Vision issues can significantly hinder a child's academic performance and overall quality of life. Despite the availability of existing vision screening and therapy methods, up to 25%-30% of school-aged children struggle with undiagnosed or untreated vision problems, primarily due to financial limitations, logistical issues, and a lack of prioritization of vision care. Untreated pediatric vision problems can lead to social issues and long-term systemic failures. Therefore, it is crucial to provide comprehensive vision care that ensures the diagnosis of all types of visual impairments and appropriate treatments. To address these challenges, computer programs and software applications have been developed as new screening and therapy approaches. This study examines the relationship between pediatric vision and academic performance in school-aged children and evaluates the feasibility of using technological advancements in this medical field. I have conducted a meta-analysis review that involved analyzing and synthesizing data from various studies to explore the link between pediatric vision and academic performance, as well as studies that evaluated the efficacy of using computer programs and software applications as screening and therapy approaches for children's vision issues. Based on the preliminary results, I found that untreated vision problems in children can negatively affect academic performance and overall quality of life. I also discovered that technological advancements have the potential to improve access to vision care and treatment for children, particularly for those who face financial or logistical barriers. I hope that these findings will raise awareness about the importance of early detection and intervention in pediatric vision care, and encourage policymakers and healthcare providers to prioritize this issue. My research also highlights the potential benefits of using technology to improve access to vision care for children and suggests that further studies are needed to explore the efficacy of different technological interventions.

Although social Virtual Reality (VR) has attracted increasing attention as a new way for people to interact, it faces challenges with harassment, a problem other social platforms face as well, online gaming communities in particular. The embodied environment social VR provides also brings new forms of harassment compared to social media, requiring effective responses from social VR platforms. We examined the safety features of four popular social VR games: VRChat, Horizon World, Altspace, and RecRoom to learn the standard safety practices. To understand how social VR communities share and respond to harassment experiences, we collected 134 posts and comments from online communities for these games on Reddit, Twitter, and Oculus Forum. We used inductive coding to identify themes and trends. We found that the four social VR games have common safety features such as Personal Bubble, Block, and Report, but these features differ in name, effect, and ease of access. This can pose an increased learning curve for players and make them less aware of these functionalities. From the online posts, we found the most common harassment experiences include hate, unwanted sexual attention, and embodied sexual harassment. The most common response to harassment experiences is suggesting strategies or resources. However, these responses include a mix of positive (e.g., empathetic, supportive), neutral, and negative (e.g., gaslighting) tones. We also found a difference between the most commonly adopted safety feature and the most recommended feature, where the former is Personal Bubble and the latter is Block. Based on the findings, we provide design implications to improve safety features and build easier-to-access and informed safety systems for social VR games. This research contributes to developing a more inclusive environment for players from diverse backgrounds and identities by identifying opportunities to provide better safety features and improve safety norms in virtual worlds.

Methanotrophs are prokaryotes that naturally consume the potent greenhouse gas methane for energy. Through metabolic engineering at an industrial scale, these microorganisms hold potential to mitigate the contribution of methane emissions to global warming. In particular, Methylotuvimicrobium buryatense can sustain robust growth both in nature and experimental settings; it is a promising engineering candidate. To develop a robust metabolic engineering platform using M. buryatense, biologists require a deeper understanding of the genetic mechanisms by which it functions. Here, I present an open-source software tool designed to interactively explore the transcriptome of M. buryatense. By integrating bulk RNA-seq datasets collected from experiments over the past decade and applying an array of unsupervised machine learning clustering algorithms, we cluster genes by their expression profiles in differing growth conditions. These gene clusters are annotated with gene ontology (GO) terms using statistical enrichment analysis to assist in functional interpretation of the clusters and the genes that comprise them. To enhance domain-expert researchers’ ability to explore and drill-down into specific queries, I unify these cluster-specific analyses in a web-hosted tool using interactive data visualization techniques centered on a ReactJS frontend and Azure Cloud backend. With both exploratory and query-focused use cases, this software tool can support M. buryatense biologist workflows for predicting functions of hypothetical proteins, showcase new or confirming putative regulatory processes, and generate new experimental hypotheses from the presented transcriptomic trends.

Two-photon microscopy enables single-cell resolution recording of neural activity via the expression of proteins that change fluorescence brightness levels based on neural activity. This technology can be used in conjunction with behavioral tests in animal models to investigate the neural mechanisms underlying cognition, sensory processing, and internal states. Here we present findings to improve 2-photon data quality through a denoising algorithm, which removes random non-neural noise from data, and subsequently extract neural-behavioral relationships through deep-learning classification. In this project, I wrote custom python scripts to perform these complex analyses using open-source packages on the UW high performance computing cluster. Two-photon in vivo images of fluorescent indicators can be contaminated by varying levels of noise, related to the recording device or the environment. Such noise is prohibitive for detecting neural structures. Here, I apply a convolutional neural network (CNN)-based denoising algorithm, DeepInterpolation, to mitigate the noise present in neural activity recordings. We hypothesize that denoising will achieve a significantly higher single-pixel signal-to-noise ratio (SNR) compared to the raw data, and enable significantly more neural structures to be detected by segmentation algorithms. Deep learning techniques have shown promising results in improving the classification of video data. Data acquired from two-photon microscopy are sequences of images across time, yet most analyses focus on pixel-averaged time-series extracted from individual neurons. The relational information between space and time that may inform of underlying neural mechanisms is therefore lost in these approaches. Here, we propose the application of Deep 3-dimensional convolutional networks (3D ConvNets) to learn spatiotemporal features of two-photon imaging data and to classify local circuit interactions related to animal behavior.As a whole, the goal of this work is to provide an open-source working example for the classification and feature extraction of two-photon imaging neural activity recordings. This pipeline can be used to gain insight into spatiotemporal dynamics related to event-related behaviors in two-photon imaging datasets.

Our research group is exploring the feasibility of utilizing nanopore sensors for protein sequencing, whose compact size and ability to facilitate extremely long, uninterrupted reads of protein strands upstage the current procedure of using complex, expensive mass spectrometry (MS) devices. My project predicts the sensor’s raw signal data using a carefully tested combination of each amino acid’s volume and charge properties. Using my model to generate predictions for a specific database of proteins, I can compare the unknown raw signal to each of the predicted signals to single out the best matching/correct sequence. While the protein space of De Novo sequencing is vast (20 to the power of protein sequence length), this method effectively shrinks the protein space to a group of substantive, feasible sequences. Employing the current predictive model on a database of synthetic and natural proteins, when compared against an unknown protein’s raw signal, I found that, on average, the correct prediction consistently ranked within the 99th percentile of matches among a predicted test set of >20,000 sequences. Advancing single-protein sequencing can revolutionize protein research by enabling the identification of low-abundance proteins. Additionally, the increased sensitivity of the nanopore sensor could shed light on the so-called "human dark proteome," composed of approximately 3,000 human proteins that have not yet been identified despite genetic evidence of their existence.

DNA methylation is a form of DNA modification that influences gene expression and plays a fundamental role in development and disease. Previous research has shown that DNA modifications, such as DNA methylation, provide a mechanism by which cells encode their response to environmental conditions. However, it is not known whether DNA methylation at an arbitrary genomic location can play a causal role in the regulation of gene expression. To determine the causal relationship between DNA methylation and gene expression, we combine an experimental assay called mSTARR-seq and machine learning, to learn the genomic relationship between DNA methylation, genomic DNA sequence, and gene expression. Specifically, mSTARR-seq is a perturbation-based assay that measures gene expression levels from millions of genomic sequences in both methylated and unmethylated conditions. We build a sequence-to-expression convolutional neural network (CNN), which takes as input genomic DNA as well as methylation state from each experimental region and predicts as output the corresponding gene expression levels. Through this process, we hypothesize that the model can learn which sequence features are sensitive to DNA methylation, shedding light on biological mechanisms underlying the impact of DNA methylation. Our experiments have shown that our model can predict gene expression values on unseen test sequences with reasonable accuracy (Pearson R=0.45). In ongoing work, we are optimizing model architecture and hyper-parameters to better understand the sensitivity and robustness of our modeling choices.