The advent of next generation sequencing (NGS) has revolutionized how we study both the genome and epigenome. However, current approaches are limited by the short read lengths associated with NGS, which limits our ability to discern genetic and epigenetic features across nearly 10% of the human genome containing repetitive genomic elements, such as telomeres and centromeres. Recent developments in long-read sequencing methodologies have enabled longer, high-quality reads with significantly more information at the single-molecule level than ever before. Our group has developed a methodology called Fiber-seq that uses an exogenous, non-specific adenine methyltransferase to ‘spray paint’ the genome with m6A methylation marks in an effort to stencil the chromatin architecture of the epigenome onto underlying DNA fibers in a manner that is interpretable by long-read sequencing. Each molecule is then sequenced using PacBio Single Molecule Real-Time (SMRT) sequencing. Methylated adenines are subsequently identified and used to directly resolve the chromatin architecture overlying that molecule of DNA, such as nucleosome footprints and transcription factor binding events. However, calling these methylated adenines on single molecules is computationally intensive, revealing the need for more efficient computational frameworks for base modification detection using PacBio sequencing. To address this, we have developed a statistical framework for identifying m6A-modified bases using raw PacBio sequencing data, leveraging the fact that each molecule is sequenced multiple times during the process. Specifically, we utilized raw data from an unmodified control sample to generate a background methylation model. We then scored data from Fiber-seq test samples against our background distributions to determine the methylation status of individual adenines across single molecules. Our initial model shows high specificity but lacks sensitivity, driving the need for further model development.We aim to generate a Deep Learning framework co-opting the data processing framework built for the statistical model.

The structure and function of a protein is determined by the amino acid sequence that makes up the protein. Understanding how proteins are likely to mutate allows us to predict how organisms will evolve. However, the complex interactions between amino acids in a protein makes it difficult to predict beneficial mutations, and specifically their impact on protein function. I introduced models grounded in statistical physics to learn effective couplings between protein residues from evolutionary data and infer the impact of mutations using covariation of amino acids in evolutionary samples. I then compared the inferred model with machine learning inference of biophysical interactions in proteins that our team has developed to characterize the amino acid preferences within structural micro-environments of proteins. The results of this project will allow us to combine evolutionary data and machine learning inferences to predict beneficial mutations that will occur in a protein. This would have myriad benefits in medicine and evolutionary biology such as being able to predict how bacteria and viruses are likely to mutate in response to treatments.

Underlying the mechanism of many biological processes are biomolecules called proteins. Proteins have dynamic 3D structures, fluctuating between an ensemble of different shapes (conformations), which often have varying physical and chemical properties. Thus, to understand how proteins function, we must understand their conformational dynamics: the kinetics and energetics associated with a protein’s conformational landscape. Here, we develop a novel combination of two spectroscopy techniques for probing protein conformational dynamics: single-molecule detection, and transition-metal-ion-fluorescence-resonance-energy-transfer (tmFRET). tmFRET is the distance-dependent transfer of energy between a donor fluorophore and an acceptor transition metal ion, and this phenomenon has previously been used to detect conformational changes in bulk samples of protein. Extending tmFRET to monitor the conformational state of single protein molecules allows us to detect intermediate states that would be averaged out and thus unobservable in bulk samples, as well as measure the rate of conformational state transitions at equilibrium. To achieve single-molecule tmFRET measurements using a maltose binding protein (MBP) model system, we: label MBP with a metal ion at an engineered metal-binding site and with a fluorophore using cysteine-specific chemistry; specifically anchor MBP molecules to a functionalized glass coverslip; and image single MBP molecules using total-internal-reflection-fluorescence microscopy. Finally, we use quantitative image analysis to recover fluorescence parameters that we can interpret in the context of MBP’s previously well-characterized conformational transitions. Preliminary results culminating in movies of individual fluorescent MBP molecules are presented. Overall, this novel combination of tmFRET with single-molecule fluorescence is a powerful new tool for researchers seeking to probe the conformational dynamics of proteins of biological interest.

Chronic inflammation in the lung often leads to airway remodeling, which can worsen symptoms in inflammatory diseases such as asthma. Airway remodeling is attributed to the excessive deposition of the extracellular matrix (ECM) by myofibroblasts, which are a differentiated form of fibroblasts. Eosinophils, when activated by interleukin-3 (IL-3), release certain soluble factors that were found to be associated with inflammation in asthmatic tissues. Hence, it is crucial to study cellular communication in airway remodeling to facilitate the development of treatments. The aim of this project is to establish an in vitro model of asthma by coculturing primary human lung fibroblasts and eosinophils to study the soluble factors that trigger airway remodeling. We hypothesize that IL-3 activated eosinophils, when immunoglobulin (IgG) is added, release soluble factors that trigger the gene expression and phenotypic changes in fibroblasts. The coculture device has two chambers, in which two types of cells can be cocultured in the shared media while being physically separated by a half wall. Eosinophils are seeded in the outer chamber of the devices and degranulated. The differentiation of fibroblasts would then be quantified by utilizing immunocytochemistry to see the differences in expression levels of alpha smooth muscle actin (ɑSMA) in fibroblasts, in addition to quantitative polymerase chain reaction (qPCR) to detect messenger RNA (mRNA) level associated with inflammation and tissue remodeling. The initial experiments were focused on the monoculture of fibroblasts to ensure that reliable readouts can be obtained from fibroblasts before initiating the coculture. The preliminary data suggest that the fibroblasts treated with transforming growth factor beta 1 (TGF-β1), which promote differentiation, result in significantly higher expression of ɑSMA. Our future experiments include initiating the coculture of eosinophils and fibroblasts to fully illustrate this crucial cellular communication in airway remodeling.

Microglia—the brain’s immune cells—change shape upon external stimuli from either environmental cues or direct injury. Quantifying changes in microglial shape is essential for understanding disease, injury, and their potential as therapeutic targets. Although we can use confocal imaging of immunofluorescent stains to visualize microglia, we lack software to robustly quantify their shapes in a high throughput manner. Therefore, we developed a microglial shape analysis pipeline using Python. In this larger collaborative project, I optimize and incorporate Visually Aided Morpho-Phenotyping Image Recognition (VAMPIRE), a machine-learning-based method that classifies shapes of microglia, into the pipeline. VAMPIRE visualizes the shapes of classified microglia and characterizes their shape heterogeneity using shape metrics. This project aims to demonstrate the robustness and validity of VAMPIRE in the pipeline and when applied to different microglia imaging datasets. To demonstrate VAMPIRE is robust in tissue, I apply VAMPIRE on two datasets: images from an ex vivo organotypic rat brain slice model of oxygen-glucose deprivation and images from an in vivo rat model of hypoxic ischemia. The ex vivo and in vivo datasets provide a representation of different tissue samples—live 300µm thick brain slices and fixed 30µm brain sections—to show how VAMPIRE is robust for classifying microglia generated from different experimental methods. The application of VAMPIRE on tissue models captures changes in microglial shape in response to injury and treatment, allowing comparisons between injury and treatment response. The overall pipeline and VAMPIRE analysis are cell and disease agnostic; therefore, the same methodology can be applied to other models, cell types, and species.

 Microglia, the resident immune cells in the brain, have multiple functions including synaptic pruning to preserve resources, phagocytosis of apoptotic cells, and isolation and removal of foreign material. Depending on local environmental stimuli, microglia can change their shape between multiple states including highly branched, branched, or ameboid. To better understand microglia responses to changes in the brain environment, I investigated morphological shape features that include changes in area, circularity, and aspect ratio among other important features. I specifically focused on the microglial response to oxygen-glucose deprivation (OGD). Oxygen-glucose deprivation is a condition where the brain fails to receive the necessary oxygen and nutrients for growth and maintenance, resulting in higher levels of stress and cytotoxicity. Investigating the effects of OGD on microglia is part of a larger effort - developing a fluorescent imaging pipeline called microFIBER. Our goal for microFIBER is to create an unbiased, detailed, and replicable analysis pipeline for the robust characterization of microglia morphology. Images are from a previous investigation into effects of OGD on neonatal rat brains in the Nance Lab. We used SciKit-Image along with other Python packages to segment, label, and quantify the geometry of fluorescent-labeled microglia cells in the images. SciKit-Image’s module RegionProps was used to quantify shape features by drawing certain properties over the objects and then measuring those drawings. I then analyzed the response of microglia in non-treated, 1.5-hour OGD exposure, and 3-hour OGD exposure via data analysis in Python and Excel. I further divided these treatment groups into regional comparisons of the cortex, hippocampus, and thalamus. Results from statistical analysis supported differences between treatment groups and brain region, including statistically relevant differences in microglial circularity, area, and axes lengths. Differences in shape features could be used in the future as markers for diseased or distressed conditions for medical diagnosis.

The Ashkenazi Jewish(AJ) genome is one of the “easiest” for ancestry testing companies to recognize, but most do a poor job of categorizing geographical AJ ancestry nor provide consumers with an in-depth understanding of Jewish history. The latter is essential for individuals who were unaware of Jewish ancestors likely entering their family tree from assimilation due to persecution. Second, very few companies offer results reporting for the other major Jewish ethnicities-Sephardim and Mizrahim. The central goal of this study was to explore various mathematical and ethical discrepancies harming the consumer experience in ancestry testing. Potential conclusions would speak to larger disparities within the development and marketing of these products. I utilized multiple family members as individual case studies for assessing the computational output between 24 ancestry testing services. The individuals selected were informed on consent, usage of data, and self-reported as either Ashkenazic or Sephardic. After viewing results and analyzing differences in each company’s biobanking and algorithm processes, online forums were consulted for results reporting experiences. Initial findings indicated a significant variation in both ancestry estimation and available information on results across all companies. This was reflected in anecdotal data from online forums, with many consumers confused regarding unknown Jewish ancestry and/or what the Jewish identity entailed. I propose that an overall lack of regulation by governing bodies within the direct-to-consumer genetic testing industry is a large factor as to why this phenomenon occurs. As my hopes were to expose the weak boundaries set by ancestry testing companies by using the Jewish genome as a case study, this project provides a framework for further researching the intersection of bioethics, technology, and identity. I close by proposing possible solutions relating to research, results reporting, UX clarity, and promoting further genetics education.