2022 Graduate Student Symposium

Help us judge student presentations!

Please fill out this form to help us score the presenters in Student Presentation Session I/II.

Marchand Research Group – Xenobiology lab: Expanding nucleic acid technology with xenonucleic acids

Hinako Kawabe

Nucleic acids are the building blocks of life as we know it. With just four bases, DNA encodes for the 20 standard amino acids and performs complex biochemistry. Xenonucleic acids (XNAs) are nucleic acid analogues that expand the biochemical space we have access to. While multiple XNAs have been previously developed, the Marchand Lab works with XNAs that are structurally similar to the standard bases; our goal is robust incorporation of these XNAs into synthetic biology. This poster will outline the tools we must develop to utilize XNAs in synthetic biology, including sequencing methods and engineered organisms. We also highlight some future applications and additional directions in XNA work, including enzymatic synthesis of XNAs and genetic code expansion.

Room-temperature Synthesis of Metals Sulfides for Battery Applications

Kevin Lee

Open-Hardware Materials Acceleration Platforms for Accessible and Democratized Materials Discovery

Maria Politi

We present two high-throughput workflows implementing low-cost and open-hardware based materials acceleration platforms (< $15k), using commercially available materials, and adopting open-science principles for the exploration of soft matter systems involving large chemical design spaces. The first platform addresses the synthesis of deep eutectic solvents (DESs), a novel class of designer solvents with appealing properties for various applications. A wide range of candidate organic precursors (>105) could be used to form DESs with variable properties, leading to a vast design space, which remains largely unexplored. To tackle such a challenge, we have demonstrated the development and use of high-throughput DES formulation workflows and data-driven design strategies for automated synthesis, rapid evaluation, and screening of physical and electrochemical properties. We further synthesized and characterized 600 DES candidate combinations to evaluate their potential use as redox flow battery electrolytes. The second workflow investigated the sonochemical synthesis of CdSe quantum dots and magic-sized clusters at 625 unique sample conditions, including variations in precursor ratios and ligand concentrations. We implemented a configurable and open-source automation platform for sonochemical applications, based on a repurposed 3D-printing platform (Jubilee) with tool-changing capabilities, in combination with an Opentrons OT-2 liquid handling robot and microplate readers for a fully automated workflow, from sample synthesis to characterization. The experimental campaign took less than 2 weeks, making it significantly faster than the traditional, low-throughput hot-injection method. We subsequently use a data-driven approach to corroborate the qualitative relationships observed from the samples’ optical characterization. These accessible and affordable systems would enable most laboratories to adopt them with limited funding. Moreover, reduced volumes that are as low as 0.5 mL per sample, allow for cost-efficient and environmentally friendly experimental procedures. Finally, the flexibility, modular nature, and reconfigurability of these high-throughput platforms, makes them adaptable to a variety of soft matter formulation studies

Mittal Research Group – Comprehensive Cancer Management Through Imaging, Genomics, and Artificial Intelligence

Rachel Ware

The Mittal group uses bioimaging and machine learning methods to gain new insights in the molecular and spatial disease profiles. We also build digital models on existing medical images for more comprehensive diagnosis and disease management. The current projects focus on digital identification of immune cells on breast cancer tissue, histological and diagnostic modeling of kidney tissue, and spatial profiling of molecular markers in cancer. In addition to data science, we are developing multiplexed spectral imaging tools, to identify molecular targets on cancer samples, enabling multi-dimensional spatial profiling that will not only guide precision diagnosis but also open new opportunities for treatment especially immunotherapy.

Carothers/Zalatan Research Group – Advancing CRISPR-Cas transcriptional programs in bacteria

Ava, Karanjia

CRISPR-Cas transcriptional tools have been widely applied for programmable regulation of complex biological networks. In comparison to eukaryotic systems, bacterial CRISPR activation (CRISPRa) has stringent target site requirements for effective gene activation. We have bypassed some of these limitations through PAM-flexible dCas9 variants, multiple transcriptional activator configurations, and promoter engineering. Our work provides a framework to choose CRISPRa machinery components for a given set of gene targets, which will expand the utility of CRISPRa/i gene regulation in bacterial systems. We will also be highlighting future directions and applications of our transcriptional tools including: single cell RNA sequencing, CRISPR-Cas translational programs, biocontainment strategies, and bacterial therapeutics.

Xenomorph: nanopore sequencing of xenonucleic acids

Hinako Kawabe

At the molecular level, all life on earth is governed by simple building blocks including the four nucleic acids in DNA. Xenonucleic acids (XNAs) are DNA base analogues that expand the biochemical space that is biologically accessible. One of the major barriers to XNA work in synthetic biology is the lack of commercially available sequencing. Nanopore sequencing is a next-generation sequencing method that is based on current measurements; in this work, we aim to parametrize a model to sequence four different XNAs using nanopore data. First, we sequence a known library of XNA containing sequences and use these measurements to build our model. We design a package called Xenomorph, which processes the raw nanopore data and calculates the outlier-robust maximum log-likelihood to output the most likely base at a given position. While different sequence contexts give variable accuracy, we find that a consensus read gives over 95% accuracy for all four of our XNA bases. Using Xenomorph, we hope to explore many future directions in XNA sequencing, including additional model building and investigation of XNA retention in vivo.

Metal Sulfide Nanomaterials as Templates for Heterostructure Formation and Aggregative Growth

Zach Wylie

Diagnostic Tool to Estimate Half-Cell Properties from Fuel-Cell Battery Experiments

Rose Lee

Development of a Python-based tool to non-destructively disaggregate individual electrode contributions to the full-cell response to advance next-gen batteries.

The effect of curing schedules on fiber-matrix adhesion in carbon fiber–epoxy resin composites

Luke Elkhoury

Fiber-reinforced polymeric composites are used in a large and growing number of applications, all requiring different property sets including the nature of the fiber-matrix adhesion to which the present work is addressed. Specifically, the number of curing cycles, curing temperature and schedule, degree of cure, use of accelerants, annealing, and the use of fiber handling agents are investigated for systems of Hexcel IM7 carbon fibers embedded in Epon862 (resin) and Epikure Curing Agent W (hardener) using the single-fiber fragmentation method. The fractional extent of cure is monitored using differential scanning calorimetry (DSC), so that comparisons are made at the same degree of cure (99%). Single-stage curing at the highest temperature produces the highest apparent adhesion, and the use of accelerants significantly increases the curing rate while maintaining the same level of adhesion. Accelerants in some cases, however, decrease the plastic yield strength of the specimens. Annealing reduces induced residual stress and apparent adhesion, but not below the baseline achieved at lower curing temperatures. Plastic yield strength and apparent adhesion decrease for any degree of cure lower than 95%, while the use of handling agents shows no effect on adhesion.

Data science integration with kidney images: a step towards accurate and comprehensive disease diagnosis.

Rachel Ware

The pathology community is rapidly adopting the process of whole slide imaging, where professionals scan and digitize biopsy tissue slides for viewing equivalent to that of traditional light microscopy. These slides are manually reviewed by pathologists to determine a patient diagnosis and subsequent treatment planning. The digitization of tissue slides provides a pathway towards computer-aided diagnostic tools that may reduce intra- and inter-reader variability, decrease manual labor, and provide new insights in the disease profile. The Mittal Research Lab leverages machine learning techniques and bioimaging to build diagnostic aids and improve patient outcomes. In this project, we collaborate with renal pathologists at the University of Illinois at Chicago to develop a digital pipeline to recognize significant diagnostic and prognostic features of stained kidney tissue. We aim to identify three structures associated with function (glomeruli, tubular cytoplasm, and tubular basement membrane) and three histological signatures of disease (fibrosis, edema, and inflammation). Our current convolutional neural network model identifies glomeruli, tubular cytoplasm, fibrosis, and inflammation with a 93% external validation accuracy on tissue sections extracted from marked areas of these categories. This accuracy drops when using the entire whole slide images with unmarked areas due to variability within the different classes. Additionally, predictions are affected by stain variation from sample to sample. In this talk, I will describe the development of our digital pipeline to achieve the current accuracy, bridging the gaps between data science and medicine, and steps to improve model shortcomings for generalizability across samples/labs.