2021 Posters

1Hayley BoigenzahnMeasures and mathematical models of prebiotic peptide formation
Studying the origin of life can help us better understand our own biology and potentially design highly adaptable systems to use as platforms for molecular discovery. It can also help us recognize the potential for life on other worlds by recognizing what molecules allowed life to form on our own planet.
We aim to quantitatively model the formation of peptides in a prebiotic environment to identify signs of molecular cooperation, autocatalysis, and other chemical processes that may have contributed to the origin of life.
2Em CraftDiscoverIT, UW-Madison, and WID Provide Your IT Services
3Michael FerriswivDM: covid vaccine allocation modeling for Wisconsin
Wisconsin made a remarkable about-face in two months, shooting up state rankings to become a leader in vaccinations. Ninety percent of the doses delivered to Wisconsin have been administered. How? Rather than a focus on high-volume sites, the state sent vaccines to public and private health-care providers, in an approach one state health department official likened to an even spread of peanut butter. This poster helps explain the model behind that allocation.
4Nisha IyerProbing Spinal Cord Diversity With Human Stem Cell Differentiation and Single Cell RNA-Sequencing
The spinal cord contains billions of neurons, with a huge diversity of subtypes enabling sensory, proprioceptive, and motor function. However, current human stem cell-based in vitro models and prospective cell transplantation therapies fail to reflect the significant regional specificity of spinal cells.
Here we recapitulate the full diversity of spinal cell types along both the rostrocaudal (R/C) and dorsoventral (D/V) axes with chemically defined, scalable protocols using human pluripotent stem cells (hPSCs). We first induce R/C patterning to generate neuromesodermal cells from a defined anatomical level, then instruct these cells to become early spinal progenitors. By providing appropriate D/V signaling, spinal progenitors can be sub-specified to generate tunable ratios of motor neurons (MNs) and locomotor interneurons (INs) from the ventral spinal cord, or TGF-β-dependent proprioceptive INs and TGF-β-independent sensory INs from the dorsal spinal cord. Cultures with over 95% neuronal yield can be generated in as little as 19 days, and these protocols can be used modularly to generate phenotypes from different anatomical levels. Single-cell RNA-sequencing reveals regionally specified neurons with discrete Hox gene profiles, representation of all major motor and somatosensory spinal cell types, and the presence of human-specific cell populations. We applied the computational alrogithm Arborteum to infer modules of coexpressed genes between subpopulations, facilitating the discovery of novel markers defining region-specific neuronal types. Altogether, this dataset enables characterization of the diversity of human spinal cells for the first time. We anticipate that access to these cells will advance a mechanistic understanding spinal development, expand the potential and accuracy of in vitro models, provide insight into novel therapeutic targets, and represent clinically relevant populations for cell transplantation.
5Julian Katz-SamuelsAn Empirical Process Approach to the Union Bound: Linear and Combinatorial Bandits
This paper proposes near-optimal algorithms for the pure-exploration linear bandit problem in the fixed confidence and fixed budget settings. Leveraging ideas from the theory of suprema of empirical processes, we provide an algorithm whose sample complexity scales with the geometry of the instance and avoids an explicit union bound over the number of arms.
Unlike previous approaches which sample based on minimizing a worst-case variance (e.g. G-optimal design), we define an experimental design objective based on the Gaussian-width of the underlying arm set. We provide a novel lower bound in terms of this objective that highlights its fundamental role in the sample complexity. The sample complexity of our fixed confidence algorithm matches this lower bound, and in addition is computationally efficient for combinatorial classes, e.g. shortest-path, matchings and matroids, where the arm sets can be exponentially large in the dimension. Finally, we propose the first algorithm for linear bandits in the the fixed budget setting. Its guarantee matches our lower bound up to logarithmic factors.
6Alexis LawtonRevealing dynamic protein acetylation across subcellular compartments
Acetylation of the ε-amino of lysine residues is a widespread, reversible post-translational modification that regulates many cellular functions, including protein-protein interactions, protein-DNA interactions, cellular localization, protein stability, and enzymatic activity. Historically, the first well-characterized example of acetylation was found on histone tails, but more recent proteomics analyses have identified thousands of acetylation sites on non-histone proteins.
To address which acetylation sites are functionally relevant, our lab has developed and validated an improved, complementary proteomics method to calculate global and site-specific acetylation stoichiometry. Using the acetylation stoichiometry analyses, the goal of this study is to understand acetylation dynamics in serum stimulated cells and to define the cellular mechanisms involved in regulating the unique responses and rapid changes in protein acetylation.
7Sarah LeichterSubstrate deformation regulates DRM2-mediated DNA methylation in plants
DNA methylation is a major epigenetic mechanism that critically regulates transposable element silencing and genomic stability. In plants, DOMAINS REARRANGED METHYLTRANSFERASE 2 (DRM2) preferentially mediates CHH (H= C, T, A) methylation, a substrate specificity distinct from that of mammalian DNA methyltransferases. However, the underlying mechanism is unknown.
Here we report the structure-function characterizations of DRM2-mediated DNA methylation. An arginine finger from the catalytic loop intercalates into the non-target strand of DNA through minor groove, inducing large DNA deformation that impacts the substrate preference of DRM2. The target recognition domain of DRM2 stabilizes the enlarged major groove via shape complementarity rather than base-specific interactions, permitting the substrate diversity. The engineered DRM2 C397R mutation introduces base-specific contracts with the +2-flanking guanine, thereby shifting the substrate specificity if DRM2 toward CHG DNA. Together, this study uncovers DNA deformation as a mechanism in regulating DRM2 CHH substrate specificity and illustrates methylome complexity in plants.

8Shruthi MageshGenetic determinants of surface colonization by abundant rhizosphere bacterium Flavobacterium johnsoniae
The rhizosphere encompasses the diverse community of microbes living on the surface of plant roots and in their vicinity in soil. This community is a key determinant of plant health because it provides nutrients, mediates disease, and influences plant development. Members of the gram-negative genus, Flavobacterium, are abundant in the rhizosphere and soil.
The mechanistic basis for Flavobacterium colonization of the rhizosphere and soil is poorly understood, but it seems likely that colonization of surfaces such as roots and soil particles is essential for the maintenance of Flavobacterium in the rhizosphere. Sand is a common and easily manipulated soil particle, making it ideal for studying surface colonization in the laboratory. To dissect the genetic determinants of surface colonization, we conducted a massively parallel mutant screen (InSeq) of F. johnsoniae in the presence and absence of sand. This high-throughput screen revealed 25 genes likely to be important for sand colonization, including genes predicted to be involved in cell membrane biogenesis; motility; transport of amino acids, nucleotides, or inorganic ions; and metabolism. Most of the genes identified by the InSeq screen are uncharacterized, encoding either hypothetical proteins or proteins of unknown function. We validated the involvement of seven genes in colonization by constructing in-frame deletions in genes fjoh_0651 (hypothetical protein), fjoh_1448 (exosortase family protein XrtF), fjoh_1449 (exosortase F system-associated protein), fjoh_2379 (LysM peptidoglycan-binding domain-containing protein), fjoh_1653 (SprA), fjoh_1555 (PorV) and fjoh_1556(PorU) and testing the resulting mutants’ ability to colonize sand and other substrates such as glass and polystyrene. Four mutants tested were affected in attachment to and persistence on all three substrates, indicating the importance of the targeted genes in general surface colonization. Our results provide insight into surface colonization by F. johnsoniae and form the basis for further study of F. johnsoniae survival on surfaces in the rhizosphere.

9Ellen MorganImplementing degron technology to study the role of histone demethylases in pluripotency maintenance
Members of the KDM3 histone demethylase family, KDM3A and KDM3B, play an essential role in maintaining embryonic stem cell (ESC) pluripotency. They appear functionally redundant in this role, as only simultaneous depletion of both factors results in loss of self-renewal and cell viability in ESCs. Despite their necessity, the specific mechanism by which KDM3B and KDM3A help maintain pluripotency remains uncharacterized.
Preliminary data from our lab showed a potential interaction of KDM3A and KDM3B with a variety of mRNA processing machinery, prompting us to determine if KDM3A and KDM3B play a functional role in mRNA splicing. In order to do so, we implemented the dTag13 system, a type of degron technology that permits rapid chemical degradation of any protein of interest (POI) using CRISPR/Cas9-mediated gene editing. With this technology we will be able to more precisely dissect the functional role of KDM3A and KDM3B in mRNA processing. On a broader scale, we can identify the events that lead to the loss of cell viability that occurs when cells are depleted of both KDM3A and KDM3B. Doing so will help identify the role of KDM3A and KDM3B in pluripotency maintenance and give us better insight into how epigenetic and post-transcriptional regulation of gene expression helps define cell identity and direct development.

10Katie MuellerSingle Cell Profiling to Define Biomarkers of Photoreceptor Dysfunction after Gene Editing within human PSC-Derived Organoids
Screening for adverse events is essential for the development of safe genome editing therapies. Here we propose to develop a generalizable and scalable approach to define biomarkers for adverse events after delivery of a genome editor.
Our strategy combines state-of-the-art, label-free optical metabolic imaging (OMI) to measure the physiological, functional, and high-content morphological status of photoreceptors, with single cell transcriptomic profiling (scRNA-seq) and regulatory network-based methods to analyze single cell data. The inferred gene regulatory networks can be used to identify a small (~50) set of biomarkers for adverse events within functional cells. Proof-of-concept studies will focus on the retina, specifically on rod and cone photoreceptors (PR) within 3D optic vesicle (OV) organoids derived from human pluripotent stem cells (PSCs). Creation of this dataset and validation of this approach will leverage these bioengineering technologies toward the development of safer genome editing therapeutics. By tackling a 3D, heterogeneous organoid culture, our approach will extend to more complex cultures. Thus, the impact of this work could be broad, with the potential to advance the development of genome editors administered to any tissue.
11Kushin MukherjeeMapping a low - dimensional space of color - concept associations
People systematically associate colors and concepts, a phenomenon that can either help or hinder the interpretation of color in information visualizations. For instance, by applying optimization algorithms on color concept association ratings, one can create palettes that are easily interpretable without legends (Schloss et al., 2018). Yet such optimization requires the designer to quantify associations between each concept and a large range of colors, to avoid the conflicts that arise when multiple concepts evoke the same strongest associates.
Collecting association ratings for all possible colors and concepts is prohibitively costly and time consuming. We therefore considered whether the space of color concept associations can be expressed using low-dimensional representations. If so, that would mitigate the need for exhaustive human ratings and enable extrapolation of a limited set of ratings to new concepts and colors. To test this possibility, we collected color concept association ratings for 30 concepts (spanning diverse concrete and abstract conceptual domains) and 58 colors (sampled uniformly over CIELAB space). Using principal components analysis (PCA) on the mean color-concept association ratings, we determined 8 ‘color profiles’ that strongly captured the structure of color-concept associations (90% variance explained). From these profiles, we fit regression models to estimate association ratings between new, unobserved concepts and colors. These models predicted both how sensitive each profile was to hue, lightness, and chroma and what blend of the 8 profiles best captured the color associations for a given concept. Using a leave-one-out approach and querying a subset of colors, we strongly predicted ratings for held-out concepts (mean correlation of 0.82 between true and predicted ratings). Our method can be used to automatically generate easily-interpretable color palettes for visual communication. Moreover, our results indicate that the mental representations underlying color-concept associations are highly structured, opening the way for a more principled understanding of color semantics.
12Zhen PengSeed-dependent Autocatalytic Cycles organize complex reaction networks underlying abiogenesis
Life is the canonical example of a complex system, consisting of many and diverse chemical components organized in specific manners, and some of these components are too large and complex to emerge spontaneously from abiotic materials. In contrast, the abiotic environment, which life feeds on and originated from, is much simpler and less organized. Such a gap between biotic and abiotic worlds, and the lack of direct observation of abiogenesis, make origin of life one of the hardest scientific questions.
However, if some abiotic systems do share features similar enough to biotic systems but different enough from other abiotic systems, it is highly likely that the mechanism underlying such features also governs systems on the path from non-life to life. With this rationale, we perform comparative analyses on realistic abiotic and biotic reaction networks with the focus on autocatalysis, which underlies propagation – a key feature of life. We propose the concept of and provide an algorithm to detect Seed-dependent Autocatalytic Cycle (SAC), which feeds on food chemicals to propagate but cannot emerge without being seeded by some non-food chemicals. We found that serial stochastic activation of SACs can result in incrementally complexifying autocatalytic networks, and that multiple seemingly different life-like features, including stochastic seeding, seed-induced autocatalysis, trophic hierarchy, interdependence, scaffolding, and adaptation, are all natural outcomes of the reaction networks organized by SACs. In addition, the SAC theory provides a candidate roadmap from a simple abiotic environment to life-like prebiotic systems and finally to primitive forms of life, by a prebiotic evolutionary mechanism not necessarily requiring linear polymers storing genetic information. This also inspires new insights into designing origin-of-life experiments aimed at detecting SACs in vitro.
13Alex PlumSpatial Structure in Autocatalytic Chemical Ecosystems at the Origins of Life
At the origin of life, autocatalytic cycles may have formed chemical ecosystems and evolved, providing the scaffolding for later stages in the emergence of life. How exactly evolvability can emerge in out-of-equilibrium chemical processes remains an open question. Autocatalytic cycles, whose constituent chemicals collectively catalyze their own continuous recreation, may exhibit limited evolvability when situated in the right sort of spatial structure.
Mathematical models of the dynamics of chemical reaction networks situated in well-mixed reactors, continuously diluted, and driven out of equilibrium by a constant flux of food chemicals demonstrate that distinct autocatalytic processes within those chemical reaction networks can act analogous to distinct species in biological ecosystems. This past modeling relied on mass-action kinetics so that concentrations were continuous, events were deterministic, and all spatial structure was abstracted away. As a result, ecological outcomes depended solely on initial conditions, dilution rates, and reaction rate constants. Here, we introduce models that incorporate spatial structure to explore the effects of adsorption, desorption, diffusion, and compartmentation on chemical ecology and its capacity to evolve. We stochastically simulate artificial, out-of-equilibrium chemical ecosystems in reaction-diffusion systems, on mineral surfaces, and in nested compartments. In each case, we demonstrate that ecological outcomes can differ from the well-mixed case because of differences in how autocatalytic cycles interact with the spatial structure. This can allow a greater diversity of autocatalytic cycles to stably coexist and widens the space of chemical traits that might be selected for by a prebiotic environment.
14Yiming (Amy) Qin

Deficiency in Sirtuin3 rewires mitochondrial metabolism and increases maximum lifespan under caloric restriction
CR is a widely studied regimen in mice that robustly extends health span and lifespan. Previous work revealed that CR stimulates Sirtuin 3 (SIRT3) expression, and through deacetylation of mitochondrial enzymes enhances metabolic flux of pathways often dysregulated in aging and age-related disorders. Here, we set out to investigate the importance of SIRT3 in CR-mediated longevity.
We report for the first time that SIRT3 is dispensable for CR-dependent longevity, and unexpectedly, find that SIRT3 ablation further extends lifespan under CR compared to WT mice. We revealed that loss of SIRT3 resulted in attenuated succinate-dependent respiration but preserved palmitate-dependent respiration in heart and liver of aged CR-treated mice. We further report that KOCR mice were less able to maintain carbohydrate-derived energy production and displayed faster fuel switch to FAO during post-absorption period, leading to a longer fasting state relative to WTCR animals. Despite longer lifespan, altered fuel utilization in KOCR is associated with reduced exercise capacity and spontaneous activity. These results provide an example of the uncoupling of lifespan and aerobic fitness, and new insights into the role of SIRT3 in CR adaptation, fuel utilization, and aging.
15Demitri ShotwellTurng Group - Polymer Processing
Through smart manufacturing techniques, the processing of difficult and unique materials, novel tooling, and novel processing methods we determine ways to improve plastic parts and plastic manufacturing. This work can be scaled up for industrial applications and offer insights in the theoretical understanding of polymers and rheology. Our work aims to improve our understanding of polymers and directly applies to every-day products such as single use plastic silverware and to high-performance products such as optical components in cell phones. In improving polymer processing we can increase part quality, reduce cost, and reduce waste in manufacturing.
16Akhilesh SoniInteger Programming for High Rank Matrix Completion
Finding a low-rank representation of high-dimensional data is a fundamental problem in data science. In the High-Rank Matrix Completion (HRMC) problem, we are given a collection of n data points where each of the data points is observed only on a subset of its coordinates, and the points are assumed to lie in the union of a small number of low-dimensional subspaces. The goal of HRMC is to recover the missing elements of the data points under these assumptions.
State-of-the-art algorithms for HRMC, such as k-GROUSE, perform well if there is modest amount of missing data, but typically fail on instances with large amounts of missing data. We propose a novel mixed integer linear programming formulation for HRMC. The formulation is based on building a set of candidate subspaces and assigning points to selected subspaces. The structure is identical to the classical facility-location problem, with subspaces playing the role of facilities and data points that of customers. We propose a Benders’ decomposition approach for solving the linear programming relaxation of the formulation combined with a column-generation approach for identifying candidate subspaces. The pricing problem to identify a basis matrix is a highly-nonconvex unconstrained nonlinear program that we solve approximately via a gradient-descent method. The integer programming formulation, solved over the candidate subspaces identified by pricing, yields high-quality solutions to the problem. We perform a significant empirical study showing our approach is a significant improvement over state-of-the-art methods for the HRMC problem, especially in the high-missing-data regime.
17Alana StempienMicropattern Platform Promotes Extracellular Matrix Remodeling by Human PSC-derived Cardiac Fibroblasts and Enhances Contractility of Co-cultured Cardiomyocytes
In native heart tissue, cardiac fibroblasts provide the structural framework of extracellular matrix (ECM) while also influencing the electrical and mechanical properties of cardiomyocytes. Recent advances in the field of stem cell differentiation have led to the availability of human pluripotent stem cell-derived cardiac fibroblasts (iPSC-CFs) in addition to cardiomyocytes (iPSC-CMs).
Here we use a novel 2D in vitro micropatterned platform that provides control over ECM geometry and substrate stiffness. When cultured alone on soft micropatterned substrates, iPSC-CFs are confined to the micropatterned features and remodel the ECM into anisotropic fibers. Similar remodeling and ECM production occurs when cultured with iPSC-CMs in a co-culture model. In addition to modifications in the ECM, our results show that iPSC-CFs influence iPSC-CM function with accelerated Ca2+ transient rise-up time and greater contractile strains in the co-culture conditions compared to when iPSC-CMs are cultured alone. These combined observations highlight the important role cardiac fibroblasts play in vivo and the need for co-culture models like the one presented here to provide more representative in vitro cardiac constructs.
18Sarah Stevens
Clare Michaud
Data Science Hub
The Data Science Hub in the Wisconsin Institute for Discovery collaborates closely with the Data Science Institute to provide data science training and implementation across campus. The Data Science Hub executes a mission for community engagement and learning opportunities for campus researchers through a variety of services, including:
- consultations with Data Science Facilitators who can recommend learning pathways and project strategies, and liaise contacts with collaborators and data science experts

– community events and co-sponsored seminars that bring together researchers and other partners around relevant data science topics
– regular trainings (including Carpentries workshops) around fundamental data science and computational skills
19Ross Tredinnick3D Scanning for Automobile Insurance Damage Assessment
In this project, we are exploring the use of commercially available mobile devices, such as the iPad Pro 2020, and iPhone 12 Pro, both newly containing embedded LiDAR (Light Detection and Ranging) sensor technology, to 3D scan automobiles for insurance documentation and assessment purposes. 3D Scanning is the process of rapidly creating a 3D model of a physical, real-world object.
Common prior methods have included using terrestrial LiDAR scanners, which are often expensive, or photogrammetry, which requires challenging processing workflows and ideal real world lighting conditions. Current insurance documentation techniques rely on 2D photographs, some of which are close-ups that can lose the greater context of what is being photographed. This projects explores the creation of new methods to provide additional information and a greater context for insurance company documentation processes for damage assessment.

20Chenglong YuStudies on Small-Diameter Vascular Grafts
Deaths caused by cardiovascular diseases have been increasing over the globe. Vascular transplantation is the most effective means of treating such diseases. However, the availability of healthy and mechanically robust tissue sources is limited, which has increased the necessity of artificial vascular graft development. Currently, many commercial synthetic materials such as Dacron and extended-polytetrafluoroethylene (ePTFE) have been successfully used for large-diameter (inner diameter>6mm) artificial blood vessels. While none of these materials has been proven suitable for the fabrication of small-diameter (inner diameter<6mm) vascular grafts due to thrombus formation and intimal hyperplasia. Endothelium consisting of a continuous monolayer of endothelial cells (ECs) is the innermost layer of native blood vessels and involved in multifaceted blood vessel activities, such as inflammation, fibrinolysis, hemostasis, and extracellular matrix (ECM) production owing to direct contact with blood. Vascular basement membrane (VBM) is a thin layer of fibrous extracellular matrix linking endothelium, and it plays a crucial role in anchoring down the endothelium to its loose connective tissue underneath.
21Yunyi Shen
Bayesian Conditional Auto-Regressive LASSO Models to Learn Sparse Networks with Predictors
Inferring a graphical structure with nodes for multiple responses and predictors is a fundamental statistical problem with broad applications from microbiome, ecology to genetics. While a multiresponse linear regression model seems like a straight-forward solution, we argue that treating it as a graphical model is flawed and caution should be taken because the regression coefficient matrix does not represent the adjacency matrix between response and predictor nodes that encodes conditional dependence.
This observation is especially important in biological settings when we have prior knowledge on the edges. Here, we propose an alternative model to the multiresponse linear regression whose solution yields a graph with edges that indeed represent conditional dependence. The solution to our model is sparse via Bayesian LASSO and is also guaranteed to be the sparse solution to Conditional Auto Regressive (CAR) model. In addition, we propose an adaptive extension so that different shrinkage can be applied to different edges to incorporate edge-specific prior knowledge. Our model is computationally inexpensive through an efficient Gibbs sampling algorithm and can account for binary, counting and compositional responses via appropriate hierarchical structure. Finally, we apply our model to a human gut and a soil microbial composition dataset.
22Yue Xie Projected Newton Methods for Bound-Constrained Optimization
We propose two methods based on the projected Newton methods for solving bound-constrained optimization problems. The first method is a scaled variant of Bertsekas's two-metric projection method. The second is a projected Newton-Conjugate Gradient (CG) method. These algorithms are designed to possess both practicality and worst-case complexity guarantees matching the best known in literature. We illustrate the competitive performance of the second method against other solvers on nonnegative matrix factorization (NMF).
23Sarah MillerTINY EARTH: A network of students crowdsourcing antibiotic discovery from soil bacteria
Tiny Earth is an international network of students crowdsourcing antibiotic discovery from soil with the goal of addressing a public health crisis, antibiotic resistance. Visit our poster to learn how we pivoted the wet-lab course to online due to COVID-19, how we incorporated AJEDI (antiracism, justice, equity, diversity, and inclusion) principles into the curriculum, what new outreach materials we developed with PBS Wisconsin, what new discoveries we found in the lab, and more.
24Travis TangenExpand your Impact: Collaborate with the Meet the Lab team
Meet the Lab is a collection of educational resources for middle school science classrooms, and is a collaboration between PBS Wisconsin Education, Wisconsin Institute for Discovery, Wisconsin Alumni Research Foundation, and the Morgridge Institute for Research.

This collection introduces students to relevant real-world issues, cutting edge research, and the human element—the people working together to research, innovate, and solve problems using science.

This collection supports national and state science standards, and creates opportunities for learners to develop their STEAM identity (“Who do I think I am, who can I be, where do I belong?” in context of Science, Technology, Engineering, Arts and Math). Meet the Lab collections integrate into the Discovery Building’s public engagement with science programs, particularly our field trip and summer camp programs.

WID labs involved in the project to date:
Ashton Lab – Nervous System Engineers
Handelsman Lab – Antibiotic Hunters
Schloss Lab – Visual Reasoning Lab – production in progress
Solis-Lemus Lab – production in progress
25Nolan LendvedWIDkipedia
Learn about a new web resource from WID to help you get what you need to be successful at WID and UW: widkipedia.wid.wisc.edu