomplete sequencing coverage. These tools will facilitate translation of tRNA-sequencing results to scalable qPCR-based tests for predicting COVID-19 symptom severity, and potentially other diseases beyond COVID-19 as well, such as various cancers in which tRNAs are implicated.

Rahul Gupta

Prof. Tao Pan, Biochemistry and Molecular Biology; Dr. Chris Katanski, Biochemistry and Molecular Biology, Pan Lab

College Research Fellow

Biological & Health Sciences, Computing Science

Pancreatic adenocarcinoma (PDAC) is extremely deadly, with only a 9.3% five-year survival rate, highlighting the need for novel therapeutic approaches. We are working to target metabolic alterations that PDAC requires due to abnormal tumor nutrient availability. Previously we have found that PDAC uses argininosuccinate synthase 1 (ASS1) to de novo synthesize arginine in an arginine-deprived tumor microenvironment. Perturbing de novo arginine synthesis either by using CRISPRi technology to knockdown ASS1 expression or removing citrulline, a necessary precursor of arginine, significantly slows PDAC growth. Although slowed, PDAC cells are still able to grow. Thus we want to explore how we can further decrease PDAC growth by inhibiting other metabolic pathways in combination with inhibiting de novo arginine synthesis. Madiraju and colleagues recently proposed a novel link in the liver between ASS1-mediated arginine synthesis and the AMPK nutrient sensing pathway. They found in hepatocytes that locally generated AMP, produced by ASS1 when generating arginine, activates AMPK. AMPK is also known to help tumors adapt and survive under nutrient stress. I hypothesized that AMPK activity helps nutrient-deprived PDAC cells survive in an ASS1-linked manner. However, after using CRISPRi technology to knockdown AMPK expression, we found no change in PDAC growth even under ASS1-inhibited conditions. As a result we are now pivoting to investigate other adaptations that PDAC cells use to cope with ASS1 inhibition. From previously published work identifying mechanisms of cellular adaptation to amino acid starvation, we identified several candidate pathways that may mediate PDAC’s ability to grow without ASS1 activity. We are currently determining if targeting these pathways in combination with perturbed de novo arginine synthesis might be effective at further slowing PDAC growth. We hope these results can provide insight into developing more effective PDAC treatments by targeting a combination of metabolic adaptations that PDAC is reliant on.

Riona Chen

Prof. Alexander Muir, Cancer Biology

College Global Health Scholar

Biological & Health Sciences

Tumor grade and mutation status are putative biomarkers of response to trans-arterial embolization of pancreatic neuroendocrine liver metastases (PNLM). We sought to evaluate the effectiveness of using computed tomography (CT) radiomic features to classify grade and mutation status of PNLM. 40 patients with NLM who underwent CT scans and biopsies at a single center from April 2009 to February 2018 were included in the database. Biopsied tumors with recorded tumor grade (Grade 1-3) and mutation status (presence or absence of MEN1/DAXX/ATRX mutation) were segmented in both the hepatic arterial (HAP) and portal venous (PV) phases using 3D Slicer. Quantitative features (n=115) from each scan were extracted from segmented tumors using PyRadiomics. Standard feature normalization was performed. Principal component analysis (PCA) was used for dimensionality reduction. Linear and radial basis function support vector machine (SVM) were used for training. K-fold cross-validation was used to estimate testing accuracy. Permutation testing was used to assess statistical significance. There were 25/40 (62.5%) patients with MEN1, DAXX, and/or ATRX mutation. There were 9/40 (22.5%) G1 tumors, 16/40 (40%) G2 tumors, and 37.5% G3 tumors. Over 95% of the variance in the data was described by the top 6 principal components, which were used for training and testing. Testing accuracy was 75% +/- 11% for predicting mutation status (p=0.008). Testing accuracy was 52.5% +/- 11% for predicting tumor grade (p=0.004). CT radiomic features can be used to predict mutation status and tumor grade in patients with PNLM.

Ryan Hoang

Dr. Etay Ziv, Memorial Sloane Kettering Cancer Center Department of Interventional Radiology

Biological & Health Sciences

Household air pollution (HAP) is the cause of increased morbidity and mortality in India. The health risks from household air pollution include ischemic heart disease, stroke, chronic obstructive pulmonary disease (COPD), and lung cancer. Exposure to HAP results from the burning of biomass fuel such as coal, wood, dung, or agricultural residues for cooking in India. As rural inhabitants are more reliant on these cooking methods than urban residents, they are at increased risk from HAP. Dr. Jayant Pinto from the University of Chicago Department of Surgery and his colleagues from the Indian Institute of Science are evaluating the negative impact of HAP on olfactory, pulmonary, and cognitive functions in rural households of Srinivaspura in Karnataka, India. Since we are in the initial stages of the project, we are currently trying to quantify air pollution levels in rural households in Srinivaspura. Due to the COVID-19 pandemic, this project was suspended until recently. In March 2021, the team in India set up the first Edimax air pollution monitor in Srinivaspura. Another Edimax air pollution monitor was installed in a nearby location on a separate network on April 1st, 2021. The team in India is currently in the process of installing a third device made by Purple Air that stores data for intermittent transmissions. We anticipate seeing higher than healthy levels of air pollution within the households in Srinivaspura from the data. The overall effort of this project to quantify and characterize the impact of HAP on rural Indians’ health will help inform public health measures in India to mitigate adverse health related outcomes from HAP and add substantially to the current literature on this subject.

Simatul Rashid

Prof. Jayant Pinto, Surgery

College Global Health Scholar

Biological & Health Sciences

Physiologic changes during pregnancy alter ways in which the body responds to infectious and inflammatory insults. In this study, we aimed to describe the symptoms and clinical presentation of pregnant patients diagnosed with COVID-19, a novel virus characterized by respiratory, inflammatory, and vascular complications. Understanding how COVID-19 manifests in high risk black pregnant patients can help inform the medical management of this vulnerable, understudied population. A retrospective, descriptive analysis of COVID-19 positive pregnant patients admitted at an urban tertiary care center in the south side of Chicago was conducted to collect maternal medical information/outcomes. Overall, 56 subjects were included. Common obstetric diagnoses included labor/contractions (n=25, 44.6%), ruptured membranes (9, 16.1%) and bleeding (3, 5.4%). COVID testing was prompted by symptoms in 18 (32.1%) patients and universal screening in 38 (67.9%) patients. Of those patients included 15 (27.3%) had a cough, 8 (15.4%) had a fever, 5 (9.3%) had diarrhea, 10 (18.5%) had shortness of breath, 6 (11.1%) had chest pain, 7 (12.9%) had a headache, 7 (12.9%) had myalgia, 6 (10.9%) had malaise, and 9 (16.7%) lost their smell or ability to taste. There were no maternal deaths, but 2 (3.9%) patients were admitted to the ICU antepartum and 1 (2.0%) postpartum. Overall, 5 patients were intubated. The median (IQR) GA of delivery was 39.07 (37.29, 39.71) weeks, with 11 (22.0%) experiencing preterm (<37 weeks) labor. Overall, 19 (37.2%) patients underwent a cesarean delivery. A large majority of patients were asymptomatic and diagnosed with COVID-19 by universal screening. The rates of preterm labor and cesarean delivery were similar or higher than the national average. This data will enable the analysis of the risk factors and inform the treatment of this virus throughout this understudied population.

Tinyan Dada

Prof. Sarosh Rana, Obstetrics and Gynecology; Melissa Kuriloff, Pritzker School of Medicine

College Global Health Scholar

Biological & Health Sciences

This study presents the unique advantages and disadvantages of cuneiform tablet photography and digitization through the example of tablets found in a school in ancient Iraq dating to 1800 BCE. Among the pressing demands of modern scholarship remains the need for accurate digitization practices to facilitate the sharing, publication, and teaching of cuneiform source material. Our research examines specific methods of cuneiform digitization, that being traditional tablet photography, Reflectance Transformation Imaging (RTI), and High Dynamic Range (HDR) imaging, and how these methods can best be used to digitize the Oriental Institute’s corpus of millenia-old Sumerian scribal tablets. The “3NT” tablets are a diverse collection of exercises and literature used to teach cuneiform literacy to aspiring scribes some four-thousand years ago in Nippur. These tablets pose challenges to digitization. They are intrinsically inscribed 3D objects, some with a reflective surface, and written by yet untrained scribes which require a more comprehensive, mixed model of photography to preserve their fine details best. Furthermore, our work on these tablets has allowed for new pedagogical approaches in teaching the cuneiform script in the modern-day classroom, as students can view high-quality, precise images of these tablets as opposed to traditional and often subjective line drawings. Immediately, the implications of this intersection of digitization with Assyriology are apparent: greater access to the source material for independent researchers, institutions, and pedagogical pursuits, as well as more accurate scholarship. We expect to find that a hybrid model of photography mediums allows for the most efficient and precise virtual archiving of the 3N-T Nippur corpus. The virtual preservation of these scribal tablets forms a new chapter in their long genealogies of teaching cuneiform literacy; as scribes once used them to learn cuneiform, so too does this digital humanities scholarship continue that legacy of cuneiform education today.

Joshua Beirich; Clara Mikhail

Prof. Susanne Paulus, Near Eastern Languages and Civilizations, Oriental Institute

College Research Fellow (Hoeft)

Assyriology & Cuneiform Studies

Turn back time over five thousand years. It’s 3200 BCE—in Mesopotamia, a stylus inscribes the first written words in world history. This region, today’s Iraq, produced clay tablets inscribed with cuneiform writing for the next three thousand years, from astronomical and medical treatises to literature like the Epic of Gilgamesh. UChicago’s Oriental Institute (OI) Tablet Collection houses thousands of these artifacts. However, a roadblock plaguing cuneiform studies is publication. Many tablets lack high-quality photographs, limiting their utility for scholarship, whether from digs centuries past like the school exercises of Nippur or recently excavated from the OI’s renewed expedition to the site. My research strives to answer how to most effectively visually convey the copious epigraphic and artifactual data in the Tablet Collection to a global network of scholars and general public. To this end, I develop digital preservation methods for tablets via scanning and photography. This year, my research team added to our methodological repertoire HDR (High Dynamic Range) photography and RTI (Reflectance Transformation Imaging) / PTM (Polynomial Texture Mapping) which allows the viewer to manipulate light sources. I use multiple photo-editing applications to create a composite image simultaneously displaying all faces of a tablet and enhancing visibility of material characteristics. I upload these images to the open source Integrated Database, research each tablet’s historical background, and aid in transliteration and translation of Akkadian and Sumerian textual content. This allows comprehensive presentation of not only text in the abstract but its materiality—form, composition, surface treatment, impressions, color, script and handwriting, organization, state of preservation, and therefore data reliability. Each unique object requires its own methodology for effective presentation. These physical aspects of the text can only be conveyed through photography and must be conveyed if we are to understand the active relationship between written word and material medium.

Madeline Ouimet

Prof. Susanne Paulus, Near Eastern Languages and Civilizations, Oriental Institute

College Research Fellow (Hoeft), College Summer Research Fellow

History, Interdisciplinary Humanities, Assyriology & Cuneiform Studies

In my role as Professor Andrei Pop’s research assistant I have been responsible for the transcription of a large corpus of documents related to the life and work of the British caricaturist James Gillray (1756 - 1815). Transcribing turn-of-the-nineteenth-century correspondence offered me an experience of the written word that was both qualitatively and visually divergent from much of the textual landscape of the 21st century. Often, the writing we see in the present is mediated by computers rather than the human hand. Further, modes of linguistic expression and conventions around textual correspondence have evolved over time, meaning that both the visual form of late-18th-century writing and the words it contains may reflect a historical difference to present-day readers. These ideas––combined with the visual archive of James Gillray’s oeuvre––inspired me to craft A Phantom Rising: a series of visual poems. Through incorporating images of handwriting, these poems aim to draw attention to the materiality of historical documents and perhaps to inspire other artists to contribute to the growing body of contemporary poetry that bears the direct influence of historical material culture.

Mahria Baker

Prof. Andrei Pop, Art History, Social Thought

College Research Fellow (Hoeft)

Creative Writing, Interdisciplinary Humanities

Discrimination in U.S. healthcare contexts is racialized, with disparities especially visible in non-white populations (Boulware et al. 2003, Hoffman et al. 2016). Accent discordant relationships between healthcare providers and their patients with limited English proficiency (LEP)—that is, when providers and patients do not share the same accent when speaking in the same language—can be a source of tension and a locus of implicit bias. This study seeks to examine racial prejudice and accent ideologies within hospitals, particularly healthcare networks that serve marginalized and vulnerable populations, through a study based at a hospital with a linguistically diverse patient population. In surveying doctors and nurses who work at Mount Sinai Hospital in New York City, this study addresses the following central question: How is communication in medical contexts mediated by linguistically coded racial and/or ethnic bias? Using a matched-guise technique comparing social perceptions of a Spanish-accented versus Standard American English (SAE)-speaking patient describing symptoms, I elicit implicit perception responses from medical professionals. On average, healthcare providers perceive the accented patient as more credible than the unaccented patient and assigned higher trustworthiness scores especially if they were a non-native English speaker or had had cultural competence training. Many participants demonstrated hesitancy about evaluating a patient’s sociodemographic based on their voice alone, which bodes well at least for the consciousness factor of implicit bias. Diversifying the linguistic makeup of healthcare providers, as well as making professional interpreter services available for all patients, is crucial in improving health outcomes for LEP patients. Furthermore, providing cultural, linguistic, and structural competence curricula for medical students and moving towards race-conscious medicine will better prepare the next generation of healthcare providers to work with an increasingly diverse patient population.

Maya Osman-Krinsky

Prof. Sharese King, Linguistics

Linguistics

The Jewish expatriate community of Japan goes back a few centuries, but particularly faced upheaval during the 20th century. As part of a broader nonfiction reporting and essay project, I researched the history of the Jewish community of Japan as documented in news archives and scholarly works. My main research question concerned how Jewish expatriates in Japan saw themselves and their expatriate-heavy communities as fitting in within the global Jewish diaspora. I mainly looked through the Jewish Telegraph Agency’s archives, from 1900 to the present, in order to see how Jewish expatriates in Japan were covered in news, particularly during World War II. Through these archives, I found that Japanese officials and academics were often supportive of the country’s Jewish expatriates, but that Japan’s alignment with the Axis complicated Jews’ position within the country. I also looked for other archival documents online that discuss this community of Jewish migrants, paying close attention to some 1950s-era accounts of the Jewish community of Tokyo from American rabbis around the time when a permanent synagogue opened in Tokyo. In order to place this history in a broader context, I also looked at scholarly works about anti-Semitism and “philo-Semitism” (love of Jews) in Japan. As little has been written about how Japan’s Jewish community fits or does not fit into global dialogues about anti-Semitism and Jewish identity today, this research gave me historical background that ties into my broader BA Thesis essay which is addressing those questions. The broader project also includes journalistic reporting and interviews with current members of the community.

Oren Oppenheim

Prof. Daniel Raeburn, Creative Writing

Creative Writing, History

With the outbreak of COVID-19, many turned to past pandemics to understand the present crisis. Google searches for “Black Death” and “Spanish flu” skyrocketed in March 2020, while sales of plague fiction (such as Giovanni Boccaccio’s "The Decameron," Daniel Defoe’s "A Journal of the Plague Year," and Albert Camus’ "The Plague") surged. The ongoing “Learning from Premodern Plagues” project provides a scholarly introduction to premodern pandemics and their lessons for COVID-19 and beyond. The project, conducted by the Newberry Library, seeks to provide a teaching resource for high school and college courses on pandemic history; to promote engagement with the Newberry’s collections; and to educate the public about the value and relevance of premodern history and culture. The project began as a video series on the Newberry’s YouTube channel, and will soon be accompanied by a Digital Collections for the Classroom (DCC) webpage. Each video features a researcher discussing an object from the Newberry’s collections, offering historical context, and explaining its relevance to the modern day. Previous videos have covered topics such as the huey cocoliztli epidemic in sixteenth-century Mexico, the 588 CE Plague of Justinian in Marseilles, and the eighteenth-century introduction of smallpox inoculation from the Ottoman Empire to England. The companion DCC will include photographs and short writeups for each document, as well as discussion questions and a bibliography. The final project will reflect a diversity of time periods, topics, and geographic locations from the premodern world.

Stephanie Reitzig

Dr. Lia Markey, Center for Renaissance Studies, Newberry Library

CRASSH Scholar

History, Art History

At 1,000 meters below sea level where sunlight no longer reaches, there exists an entirely different world or frontier. This sphere of existence known as the deep sea or the midnight zone reflects a world so drastically different from our own that it can be hard for the human mind to fathom. In this project, I took on the challenge of representing this alien frontier of the deep sea through the medium of music. To do so, I rigorously investigated a particular sub-genre of experimental music called underwater music, seeking to understand the ways through which musicians have used water as an instrument to communicate a particular meaning. I used diving narratives and images of the deep sea to form an understanding or interpretation of the mystery of the sea below. I present my findings in the form of a musical composition which attempts to answer the question of how we can use sound and music to represent the unfamiliarity or unknown of the deep sea.

Teis Jayaswal

Prof. Stephanie Soileau, English, Creative Writing

College Summer Institute Scholar, Dean's Fund for Undergraduate Research Awardee

Interdisciplinary Humanities

Though the works of author Hanya Yanagihara have been lauded for their engrossing novelistic realism, simply accepting them as such can preclude significant and relevant analysis of the themes of desire and power that appear throughout Yanagihara’s writings. My research draws from queer theory, literary studies, and the pederastic literary tradition more broadly in order to illuminate the effects of allegory in Yanagihara’s 2013 novel "The People in the Trees." I argue that although the form of the novel as a collection of putatively factual evidence (such as newspaper articles, letters, testimony, and footnotes) attempts to root the work in realism, less overt aspects of the book’s structure and content ultimately render it allegorical. By bringing in the queer theories of Michel Foucault and Kadji Amin and by utilizing examples from the broader body of pederastic literature (including authors Jean Genet and André Gide), I posit that this shift from realism to allegory forces the reader to also shift from a critique of the protagonist's exploitative and pederastic actions to a critique of the larger structures of power that model and enable such behavior. This analysis is part of my research on depictions of pederasty in literature which, in turn, is one approach to my broader inquiry into pederasty and the relationship between desire, power, and autoeroticism.

Willem Finn Harling

Prof. Kris Trujillo, Comparative Literature

PRISM Research Scholar

Gender and Sexuality Studies, Literature

In the United States alone, there are over 500,000 people who use American Sign Language (ASL) on a daily basis. As speech recognition technology continues to develop and becomes ubiquitous in everyday life, it is important to ensure that applications of speech recognition, such as voice-activated assistants, remain accessible to people who rely on communication in a visual modality. However, sign recognition and generation systems can be difficult to develop, as large, labeled datasets of sign language require a great deal of time and energy to construct. Thus, self-supervised models that can be trained on unlabeled data are of the utmost importance. Vector-Quantized Variational Autoencoders (VQ-VAEs) are one such model that has been used both for phoneme extraction from speech and high-quality image generation. Such models learn to encode an input image into a smaller vector space, sample from this space, and then reconstruct the input image from this sample encoding. This project explores two primary research goals: understanding to what extent VQ-VAEs encodings implicitly learn linguistic features of sign language by virtue of being trained on sign language data, and how such features might be used to build automatic sign generation systems. To help constrain the phonological features the model may learn, this project uses data of ASL Fingerspelling, which consists of 24 handshapes with 8 phonological features that vary among them: including the number and flexion of selected fingers, opposition of the thumb, and flexion of non-selected fingers. Based on experiments conducted so far, even with constraints placed on the phonological space the model may learn, VQ-VAEs may not be particularly adept at implicitly learning linguistic features of ASL Fingerspelling, nor do they appear efficient for automatic sign language generation. Ongoing work continues to explore how tuning hyperparameters of such models may improve performance.

Zack Crenshaw

Prof. Diane Brentari, Linguistics; Prof. Karen Livescu, Computer Science, Toyota Technological Institute

Dean's Fund for Undergraduate Research Awardee

Computing Science, Interdisciplinary Humanities, Linguistics

Quantum computers use the principle of superposition to store combinations of 0s and 1s called qubits. Through entanglement, qubits can be inextricably linked so that changing the state of one qubit instantaneously affects its entangled partner. These properties allow quantum computers to step in where traditional computers struggle, namely computationally intense tasks in AI, molecular design, financial technology, cryptography, and more. However, qubits suffer from decoherence: the decay of a superposition due to environmental interference. The feasibility of quantum computing predicates on a long-lived decoherence time. There is no consensus on the most effective way to combat decoherence. Complex pulse sequences are necessary, but qubit analysis under such conditions is difficult, so a tool that illustrates the qubit’s evolution would be indispensable. Using the QuTip programming language, I will create a tool that searches for, visualizes, and optimizes the pulse sequences that optimize qubit coherence. Visualizing qubit evolution as an animated arrow on the Bloch sphere is a critical step towards the goal of implementing machine learning to find the pulse sequences that protect qubits against decoherence. Ultimately, this project will facilitate our understanding of the prominent decoherence mechanics in experimental platforms like spin-spin interactions and spin-lattice relaxations in solid-state spin qubits.

Ahmed Malik

Prof. Tian Zhong, Molecular Engineering

Liew Family Research Fellow

Computing Science, Physics

Self-assembly of functional nanocrystals has been used to form highly-ordered crystalline superstructures with condensed matter properties, particularly of interest for their electronic band structures. However, these nanocrystals are typically studied using bulky organic ligands for stabilization which prevent dense packing and obstruct extended electronic properties of the system. As an alternative, our approach has examined using small, anionic chalcogenidometallate ligands to better enable densely packed structures. This primarily results in electrostatic-mediated assembly, which correlates closely with the classical colloidal electrostatic stabilization mechanisms predicted by Derjaguin–Landau–Verwey–Overbeek (DLVO) theory. Here, we use DLVO theory and modified DLVO potentials to computationally model the assembly conditions for nanocrystal systems with ionic ligands. Initial results show that these models agree with the trends observed experimentally, showing that DLVO theory is an attractive method for understanding electrostatic assembly in charged nanocrystal systems. However, we also show that traditional DLVO theory alone is incomplete for fully describing nanocrystal systems, which may be due to Pauli repulsion of the ligands. As such, this model has tentatively been improved by introducing a short-distance repulsion term to account for the dynamic presence of ligands on the particle surface. Future work will focus on applying these principles to assemble various semiconducting nanocrystal systems including PbS and PbSe and to explore additional experimental conditions including ionic strength and temperature. Investigating the assembly of ionic nanocrystal systems will help to better understand the foundational mechanics behind strongly coupled nanocrystal superlattices.

Alexandra Hinkle

Prof. Dmitri Talapin, Chemistry

College Summer Research Fellow

Chemistry

Understanding the radial distribution of galaxies inside galaxy clusters is necessary to place constraints on a variety of astrophysical effects and dark matter properties. Of particular importance is the distribution of galaxies in the innermost regions of clusters, where high matter densities lead to interesting interactions such as galaxy stripping and disruption that play an important role in current models of galaxy evolution. However, these cluster centers are difficult to study, as the high density of galaxies produces a blending effect that interferes with our ability to accurately recover the properties of individual cluster members. In this work, we aim to make highly accurate measurements of radial density profiles in the inner regions of galaxy clusters by applying corrections derived from Balrog, an image simulation and bias-detection approach developed within the Dark Energy Survey (DES). In this approach, simulated galaxies with known properties are injected into raw observational data of galaxy clusters. The images are then processed using the official DES reduction pipeline, and the measured properties of the simulated galaxies recovered by the pipeline are compared to the initial properties assigned by Balrog. Any discrepancies can then be used to estimate the detection efficiency and measurement biases of the DES pipeline, which are applied as corrections to measurements of real galaxy properties. Using this approach, we aim to probe the innermost regions of galaxy clusters more deeply than previous studies. The resulting cluster density profiles have the potential to provide new insights into galaxy evolution processes. By comparing our measurements to simulations of cluster evolution under different dark matter frameworks, we will be able to provide new evidence about intra-cluster interactions that until recently have been out of reach due to blending effects.

Alexandra Masegian

Dr. Yuanyuan Zhang, Cosmological Physics, Fermi National Accelerator Laboratory

Astronomy and Astrophysics

Often diagnosed in the late stage due to its vague symptomatology (e.g. bloating, abdominal pain) and a lack of reliable screening tests, ovarian cancer is considered the fifth deadliest cancer for women in the U.S. Although treatment usually begins with cytoreductive surgery and chemotherapy, metastatic ovarian cancer is prone to relapse and chemoresistance, making radiotherapy a common second-line treatment. However, whole abdominal radiation (WAR) is quite toxic at the necessary therapeutic levels, which can lead to serious chronic side effects that decrease patient quality of life. To counter this issue, we have developed europium-doped yttrium oxide nanoparticles (Y2O3:Eu NP) for use in low-energy x-ray activated photodynamic therapy (X-PDT), in which the NP produce cytotoxic reactive oxygen species (ROS) when irradiated with x-ray. The ROS production enhances the radiation treatment, allowing lower x-ray doses to be used, and thus reducing toxic side effects. This study aimed to optimize further modifications of these NP for in vivo use. To prevent NP aggregation and improve biocompatibility, a silica shell coating was added to the NP using a modified Stöber reaction. Transmission electron microscopy (TEM) was performed on samples taken at 6-, 10-, 12-, and 24-hour timepoints to monitor silica shell thickness (5-10 nm is ideal to ensure sufficient ROS production), with shorter reaction times giving more ideal NP sizes. In order to target the tumors specifically, the silica-coated NP were also conjugated with monoclonal antibody 47 (mAb47), which binds interleukin 13 receptor alpha 2 subunit (IL13Rα2), a protein overexpressed in most ovarian cancer cell lines, with high affinity. Antibody conjugation to the NP was proposed to work via a PEG-NHS linker, and fluorescence spectroscopy studies were conducted to confirm pegylation. Ultimately, these mAb47-conjugated nanoparticles will be used to conduct in vivo studies with mice to determine their efficacy.

Ariel Pan

Prof. Chin-Tu Chen, Radiology

Dean's Fund for Undergraduate Research Awardee, Liew Family Research Fellow

Chemistry

A manifold is a mathematical space that can locally be mapped with Cartesian coordinates, but may have complex properties globally. One example of a manifold that is easy to conceptualize is a sphere, which resembles a 2-dimensional plane when one only looks at points near a fixed center. Manifolds have been widely studied in many areas of mathematics and statistics, which has led to the formation of the Manifold Hypothesis. This hypothesis states that empirical probability distributions in various sciences (e.g. the results of mRNA sequencing in Biology) tend to possess an underlying, low-dimensional structure. Reducing the dimension of complex datasets while retaining key information is an important focus of modern methods of data analysis, and the Manifold Hypothesis gives reason to believe this can always be achieved. One of the most common dimensionality reduction procedures is Principle Component Analysis (PCA), which empirically determines the best linear approximation of specified dimension to a dataset. Little et al. (2017) proposed an adaptation of PCA called multiscale Singular Value Decomposition (mSVD) that is able to recover the local Cartesian structure of point clouds with high fidelity. It is well known that local Cartesian charts like this can be “stitched together” to allow for a unified, global representation of the underlying manifold. My research over the past academic year has focused on the development of computational methods to perform this stitching via manifold triangulation. By creating a robust process for triangulating simpler and eventually more complex manifold structures, we grow one step closer to learning the global manifold representations underlying empirical point clouds.

Isabella DeClue

Prof. Lorenzo Orecchia, Computer Science; Ryan Robinett, Computer Science

College Research Fellow

Computing Science, Mathematics, Statistics

The number of detected exoplanets has boomed over the last decade. Over 4,000 planets have been confirmed, primarily by the Kepler mission, with thousands more expected from TESS and PLATO. This wave of detections has revealed an incredibly diverse population of planets and system architectures. Of particular interest are resonant chains, planetary systems containing multiple adjacent planet pairs with orbital periods near an integer ratio (2/1, 3/2, 4/3, ...). These systems are a natural outcome of standard planetary formation theories and are a potential fossil record of planetary migration. Here we search for evidence that resonant pairs are correlated with one another; if so, the probability that a given planet pair is resonant is higher if an adjacent pair is resonant as well. We conduct this study using Bayesian statistics and Monte Carlo simulations. We first model the observed period ratio distribution as a smooth continuum with Gaussian spikes at the resonances. From this model, we calculate the likelihood that a given period ratio is resonant or merely part of the continuum. For each observed exoplanetary system, we then calculate the probability of forming a similar system, by random sampling of period ratios. Our preliminary results indicate there is an excess of exoplanetary systems with three or more adjacent resonances, suggesting a correlation between resonant pairs. We are now investigating whether this trend is an observational artifact, by investigating a series of prior distributions for the planet sizes, orbital periods, and inclinations.

Jared Siegel

Prof. Daniel Fabrycky, Astronomy and Astrophysics

College Research Fellow (Hoeft)

Astronomy and Astrophysics

As computing becomes less expensive, the interdisciplinary techniques of computational biology have become salient and effective. Molecular dynamics (MD) is a computational strategy which probes physical properties by simulating the movements of atoms. MD and other bioinformatics techniques can address properties about a system of interest more easily than physical experimentation. Computational simulation is typically more cost efficient than physical experimentation and allows for high-throughput virtual screening to facilitate expedited drug discovery and molecular engineering. While improvements in algorithms have allowed for the simulation of complex billion atom systems, increases in the length of simulated time have proven elusive. Presently, MD techniques are limited to microsecond timescales on commercially available processors. This means key rare events (e.g., enzyme docking, protein folding, protein aggregation), are sampled rarely and have high uncertainties. Thus, advancing the length of simulated time, the “time-scale barrier,” is of key research importance. In this work, I evaluate the Latent Space Simulators approach (LSS), first formalized by the Ferguson Lab in August 2020. LSS is a machine learning approach that learns a kinetic model over the limited MD training data, which enables the observation of rare states like protein folding, enzyme docking, and protein aggregation. LSS represents an enabling technology to simulate the kinetic behavior and properties of large molecular systems of biological relevance that are currently outside the reach of existing computational approaches.

Joseph Aulicino

Prof. Andrew Ferguson, Molecular Engineering

Liew Family Research Fellow

Computing Science, Engineering

Periodic spatial patterns of vegetation growth have been observed in dryland ecosystems. These patterns are thought to arise through self-organization in the water-limited environments that support them, and reaction-advection-diffusion models have suggested that the patterns are a precursor to ecosystem collapse as water becomes increasingly scarce. On gently sloped terrain, dryland vegetation patterns often appear as repeating bands of dense vegetation that are decameters wide and spaced on the order of hectometers apart, with bare soil in between. While observations indicate uphill migration of the bands on a century timescale, the water is input during rainstorms that last just a few hours. We explore the impact of assumptions about the fast hydrology associated with overland flow and infiltration during rainstorms on the slow dynamics of the patterns. We consider both temporally periodic and stochastic rain input within a conceptual fast-slow switching model that exploits the difference in timescales involved.

Lily Liu

Prof. Mary Silber, Computational and Applied Mathematics

Liew Family Research Fellow

Mathematics

The origin of superconductivity in the doped semiconductor strontium titanate (STO) is not well-understood. Its persistence even at very low carrier densities violates key assumptions of conventional Bardeen-Cooper-Schrieffer (BCS) theory, particularly in that the phonons in this regime are much faster than the electrons to which they are coupled. The formation of bipolarons, two-electron bound states in a polar medium, serves as a potential electron pairing mechanism beyond the BCS paradigm. Here, the existence of bipolarons in STO is probed using semiclassical analyses and variational calculations to estimate the system’s ground state energy in the weak- and strong-coupling limits of the electron-phonon interaction strength in STO, respectively. It is found that no binding exists in the weak-coupling regime, whereas the strong-coupling regime exhibits strong binding. We explore the intermediate regime through Quantum Monte Carlo (QMC) simulations, which enable exact calculation of the ground state energy.

Lisa Lin

Prof. Peter Littlewood, Physics

College Research Fellow, Dean's Fund for Undergraduate Research Awardee

Physics

Massive objects create gravitational wells that warp the path of light. When the light of background (source) galaxies passes close to foreground (lens or tracer) galaxies, it gets perturbed, distorting the image of the source galaxies we receive. We can measure this distortion, known as gravitational lensing shear, to estimate the mass of objects not visible to other techniques. In the case of a spherical distribution of matter, the shear at any point will be oriented tangentially to the direction of the center of symmetry. Due to the random intrinsic orientation of galaxies, this distortion of galaxy shapes is rarely noticeable or measurable for a single object. To obtain a signal, we must average the tangential component of the shear over many lens-source galaxy pairs. For our research, we performed this measurement, called galaxy-galaxy lensing, to obtain the tangential shear profiles of low-surface brightness galaxies: faint, diffuse galaxies composed primarily of non-baryonic matter or dark matter. Owing to their dimness, we know little about the structure or formation of these galaxies. Our next steps include fitting these shear measurements with a Halo Occupation Model (HOD) to profile the dark-matter make-up of these galaxies and measure their mean halo mass.

Nathalie Chicoine

Prof. Chihway Chang, Astrophysics, Dr. Judit Prat, Astrophysics, Chang Lab

Astronomy and Astrophysics

One of the most striking features of Ellesmere Island Ice Shelves is the rolling surface topography, known as ice rumples. However, the mechanisms governing how these sinusoidal surfaces arise, as well as how these features persist, have not fully been tested. In this research, we seek to test which natural forcings could initiate and maintain ice rumples through numerical modeling and theoretical analyses. Identifying the dominant forcings and associated time and pressure scales in formation and persistence is important groundwork for understanding ice shelf stability, where mass flux and ice shelf ablation directly contribute to global mean sea level.

Niall Coffey

Prof. Douglas MacAyeal, Geophysical Sciences

College Research Fellow

Geophysical Sciences

Many tools have been created to analyze various representational systems; the Inca Khipus present unique opportunities by virtue of being a knot-cord system. Khipus were one of the primary communication methods of the pre-Hispanic Inca Empire. Until now, we have understood their numerical codes. The research of Gary Urton has shown the majority of patterns on Khipus contain numerical information, however we still need to decode non-numerical patterns and fully analyze the patterns contained in the numbered parts. Under Dr. Jon Clindaniel, I have worked to analyze Khipu data and create resources for other researchers. Throughout the process, I have incorporated social scientific and qualitative studies from people such as Gary Urton and Marcia and Robert Ascher. Using the Open Khipu Research Laboratory data, I have modified a number pattern recognition code from Dr. Clindaniel’s thesis to work for colors, knot types, and other descriptive elements of the Khipus. Furthermore, I have prepped the Open Khipu Research Laboratory Github for use by both technical and non-technical parties. In the presentation, I will discuss these advances and how they will be used both by me and other researchers to continue investigating questions on the Inca Khipu system.

Nyaga Kariuki

Prof. Jon Clindaniel, Computational Social Science

College Research Fellow

Computing Science, Interdisciplinary Humanities

Recent years have witnessed rapid advancement of machine learning (ML) technologies. Speech recognition, medical diagnosis, smart assistants, self-driving cars, and many more technologies backed by ML are drastically transforming our lives. In particular, numerous cloud service providers (e.g. Google Cloud and Amazon Web Service) have developed a range of machine learning products that market on their accessibility to software developers. These ML application programming interfaces (APIs) enable developers to easily incorporate machine learning solutions into software systems. Unfortunately, ML APIs are challenging to use correctly and efficiently, given their unique semantics, data requirements, and accuracy-performance tradeoffs. Much prior work has studied how to develop ML APIs or ML cloud services, but not how software applications are using ML APIs. We present the first in-depth study of real world applications using ML cloud APIs. We manually study 360 representative open-source applications that use Google or Amazon AWS cloud-based ML APIs, and identify 70% of these applications as containing API misuses in their latest versions that degrade functional, performance, or economical quality of the software. These misuses lead to various types of problems, including 1) reduced functionality, such as a crash or a quality-reduced output; or 2) degraded performance, like an unnecessarily extended interaction latency; or 3) increased cost, in terms of payment for cloud services. Their root causes are all related to unique challenges for ML APIs. We have generalized 8 anti-patterns based on our manual study and designed several checkers and small API changes (in the form of wrapper functions) that both check for and handle common errors. Hundreds of more applications are identified as containing misuses by our checkers, beyond the 360 projects in the initial study.

Shicheng Liu

Prof. Shan Lu, Computer Science

College Research Fellow

Computing Science

The COVID-19 pandemic took a major toll on the mental health of students and teachers, especially low-income students. Historically, schools have supported student and teacher wellbeing through social and emotional learning (SEL), which is the process through which people develop social and emotional skills such as self-management and relationship skills. Drawing on interviews with teachers from across the United States, this study focuses on the impact of the COVID-19 pandemic on SEL implementation in Title I schools. Using CASEL’s Guide to Schoolwide SEL as the basic SEL framework for schools, this research examines teachers’ experiences of implementing SEL during the pandemic. This study finds that only schools that had fully integrated SEL into the school structure continued their SEL practices, while teachers created new SEL techniques and methods in response to the pandemic. Policy recommendations include integrating SEL into school structures and incorporating teacher voice and leadership into SEL planning and implementation.

Amara Cohen

Dr. David W. Johnson, Consortium on School Research

Dean's Fund for Undergraduate Research Awardee

Education & Scholarship of Teaching

Although the recession of 2008 revealed the catastrophes of over-indebtedness, the ensuing decade witnessed a significant rise in auto-loan debt, which stood at its highest level in history in 2019: over a trillion dollars. Simultaneously, auto-loan delinquencies also steadily increased, with heavy concentrations in the South. This is especially surprising since the South has the fewest proportion of debtors in the country, and the North-East, which has the highest proportion of auto-loan debtors, has the lowest proportion of delinquencies. What accounts for this spatial mismatch between hotspots of debt and delinquency? My first hypothesis is that those areas which receive greater federal welfare for medical expenses, housing, and income support will not be as delinquent because these supports mitigate financial pressures. Further, I hypothesize that those areas in which employment centres are more accessible will also be less delinquent due to lower dependencies on automobile ownership. Contrary to my first hypothesis, I find that the relationship between income support and delinquency is such that the areas which receive greater income support are also more likely to be more delinquent, which shows that instead of providing substantial pecuniary support, income support programs are proxy measures for a precarious economic situation. However, in key parts of the population, medical welfare programs do mitigate delinquencies: among counties where the majority of the population are people of colour and low-income counties, those which receive greater medical welfare are likely to be less delinquent. With respect to my second hypothesis, I find that especially among counties in the South where delinquency ranks the highest in the country, those with shorter commutes to work are also likelier to have lower levels of delinquency. Above all, my findings reveal that different regions face different challenges with respect to their debt burdens, and thus require case-specific solutions.

Atman Mehta

Prof. Luc Anselin, Sociology, Center for Spatial Data Science; Dr. Julia Koschinsky, Center for Spatial Data Science

Social and Behavioral Sciences

Rationality is a key assumption undergirding many of the predominant theories in International Relations. Specifically, many statist theories assume that although individuals might have imperfect rationality, a state’s entire bureaucratic apparatus necessarily has constraining effects on imperfect rationality. As a result, these theories generally believe state decision makers are forced to operate nearly perfectly rational. I attempt to empirically test these assumptions, through a textual analysis of almost two decades of Presidential intelligence briefings. Specifically, I ask if decision maker’s individual emotive responses can be seen increasing in times of crisis and heightened tension. To do so, this paper leverages LIWC’s dictionary and wordcount based methods for measuring textual sentiment to determine emotional language use at the entry level within the PDB corpus. We measure across several dimensions, but specifically focus in on positive and negative language use in entry references involving certain states during periods of conflict and crisis. I find that over the course of crises and engagements, briefing language becomes more emotionally charged than otherwise, with a notable large increase in negative sentiment becoming more prevalent. Individual emotions can cloud judgement, and these findings confirm this. Since this corpus’s output was briefings that sat atop the US intelligence bureaucracy, the finding of emotional language here is odd from a statist view. These findings show that in fact state decision makers cannot simply be constrained to rationality by the state and its bureaucratic apparatus. Moreover, these findings broadly serve to raise questions around rational actor assumptions in crisis negotiation and conflict resolution as it reveals state decision maker’s clouded judgement in these situations.

Austin Christhilf

Prof. Austin Carson, Political Science

CRASSH Scholar

Political Science

Unsustainable development and discriminatory urban policies have precipitated blight in Chicago’s Black and low-income communities. Brownfields—sites of former industrial activity that have the perception of contamination—are indicators of blight and disinvestment. To address the issue of brownfields in the city, Chicago launched the Chicago Brownfields Initiative in 1993 to remediate sites and attract developers to help spur economic revitalization. This paper investigates the issue of brownfields and brownfield redevelopment in Chicago, ultimately using findings to help craft a policy proposal for the future equitable and sustainable development of these sites. This paper uses sociological, environmental justice, and human rights concepts to create a theoretical framework to evaluate aspects of brownfields and redevelopment, analyze the impact on communities, and inform policy recommendations. Spatial analysis and qualitative methods are used to explore this important issue in Chicago and investigate its impact on local communities. This study addresses gaps in previous research and provides a much-needed update on sites remediated under the Chicago Brownfields Initiative. Given the recent pushes for fostering sustainable urban development and addressing issues of historic inequity, this paper aims to signal the potential of brownfield redevelopment in larger sustainability and social justice efforts.

Cinque Carson

Prof. Chad Broughton, Public Policy

Environmental & Urban Studies

Throughout the twentieth century, the United States promoted instability, poverty, authoritarianism, and incredible violence, etc., in Latin America through decades of multi-dimensional American intervention. For instance, in Chile, the United States helped create the socioeconomic conditions which engendered the 1973 coup d’état; furthermore, America supported the Pinochet Regime which fostered inequality, repression, etc. Yet, in the Central Intelligence Agency’s President’s Daily Briefings (PDBs), America’s role in the devastation of Chile, and in the devastation of Latin America more broadly, goes largely undiscussed; that is to say, the reality and consequences of American intervention are ignored and seemingly “forgotten.” Why? Through a socio-historically contextualized analysis of PDBs from the late twentieth century, I aim to use the case study of American intervention in Chile to theorize political amnesia: a political phenomenon by which the American government, and to some extent American society, “forgets” and thus abdicates its moral responsibility for past atrocities. First, I will briefly historicize my discussion by illuminating the reality of American intervention, and its consequences, in late twentieth century Chile. Then, through an in-depth analysis of contemporary PDB entries regarding Chile, I will show how even at its highest, most classified level, the American Government tonally and substantively “forgets” its role in promoting instability, poverty, and authoritarianism, etc., in Chile. Finally, I will suggest that this phenomenon, in Chile and in other contexts, has enabled the United States to abdicate its moral responsibility to certain populations, notably Latin American immigrants to the United States. Overall, along with implicitly characterizing PDBs as inherently political documents, I hope to suggest that the United States avoids righting its past wrongs by simply pretending they never happened.

Diego Quesada

Prof. Austin Carson, Political Science

College Research Fellow

Political Science, Human Rights

For decades, scholars have endeavored to put together a comprehensive theory of African law that takes into account both the influence of colonial legal structures and surviving traditional customs. My research project engages with this effort by exploring the practices of traditional tribunals in rural villages in Balaka, Malawi: how do these communities construct and engage with concepts of law and justice through tribunals? Based on this knowledge, can we better understand whether customary law and constitutional law are competing for dominance, integrating into cooperative but dual polities, or evolving together into a customized hybrid system? My data for this project include informal interviews with twelve traditional headmen, observations of tribunal proceedings in two locations, and tribunal case records from three villages, all collected from the region surrounding the town of Balaka. First, I use these sources to identify what kinds of laws are organizing these village communities. I find that headmen enforce rules that originate from both traditional customs and state laws and do not differentiate between the two when describing their role in the community. As a result, traditional leaders observe a rule of law whose source of authority is ambiguous. Second, I look at how traditional tribunals approach conflict resolution and pursue justice. These tribunals have been influenced by the constitutional law of Malawi but are rooted in traditional African customs and the practical experiences of these tight-knit, interconnected village communities. They therefore prioritize communal values such as reconciliation and family unity over individual liberty. Overall, my analysis describes a hybrid system of integrated government where tradition and practicality interact with centralized democracy to inform the growth and development of rural Malawian communities. This hybridized interpretation of local law challenges the dichotomy that is typically drawn between customary law and constitutional law in post-colonial African states.

Emily Williams

Dr. Sarah Johnson, Law, Letters, and Society; Prof. Jenny Trinitapoli, Sociology

College Research Fellow (Hoeft)

Legal Studies

In recent years, historians have produced a body of scholarship that has benefitted from the merging of two previously separate fields of historiography: that of the civil rights and Black liberation movements of the 1950s and 1960s, and that of Cold War international relations. A key argument advanced by this body of scholarship is that the American claim to leadership of the free world on the international stage was undermined by obvious racial injustice at home, prompting U.S. executive authorities to favor federal civil rights legislation. Scholars have treated the heating up of the Vietnam War in the mid-60s as the endpoint of the Cold War-civil rights connection, arguing that American military intervention in Vietnam displaced American racial injustice as the crucial factor in international prestige. However, using the recently declassified Presidential Daily Briefs (PDBs), daily intelligence documents produced by the Central Intelligence Agency (CIA) for the President’s personal review from the Kennedy to the Ford administration, I argue that the periodization of the Cold War-civil rights connection can be extended into the late 1960s. These new sources, specifically a cluster of PDBs offering detailed reports on North Vietnamese broadcasts on American racial injustice, demonstrate that foreign perceptions of American race relations continued to be an issue of significant concern to the late Johnson administration. But here, the influence of Cold War foreign policy on racial justice at home is qualitatively different from its form in the previous decade. In the context of rising antiwar dissent and perceptions of weakness abroad, the attitude of U.S. authorities towards the racial justice movements of the late 1960s turned more repressive, identifying Black activists with the antiwar movement and the revolutionary threat of the North Vietnamese themselves.

Ethan Hsi

Prof. Austin Carson, Political Science

College Research Fellow

History

“Who’s Minding the Neighborhood?: The Role of Neighborhood Associations in Neighborhood Planning,” is aimed towards creating a better understanding of the role of neighborhood associations in urban planning in Chicago. In order to ascertain the scope and consequence of neighborhood-level organizing when it comes to guiding the future of the neighborhood, we interviewed the leaders of 17 neighborhood associations in Chicago. We wanted to know, first, how neighborhood organizations view themselves and their role, and second, how neighborhood organizations approach neighborhood planning. Through these interviews, we found that these organizations generally fall within one of three types, and that each type employs a different kind of approach to neighborhood planning: servicing, protection, and promotion. Servicing-oriented associations are primarily concerned with providing more social services and resources to residents as a means of empowerment. In the case of protection-oriented associations, planning is focused on blocking developers and changing zoning codes, mainly in order to preserve the quality and character of the neighborhood. Finally, promotion-oriented associations serve primarily as amplifiers of small events and tactical urbanism geared towards promoting the economic vitality of retail corridors. While this research is limited to a relatively small number of associations within one American city, our findings point towards an overall limited role played by neighborhood associations in neighborhood planning activities. In none of these approaches did we find neighborhood planning taking place in a proactive sense—where residents come together to formulate and implement a vision for their neighborhood.

Eugen Craciunescu

Prof. Emily Talen, Urbanism Lab

College Research Fellow

Environmental & Urban Studies, Public Policy

This thesis analyzes the impact of Medicaid expansion on substance use disorder (SUD) referral patterns in order to assess whether lowering the cost of care changes how individuals enter treatment. While existing literature examines the overall effect of Medicaid on SUD utilization, this paper utilizes both quantitative and qualitative data to assess whether expanding Medicaid changed the referral source of patients, including self-referrals and referrals from criminal justice sources. TEDS-A data from 2010 to 2018 is used to conduct a differences-in-difference and event study model to assess the causal effect of Medicaid expansion. Qualitative interviews with healthcare providers, policy advocates, and researchers provide a broader understanding of the mechanisms behind how Medicaid expansion impacts referral source. This paper finds that while Medicaid expansion did not necessarily change patient referral habits, it provided a payment source for otherwise uninsured patients entering treatment. These findings suggest that payment is a necessary but insufficient condition for care.

Jacob Sims Speyer

Prof. Harold Pollack, Social Work, Policy, and Practice, UChicago Urban Labs; Samantha Steinmetz, Chapin Hall

Public Policy

The Covid-19 pandemic has negatively impacted almost all small businesses in the US and around the world. Childcare centers and home providers in the US are no exception. Many programs had to close during the first few months of the pandemic, thus exacerbating pre-existing financial vulnerabilities in the childcare industry. Our research is part of a larger study that assesses the effects of the pandemic on childcare centers and home-based providers in Cook County and downstate Illinois. We consider financial challenges facing center and home-based programs. Our research investigates the different financial challenges providers have experienced because of the pandemic as well as their experiences applying for and receiving governmental and non-governmental COVID-related assistance. The sample includes 75 childcare program directors purposively sampled by type of care, region, size, and dependency on public funding. Interviews were approximately 60-minutes in length, conducted via Zoom, and transcribed, coded and analyzed using Dedoose software. Providers who remained open or reopened experienced enrollment declines due to reductions in demand and/or public health measures related to social distancing and capacity restrictions in classrooms. These conditions led to a substantial reduction in revenue, while at the same time costs to care provision increased because of new expenses related to additional cleaning supplies and personal protective equipment. Our findings will consider variation in experiences by type of care (centers vs homes), program size, dependency on public funding, and geography. The implications for the child care and early education market will be discussed.

Jacqueline Lewittes; Fady Shokry

Prof. Julia R. Henly, Crown Family School of Social Work, Policy, and Practice

College Research Fellow (Global Health)

Education & Scholarship of Teaching, Public Policy, Social and Behavioral Sciences

Current scholarship uses machismo, a concept characterized by complex historical and socioeconomic interactions manifested through values, ideology, and behaviors, to examine Latinx masculinity. However, researchers rarely consider how machismo impacts Latinx, openly queer men and their gender performance. Building upon concepts from masculinity studies, the present study aims to create a framework to reconceptualize the cultural value of machismo within U.S. born, second-generation immigrant, queer Latinx males. Drawing from 15 qualitative interviews, this study examines how the participants negotiate their cultural expectations of machismo within Latinx communities when they disclose their non-heteronormative sexuality. Specifically, I focus on the strategies my participants deploy to navigate potential violence under the expectations of machismo while attempting to maintain personal autonomy. These findings highlight the need to contextualize studies of masculinity within narratives and intersections of both ethnic spaces and sexuality in order to get a clearer understanding of structural properties affecting and being reproduced by individuals.

José Morin

Prof. Kristen Schilt, Sociology, Center for the Research of Gender and Sexuality; Xiaogao Zhou, Sociology; Agatha Slupek, Political Science

Gender and Sexuality Studies, Social and Behavioral Sciences, Latinx Studies

Attributions of innovative behavior positively correlate with the agent's interest in thinking (Need for Cognition). Need for Cognition is a self-assessed measure of one’s enjoyment of problem-solving, but does not relate toone’s problem-solving abilities. In previous studies, problem-solving relating to Need for Cognition has not been differentiated into more than one type; however, this study suggests that one’s Need for Cognition rating may be impacted by the type of problem-solving that they are considering at the time of the rating. Here, we looked at two distinct types of problem-solving. The first involves a process in which one identifies a problem and then tries to come up with a solution (known as problem-first problem-solving). In the second, one encounters an object and reasons about its potential uses, which might lead one to understand the object as relevant to solving a problempreviously unknown to one (a process referred to as need-solution pair recognition). In an experiment (N = 61), we examine how individual differences in Need for Cognition relate to both forms of solution-generation. In two different blocks of trials participants were instructed to generate solutions for a given problem or to come up with uses for novel objects that could solve a problem with the order of these blocks counterbalanced across participants. We measured the number of unique solutions generated for presented problems (problem first block) and number of unique uses for presented objects (object first block). Results demonstrate that object-based solution-generation is positively correlated with Need for Cognition, while problem-based solution-generation is only marginally so. Ourfindings demonstrate the importance of expanding our understanding of innovative behavior to also include solution-finding via need-solution pair recognition.

Madeleine Ferrara

Prof. Howard Nusbaum, Psychology; Dr. Shannon Heald, Attention, Perception, EXperience (APEX) Lab

College Research Fellow

Social and Behavioral Sciences

Randomized controlled trials (RCTs) are considered the highest level of evidence in the hierarchical system used to classify the vast number of research publications available. Evidence derived from RCTs is thus given particular importance in evidence-based medicine, and consequently in clinical guidelines. Despite this, RCTs can lack external validity due to the narrow clinical settings in which they are conducted. Since race and ethnicity are associated with prevalence and outcome of diseases, racial and ethnic composition of patients included in RCTs can alter their results and thus the generalizability of the conclusions of the studies to the wider population. The primary aim of this study was to determine if the racial and ethnic background distributions of RCT study populations reflect that of the U.S. general population. In this retrospective analysis of the racial and ethnic compositions of RCTs in the American Journal of Obstetrics and Gynecology and Obstetrics and Gynecology from January 2010 to April 2020, the racial distributions of study participants were compared to those of the parous and general female populations of the United States. Chi-square analyses showed significant deviations from the U.S. general population in both Obstetrics and Gynecology (p<0.05). We observed an overrepresentation of Black participants and an underrepresentation of Asian participants in overall Obstetrics and in most subcategories as well as in the overall Gynecology category and most Gynecology subcategories. White participant representation was similar to the general population on average, but with wide variation across studies especially among the subcategories. Hispanic ethnicity was overall underrepresented. RCTs in the field of Obstetrics and Gynecology conducted in the U.S. deviate from the general population with regard to racial and ethnic distributions, prompting a mindful consideration of the characteristics of study conclusions as they are often incorporated into practice guidelines geared towards the general population.

Noel Cercizi; Yifei Wu; Crystal Park

Dr. Yuzuru Anzai, Obstetrics and Gynecology, Lenox Hill Hospital

Biological & Health Sciences, Social and Behavioral Sciences

This presentation introduces a theory of a new form of the Chinese authoritarian regime after 2013, in contrast to Maoist and bargaining authoritarianisms presented in earlier literature. The new form is characterized by the fact that the central government directly regulates local officials and social actors by selected institutionalization and repression divided by a no-trespassing line. Based on ethnographies and interviews, the author illustrates the new form with a case study of xinfang in the realm of stability maintenance. Specifically, the author shows how both local officials and petitioners are institutionalized and repressed in different scenarios. This case draws together central-local relations and state-society relations in an authoritarian regime and addresses the institutional and non-contentious elements of stability maintenance, which are largely neglected in previous works on authoritarian resilience.

Tianhao Hou

Prof. Elisabeth S. Clemens, Sociology

College Summer Research Fellow

Social and Behavioral Sciences

There is substantial variation in individuals’ responses to potentially stressful experiences. One source of resilience may include psychological factors drawn from self-determination theory (SDT), which posits that individuals better cope with adverse circumstances when pursuing intrinsically motivated activities. Here, we test the hypothesis that such eudaimonic living buffers psychological distress, particularly during the COVID-19 pandemic. Adults (N = 240) from New York City, North Carolina, and South Carolina completed measures of mental wellbeing as well as intellectual and civic virtues. Our findings suggest that greater levels of wise reasoning and resilience buffer pandemic-related psychological distress. Higher reported distress was associated with increased civic engagement. Additionally, wise reasoning, but not resilience, mediated the association between COVID-19 stressors and psychological distress. We discuss possible accounts of the relationships between intellectual and civic virtues with respect to mental wellbeing, suggesting that some societal challenges uniquely require eudaimonic living for adapting to changing social norms.

Yena Kim

Prof. Howard C. Nusbaum, Psychology, Attention, Perception, & EXperience (APEX) Lab

CRASSH Scholar

Social and Behavioral Sciences