Samples of Student Work

  Samples of Student Work : Student abstracts reproduced with permission.

Alan Chang, Class of 2019, Regeneron STS Top Scholar, Published in Cell Systems

"Utilizing a novel machine learning pipeline for single-cell transcriptomic characterization of a remodeled tumor microenvironment"

Evading immunosurveillance is a key hallmark of tumor progression. From suppressing immune activity to exploiting intrinsic cell mechanisms, cancer cells utilize diverse genetic perturbations to resist current methods of immunotherapy. Exploring the effects of these tumorigenesis drivers can drastically increase treatment effectiveness. Prior to this study, convergent CRISPR screens identified that mutating the gene Prkar1a facilitates tumor growth within a fully functioning immune system. However, the mechanisms by which Prkar1a-mutant cells escape immune surveillance remain unexplored. In this study, various machine learning approaches are employed via R and Bash shell scripting to analyze single-cell RNA sequencing (scRNA) data from Prkar1a-mutant tumors. This study introduces a novel in silico pipeline that separates, identifies, and characterizes remodeled cell clusters using a scRNA-seq dataset. The results provide convincing evidence of not only the immune cell type identities within the Prkar1a-mutant tumor microenvironment, but also the pathway activity of all cell populations. Alan identified immunosuppressive signatures post-Prkar1a mutation, including an anti-inflammatory M2 macrophage population, a minuscule T-cell population, and several exploited pathways. The novel characterization pipeline described in this study enables future research to define the functional roles of tumorigenesis drivers within the tumor microenvironment and thereby enhance the effectiveness of cancer immunotherapy.

Mentor: Dr. Sidi Chen, Yale University, New Haven, CT

Ethan Jacobs, Class of 2019, Regeneron STS Top Scholar, Google Science Fair Global Finalist 

"Optimizing and applying environmental DNA (eDNA) detection methods to analyze the presence of river otters in the Northeast"

Analyzing populations of species has come a long way from simple field research, with recent detection methods being based on DNA detection assays. Collection and analysis of environmental DNA (eDNA), the DNA contained in urine, scales, skin, hair, or other excretes, can now be analyzed to monitor biodiversity and map population distribution with finite precision. This research aimed to optimize eDNA analysis for the detection of undermapped, elusive, non-marine mammal species in the Northeast, specifically North American river otters, beavers, muskrats, and raccoons. Target eDNA was amplified in water samples with the cutting-edge GoFish nested PCR method. The presence of river otter and other target mammal DNA was detected multiple times in the three tested rivers, most importantly providing support for effective protocol and proper primer design. This study is the first of its kind to use eDNA as a method of environmental mapping of non-aquatic and semi-aquatic mammals in the Northeast, and is the first to use GoFish Nested PCR with non-marine mammal DNA and suburban river water samples. Ethan’s results support the use of eDNA-based research, an accessible cost- and time- efficient system of population monitoring, to aid conservation efforts and hoist field research into the imminent future.

Mentor: Dr. Mark Stoeckle, The Rockefeller University, New York, NY

Brent Perlman, Class of 2019, Regeneron STS National Finalist, 7th in nation, International Science and Engineering Fair Finalist, Patent Pending

"Human photosynthesis: Functional chloroplast sequestration in human mesenchymal stem cells"

Photosynthesis is vital to the survival of life on Earth, providing organisms with the ability to harness energy from sunlight and evolve oxygen. Although of the utmost importance, photosynthesis is a process that mammals—specifically humans—are unable to naturally conduct. However, recognizing the previous success of an endosymbiotic, chloroplast-based approach to inducing photosynthesis in murine fibroblasts (Nass, 1969), this study documents the novel ability of human mesenchymal stem cells (hMSCs) to endosymbiotically incorporate and sequester isolated spinach chloroplasts in coculture. Chloroplast-hMSC symbionts retained the de novo ability to conduct human photosynthesis over an 11-day culture period. Chloroplast sequestration had no apparent negative effects on hMSC metabolic activity or cellular viability at the end of the culture period. Hitherto unreported in the literature, this patent-pending process, which allows for the sterile culture of chloroplasts with continued viability, provides future chloroplast therapies with great clinical promise. With potential applications in post-ischemia interventions, engineering of full tissues and organs in vitro, targeted cancer treatments, and even in vivo production and delivery of biopharmaceuticals, sterile chloroplast culture and the phenomenon of human photosynthesis may revolutionize the path of future scientific advancement.

Mentors: Dr. Glenn Gaudette and Dr. Joshua Gershlak, Worcester Polytechnic Institute, Worcester, MA

Alexandra Brocato, Class of 2018, Siemens Semi-finalist, Regeneron STS Top Scholar
“Illuminating non-neuromuscular phenotypes and their temporal trajectory in Spinal Muscular Atrophy (SMA) using electronic health records”

Spinal muscular atrophy (SMA) is a recessive genetic disease targeting motor neurons (MNs) in the spinal cord that results from decreased survival of motor neuron (SMN) protein. SMA is the leading genetic cause of infant mortality, affecting approximately 1 in 10,000 babies in the U.S. Promising therapeutics increase production of SMN protein, preserving MN levels and delaying neuromuscular phenotypes. However, to be effective, therapies should be administered pre-symptomatically, before significant cellular death. Unfortunately, if given too early, potential side effects could be more harmful to less severe, later onset cases of SMA. Without a clear disease trajectory, developing relevant treatments become nearly impossible. Furthermore, since current treatments focus specifically on motor neurons, effective interventions may need to also address phenotypes outside the nervous system. Our study is the first of its kind to characterize SMA disease progression across both neuromuscular and non-neuromuscular phenotypes and to create a timeline of their trajectories. Focusing on 860 SMA individuals, extracted from a population of 45 million registered Aetna Insurance record holders, we conducted all data analysis in R, a programming language for statistical computing. Using this approach, we provide a framework for pre-symptomatic identification of patients who could benefit from preventative medicine. By understanding phenotype trajectories, therapeutics can be developed and administered before major inflection points in SMA without harming individuals with quality years of life remaining before symptom onset.

Mentors: Dr. Lee Rubin and Dr. Scott Lipnick, Harvard University, Cambridge, MA

Alexis Aberman, Class of 2018, Regeneron STS Top Scholar
“A direct comparison of infants’ comprehension of unique versus generic versions of objects”
Infants learn about abstract, generic categories in language through exposure to unique exemplars. In order for humans to use language effectively, they need the ability to make abstractions. There are few studies that examine infants’ ability to make abstractions. This study tested if 15 infants (12-18 months) had the ability to comprehend abstract ideas in language. In a looking-while-listening procedure, infants were directed to look at one of two images. In half of the trials, the two distinct images were of the baby’s own, unique items, while in the other half, the two images were the generic versions of the same objects that the infant had not previously been exposed to. Infants’ success was measured by the corrected proportion of target-looking after receiving instruction to look at the target image. Infants were found to show a greater proportion of target-looking in the unique trials, although there was no significant difference in the infants’ proportion of target-looking between unique and generic trials. Because infants demonstrated a better understanding of their own, unique objects, this study may suggest the existence of a milestone in their ability to make abstractions and must be further confirmed through a larger study. A more precise understanding of the major milestones in language development could aid in both our knowledge and treatment of developmental delays, as well as give us a better understanding of how infants view the world around them.

Mentor: Dr. Elika Bergelson, Duke University, Durham, NC

Jeremy Ma, Class of 2018, Regeneron STS Top Scholar, Neuroscience Research Prize national finalist
“Perceptual interactions in visual depth perception: A quantitative EEG study”
Percepts of different sensory modalities have been shown to interact with one another. Previous studies have qualitatively looked into the results of the interactions between stereo depth and specific pictorial depth cues, but failed to address the interaction themselves. My study will quantitatively investigate perceptual interactions between pictorial (two dimensional) and stereo (three dimensional) depth perception, the combination of which I term combined depth perception. Using a steady-state visually evoked potential (SSVEP) paradigm and a high density EEG net, the neural activity of eight subjects was recorded during the alternation and detection of different types of depth. I proposed and implemented the Relative Peak Strength variable, in order to quantitatively compare and plot the response strengths of each electrode. The perception of pictorial depth was observed to induce neural activity in the ventral stream, while stereo depth was observed to induce activity in the dorsal stream, suggesting these percepts have different functions when perceiving depth. The heat map for combined depth perception was significantly different from the mere sum of pictorial and stereo depth, suggesting combined depth behaves like a Gestalt, in which pictorial depth and stereo depth are not processed parallel to each other. Furthermore, heat maps of combined depth resembled the heat maps of stereo depth scenarios, suggesting stereo depth is the predominant type of depth perceived in combined depth. Pictorial depth was also observed to lower neural activity when comparing different amounts of depth; but has no major contributions when detecting depth. These results suggest that pictorial depth can have different roles on combined depth depending on the task; for example, pictorial depth can play a supplementary role when comparing depth, but have no influence when detecting depth. When pictorial depth is not able to supplement stereo depth, the increased amount of neural activity could be a crucial reason of visual fatigue when viewing stereoscopic displays. This connection between depth percepts and streams should be further investigated, as it can lead to a better understanding of the perceptual mechanisms underlying the reconstruction of the visual world.

Mentor: Dr. Pawan Sinha, Massachusetts Institute of Technology, Cambridge, MA

Alexandra Remnitz, Class of 2018, Neuroscience Research National Finalist and Presenter at the American Academy of Neurology
“Behavioral lateralization and scototaxis unaltered by near future ocean acidification conditions in Poecilia latipinna (Sailfin Molly)”
Rising anthropogenic emissions of CO2 have increased ocean acidity by 25% (NOAA, 2017). In more than 40 studies to date on fish, including species of coral reef and pelagic fish, this increase in CO2 has been shown to alter behavior related to specific sensory systems, such as olfactory, auditory, and visual, in addition to behaviors representing broader cognitive function such as learning, activity, and boldness. The underlying cause behind these behavioral disruptions is hypothesized to be the alteration of ion gradients across the GABAA receptor, the major inhibitory neurotransmitter found throughout the vertebrate nervous system. There are few studies that examine CO2-induced behavioral alterations in estuarine species, specifically fish that regularly experience diel and seasonal CO2 fluctuations in their natural environment. This study examined the effects of predicted near-future CO2 concentrations (~1000 µatm) on behavioral lateralization and scototaxis of the Sailfin Molly (Poecilia latipinna). Behavioral lateralization is the tendency of an individual to favor one side of the body and scototaxis refers to light/dark preference and is a proxy for anxiety measurements. Elevated CO2 levels were not found to affect these behaviors in this understudied species. Findings from this study suggest the Sailfin Molly could be more resilient to high CO2 levels in comparison to other tested species; however, more research is needed to fully assess behavioral tolerance. These novel findings may inspire researchers to further explore the mechanisms leading to species-specific differences in behavioral tolerance, potentially allowing for better assessment of adaptive capacity. Understanding resilience in various species of fish may aid in the maintenance of biodiversity throughout the changing oceanic environment.

Mentors: Dr.Rachael Heuer and Martin Grosell, Rosenstiel School of Marine and Atmospheric Sciences at the University of Miami, Miami,

Audrey Saltzman, Class of 2017, Regeneron STS National Finalist
“Swift XRT and UVOT investigation of low-mass X-ray binary 1RXS J180408.9-342058
The radii of neutron stars are difficult to determine, and there is little consensus as to the correct models for neutron star spectra. Here, the spectra of observations taken by the Swift Xray Telescope (XRT) of the neutron star in the low-mass X-ray binary (LMXB) 1RXS J180408.9-342058 were analyzed, and the neutron star’s radius values were calculated for each observation using the spectral continuum method. The spectra were best modeled using a blackbody to account for thermal emission and a power law to account for inverse Comptonization. The model’s parameter values and the radius are reported for each of the observations providing a longer term look at the star’s behavior. Physically realistic radius values could not be calculated as the color correction factor is unknown; therefore, the color correction factor’s minimum and maximum values were determined using realistic radius values from the literature. A burst rate of one per 4,168 seconds was calculated. Furthermore, this study is one of the first to trace the UV and X-ray thermal evolution simultaneously for a LMXB with a transient, occasionally outbursting, neutron star. In one of the clearest detections to date, the UV flux was determined to be due to reprocessed emission.

Mentor: Dr. Jon Miller, University of Michigan, Ann Arbor, MI

Michelle Morgenthal, Class of 2017, Junior Science and Humanities Symposium National Finalist “Identifying protective and risk factors associated with behavioral misadventure in high sensation seeking adolescents"

This study examines a cohort of vulnerable adolescents, due to their high sensation seeking tendencies. The goal is to characterize the protective and risk factors associated between youth high in sensation seeking and low behavioral misadventure (HSS/LBM) in contrast with those high in sensation seeking and high in behavioral misadventure (HSS/HBM). These cohorts are important because previous research has mostly focused on high versus low risk takers without taking sensation seeking into account. However, this distinct identification of cohorts can help inform prevention methods more effectively by looking at those who would be willing to take risks, but somehow hold themselves back. A secondary data analysis was conducted, drawing on subsample of data derived from wave one of the Adolescent Health Risk Behavior Study (N = 2017) at the University of Michigan, headed by Dr. Daniel Keating. A modified version of the Brief Sensation Seeking Scale (mBSSS) and the Behavioral Misadventure Scale (BMS) were used in this study to identify both groups for comparison with demographic characteristics as well as protective and risk factors associated with adolescent risk behavior. ANCOVA and MANCOVA were used to contrast the HSS/LBM and HSS/HBM groups across these factors. There were four significant findings. First, race was a significant demographic factor (p < .001, Φ = .35), with Caucasians more likely to be in the HSS/HBM group than African Americans or Asians. Second, BMI (p = .035, Φ = .12) was identified as a protective factor against unsafe risk participation; those who were overweight were more likely to be in the HSS/LBM group. Next, although hypothesized that Physical Activity would be a protective factor, it was found to be a significant risk factor (p = .001, r = .20), as those who reported less exercise were more likely to be in the HSS/LBM group. In contradiction with previous research, Supportive Family Context, Pubertal Timing, School Engagement, Peer Influence, Socioeconomic Status, and Religiosity had null effects on the placement of adolescents in the HSS/LBM or HSS/HBM group. We conclude with plausible rationales for our results.

Mentors: Dr. Edward Huntley, Dr. Daniel P. Keating and Dr. Meghan Martz, University of Michigan, Ann Arbor,

Isabelle Chong, Class of 2017, Regeneron STS Top Scholar
“Using an Inertial Navigation System (INS) and a Laser Range Finder (LRF) to create a novel Electronic Navigational Aid (ENA) for the blind”
The current navigation aid of choice for the blind is the white cane, which, although lightweight and easy to acquire, has a limited range and requires extensive training to use. While Electronic Navigational Aids (ENAs) have been developed to improve upon the white cane, the need for certain environmental conditions and preconfigured infrastructure in some approaches (e.g., radio frequency identification and structured light) remains an issue. My objectives were to (1) design the conceptual and mathematical methodology of an ENA for device location and obstacle detection without the use of preconfigured infrastructure, (2) build an ENA by combining a Laser Range Finder (LRF) and an Inertial Navigation System (INS), (3) code a real-time algorithm for obstacle detection and Kalman filtering in C++, and (4) test my ENA’s functionality from both an engineering and human subjects standpoint to obtain quantitative and qualitative feedback. The completed ENA can detect obstacles within a six meter range without preconfigured infrastructure, raising an alarm to the user through sound and haptic feedback if an obstacle has been detected. This device has the potential to provide a robust alternative method of blind navigation in the future.

Mentor: Dr. Yao Wang, New York University School of Engineering

Yasamin Bayley, Class of 2017, Regeneron STS Top Scholar
“Species-specific responses of coccolithophores’ growth rates and calcification to various light intensities: A comparative study of the species Emiliania huxleyi and Coccolithus pelagicus”
Coccolithophores are a type of single-celled marine phytoplankton vital to the biogeochemical cycles of the ocean, specifically the sinking of inorganic carbon. Due to current projections of increasing surface water temperatures, coccolithophores may begin to shift their position upward in the water column, resulting in exposure to higher light intensities. Coccolithophores have been noted to have somewhat contradictory responses to changes in their environment. It is postulated that this difference in responses is due to the species-specific responses coccolithophores seem to have. We seek to determine how two different species of coccolithophores (Emiliania huxleyi and Coccolithus pelagicus) respond to varying light intensities. Specifically, we looked at how these two species’ growth rates and calcification differed at these varying intensities. We used cell counts to determine growth rates and scanning electron microscopy to determine structural changes that high light intensities have caused on each species. We found that Emiliania huxleyi’s growth rate and photosynthesis are largely unaffected by a high light intensity, whereas we present the novel finding that Coccolithus pelagicus produces more calcite per cell at high light, but fewer cells per ml (less calcite per ml than at a low light). We provide further evidence to address the question of why coccolithophores evolved to calcify. We note a correlation between coccolith production and light intensity, which suggests that the production of coccoliths may be triggered by high levels of light. Additionally, this finding suggests that the current increases in surface water temperature may have wider implications in the ecosystem overall, such as a decrease in the downward flux of calcite to the benthic levels.

Mentor: Dr Glen Wheeler, Marine Biological Association of the United Kingdom, Plymouth, UK

Brian Singer, Class of 2016, Neuroscience Research Prize National Finalist and Presenter at the American Academy of Neurology

"To Brux or not to Brux: The Development of Two Novel, Non-Invasive Devices for the Detection of Bruxism"
Bruxism is a disorder in which a patient excessively grinds or clenches their teeth. Symptoms include tooth wear, headaches, back pain, and neck pain. The most common method of treating bruxism is through the use of a mouthguard. The mouthguard does not cure bruxism but only prevents the symptom of tooth wear. Researchers have attempted to reduce bruxism through biofeedback systems. Current bruxism biofeedback devices such as intra-oral pressure sensors and EMG-based systems are intrusive to wear. This study proposes two separate, novel devices that detect bruxism in a less-intrusive manner. The first device is EEG-based and collects data from the F7 electrode located above the left ear. The device uses a machine-learning discriminant-analysis algorithm to detect bruxism from the EEG data. The second device uses Eulerian Video Magnification to amplify temporal color changes in the masseter muscle as seen in a video recording (or live video feed) of bruxism. Both techniques appear to be novel approaches for the detection of bruxism. Both devices were compared to a commercial bruxism detection device to gauge effectiveness and obtain qualitative feedback. Both of the proposed devices demonstrated statistically significant improved efficacy while being less intrusive when compared to the commercially available device.

Adam Ingber, Class of 2014
"Cerebrospinal fluid biomarkers and reserve variables as predictors of future “non-cognitive” outcomes of Alzheimer’s disease"
Alzheimer’s disease (AD) is a devastating neurodegenerative disease that affects a steadily increasing portion of the elderly. It is imperative that early detection and treatment strategies be developed to identify the disease in its early stages and begin potential therapeutic options to halt or prevent conversion from preclinical to symptomatic AD. Using longitudinal data from participants enrolled in studies at the Washington University Knight ADRC, I used linear mixed models to examine the way in which cognitive and brain reserve variables mediate how AD biomarker levels in cognitively normal persons predict future changes in function, weight, mood, and behavior. While education was not shown to have a significant effect on predicting future non-cognitive decline with time, total brain volume exhibited a strong and significant effect when combined with biomarker values to predict decline due to AD over time. My findings suggest that brain reserve plays a stronger role than cognitive reserve in building protection against non-cognitive impairment in AD. This study will contribute to the growing literature on predictive biomarker models of AD and may aid in the future development of individualized risk profiles that can predict future consequences of AD years in advance, with tremendous potential medical and financial implications.

Mentor: Dr. Catherine M. Roe, Washington University School of Medicine, St. Louis, MO

Rachel Cawkwell, Class of 2010, Intel STS National Finalist

"Effect of Tumor Microvesicles on Macrophages in Cancer"
A normal, though understudied, physiological process is the shedding of microvesicles, membranous sacs, from cells. These microvesicles are representative of the cell type they are derived from and can transfer membrane receptors, proteins, mRNA, and organelles in the derived cell. Microvesicles are found in higher numbers in cancer patients and appear to be a vital method of communication for tumors. This project studies how microvesicles work as tumor messengers with regard to macrophages. Macrophages are a type of white blood cell that the tumor uses to direct processes such as blood vessel recruitment and invasion of other tissues. Tumor microvesicles were isolated to see whether they could transfer mRNA to macrophages and increase macrophage proliferation and migration. The tumor microvesicles did transfer mRNA to macrophages as confocal microscopy and qRT-PCR revealed uptake of fluorescence expressing mRNA from microvesicles. A proliferation assay showed no significant change in macrophage proliferation, but an invasive assay demonstrated that tumor microvesicles can increase macrophage migration. This means that tumor microvesicles could potentially recruit macrophages to the primary tumor site, perhaps by transferring mRNA. Further investigation into the details of tumor microvesicle-macrophage communication should confirm and extend these results.

Mentors: Dr. David Lyden and Dr. Hector Peinado Selgas, Weill Medical College of Cornell, New York, NY