GradBMES hosts 2022 Bioengineering Graduate Student Research Symposium
The Bioengineering Graduate Research Symposium is an annual event hosted by the graduate Biomedical Engineering Society (GradBMES). During this highly anticipated gathering, bioengineering students and faculty have the opportunity to learn more about different research in the department. "This event shows the younger students that they can do great things here. For the older students, it gives them ideas for who they can collaborate with," said GradBMES president Joseph Tibbs. "Seeing everyone, asking questions, and getting responses in real-time feels much more personal than the online symposiums we've had to host during the pandemic," he said.
Bioengineering Ph.D. students Jongwon Lim, Tarun Mahajan, Fu Li, Hyeon Ryoo, Skye Shepherd and Xiaohui Zhang presented at this event. Lim and Shepherd were also selected for travel awards in recognition of their outstanding presentations and to encourage them to showcase their work at upcoming conferences.
Title: Microfluidic point-of-care device for detection of early strains and B.1.1.7 variant of SARS-CoV-2 virus
PI: Rashid Bashir
Abstract: Current gold standard for detecting the virus and its variants is based on PCR-based diagnostics using complex laboratory protocols and time-consuming steps, such as RNA isolation and purification, and thermal cycling. These steps limit the translation of technology to the point-of-care. While PCR-based assays currently offer the possibility of multiplexed gene detection, reports of isothermal assays at the point-of-care with detection of multiple genes are lacking. Here, we present a microfluidic assay and device to detect and differentiate the Alpha variant (B.1.1.7) from the SARSCoV-2 virus early strains in saliva samples. The detection assay, which is based on isothermal RTLAMP amplification, takes advantage of the S-gene target failure (SGTF) to differentiate the Alpha variant from the SARS-CoV-2 virus early strains using a binary detection system based on spatial separation of the primers specific to the N- and S-genes. We use additively manufactured plastic cartridges in a low-cost optical reader system to successfully detect the SARS-CoV-2 virus from saliva samples (positive amplification is detected with concentration ≥10 copies per μL) within 30 min. The reliability of the developed point-of-care device was confirmed by testing 38 clinical saliva samples, including 20 samples positive for Alpha variant (sensitivity > 90%, specificity = 100%). This study highlights the current relevance of binary-based testing, as the new Omicron variant also exhibits S-gene target failure and could be tested by adapting the approach presented here.
Title: Relative stability of mRNA and protein severely limits inference of gene networks from single-cell mRNA measurements
PI: Roy Dar
Abstract: Inference of gene regulatory networks from single-cell expression data, such as single-cell RNA sequencing, is a popular problem in computational biology. Despite diverse methods spanning information theory, machine learning, and statistics, it is unsolved. This shortcoming can be attributed to measurement errors, lack of perturbation data, or difficulty in causal inference. Yet, it is not known if kinetic properties of gene expression also cause an issue. We show how the relative stability of mRNA and protein hampers inference. Available inference methods perform benchmarking on synthetic data lacking protein species, which is biologically incorrect. We use a simple model of gene expression, incorporating both mRNA and protein, to show that a more stable protein than mRNA can cause loss in correlation between the mRNA of a transcription factor and its target gene. This can also happen when mRNA and protein are on the same timescale. The relative difference in timescales affects true interactions more strongly than false positives, which may not be suppressed. Besides correlation, we find that information-theoretic nonlinear measures are also prone to this problem. Finally, we demonstrate these principles in real single-cell RNA sequencing data for over 1700 yeast genes.
Title: Advanced image reconstruction for accurate and high-resolution breast ultrasound tomography
PI: Mark Anastasio
Abstract: Breast cancer is the most common cancer among women, accounting for 1/3 of cancers diagnosed. Currently, mammography is the state-of-the-art method for detecting and diagnosing breast cancer. However, mammogram does not detect certain subtle breast cancers, especially for people who have denser breast tissue. A new technology, Ultrasound Computed Tomography (USCT), can be useful for breast cancer imaging that utilizes ultrasound and tomographic principles to obtain quantitative acoustic properties estimation of soft tissues. The acoustic properties include speed-of-sound, density, and acoustic attenuation. USCT has a number of advantages over existing imaging modalities. Compared with mammography, it is radiation-free, breast-compression-free, and relatively inexpensive. Compared with conventional B-mode ultrasound, it produces images with a large field-of-view whose quality is independent of the skill of the operator. In this talk, the background of USCT and wave physics of image formation will convey. We will introduce our recent works on numerical breast phantom modeling, the development of advanced image reconstruction algorithms, and accurately modeling of the underlying physics of USCT imaging systems. The results demonstrate the improved image accuracy by the use of our advanced algorithms and numerical modeling and the promise for better clinical use.
Title: PEG Microgels Enable Distinct 3D Cellular Microenvironments
PI: Gregory Underhill
Abstract: Non-alcoholic fatty liver disease (NAFLD) is a growing epidemic projected to become the major cause of liver related morbidity and mortality. If left untreated, it can develop into liver fibrosis and cirrhosis. Liver fibrosis is characterized by a highly heterogeneous microenvironment, which can make its modeling in vitro difficult. Hepatic stellate cells (HSCs) are implicated to be the main effectors of liver fibrogenesis in the diseased liver. We have developed a 3D system to modulate HSC behavior using polyethylene glycol (PEG) microgels of different stiffnesses, functionalized with well-defined protein components of the extracellular matrix (ECM). Multi-arm PEG modified with norbornene is reacted with thiolated ECM proteins, emulsified, and triggered to react with crosslinkers to form spherical microgels of <25 µm. HSC or any cell type, such as iPSCs, can then be cultured with these fluorescently tagged, heterogeneous microgels as neighbors using microwells. When working with already activated HSC, we observed that fibronectin, an ECM protein critical for liver fibrogenesis, showed the most elongated phenotype, while microgels without ECM led HSCs to aggregate and round up. qPCR studies showed the least activated phenotypes in HSCs cultured with the stiff, fibronectin microgels in terms of ACTA2, TGFB1 and PDGFRB. Other adhesive proteins such as collagen I, laminin and n-cadherin also showed lower activation markers compared to microgels without protein. This study serves as a proof of-concept of the use of protein-conjugated microgels of varied stiffnesses as platforms to study in vitro 3D cell behavior.
Title: Target Recycling Amplification Process for Digital Detection of Exosomal MicroRNAs through Photonic Resonator Absorption Microscopy
PI: Brian Cunningham
Abstract: Exosomal microRNAs (miRNAs) have considerable potential as biomarkers for diagnosing cancer development and monitoring the progression of diseases. Therefore, developing an ultrasensitive sensor to effectively quantify exosomal miRNAs that is simple, cost-effective, and requires low sample volumes for frequent testing plays a crucial role in the early diagnosis of cancers and provides insight into the role of miRNA in evaluating therapeutical efficiency and disease progression. Therefore, we designed an efficient target recycling amplification process (TRAP) for ultrasensitive detection of exosomal miRNAs through photonic resonator absorption microscopy and toehold-mediated DNA strand displacement reactions. By utilizing target recycling, we achieved multiplex digital detection of miRNAs with a subattomolar sensitivity in 20 min with broad dynamic range from 0.1 aM to 1 pM. The TRAP shows robust selectivity at a single-base mismatch precision. Extracted exosomal total RNAs derived from human cancer cell lines were also investigated with the TRAP system and validated with qRT-PCR quantification, where our system achieved a detection limit of 1.2. copies/μL. This TRAP approach is well-suited for circulating biomarker quantification, particularly when are miRNA present in low concentrations or in low sample volume, with potential for point-of-care testing in cancer diagnosis or for treatment and progression monitoring.
Title: Identifying functional brain networks from spatial-temporal widefield calcium imaging data via a recurrent autoencoder
PI: Mark Anastasio
Abstract: Exploring functional brain networks (FBNs) from functional neuroimaging data provides unique insights into the functional architecture and the organization of the brain. Wide-field calcium imaging (WFCI) with genetically encoded calcium indicators enables monitoring of cortex-wide neural dynamics in mice of specific cell populations with higher spatial and temporal resolution and signal-to-noise ratios than traditional measured hemodynamics. While seed-based correlation analysis and independent component analysis have been applied to explore the FBNs based on WFCI data, these methods analyze WFCI data on either spatial or temporal domains. However, the brain is a network that varies spatially and temporally. Thus, a method capable of identifying FBNs from spatial-temporal calcium dynamics is desired. Recently, representation learning that employs deep neural networks demonstrated the ability to learn reduced-dimensional latent embeddings that could capture the important spatial-temporal information in neuroimaging data. In this study, a recurrent autoencoder was adapted to learn spatial-temporal latent embeddings from WFCI data. Ordinary least square regression was employed to derive the spatial maps of corresponding FBNs. A template matching procedure was performed to identify the FBNs using seed-based functional connectivity maps as templates. Examples of FBNs identified by the model are presented. Spatial similarities are shared between FBNs estimated from learned embeddings and those derived by seed-based correlation method. The proposed method will facilitate further investigations of FBNs from the spatial-temporal calcium dynamics recorded by the WFCI data.