48440 research outputs found
Sort by
Distributed Computing for Association Analysis of Whole Genome Sequencing Data
No full textIdentifying genetic variants associated with complex diseases in Genome-Wide Association Studies requires scalable computational tools capable of integrating large-scale genomic data. Our key contribution is the implementation of two novel variant-set statistical frameworks within Hail, a cloud-native genomics analytics platform. Our framework enables mapping novel statistical methods to large datasets distributed across compute resources, performing powerful analyses in parallel. This work supports future genetics research by increasing access to advanced statistical tools, with potential for broad adoption after future publication
Engineering of a Modern Storage Bed with Integrated Electrical
No full textThis project followed the engineering design and construction process of a modern storage bed frame. The goal of this bed frame was to balance the minimalist design characteristics of modern architecture without sacrificing functionality or usability. It had to have a robust design that could withstand regular use while still being modular, allowing for easy transportation and installation when moving apartments. It has around 30 cubic feet of under-bed storage drawers, a unique floating rustic design, motion sensor footwell lighting, and integrated reading lights on an oversized headboard with floating end tables. These end tables feature small bedside drawers, a laptop slot, integrated outlets, and wireless charging capabilities. The bedframe was constructed in a manner that allowed sections to be small enough to carry with two people and weigh less than 100 pounds each. Load calculations were used to ensure that structural members were correctly sized and that sufficient fastening methods were employed between different components. Iterative design and prototyping were employed to optimize the efficiency of the hidden wireless chargers and other elements that required fine craftsmanship. Finally, the National Electrical Code and several wood construction standards were examined to ensure a safe design. The following report highlights both the design and engineering that went into meeting these goals, as well as the construction process and the challenges that went into creating a finely crafted, highly functional, and aesthetically pleasing custom bed frame
CS MQP - Congestion Control over a Satellite Network
No full textThe existing algorithms for detecting congestion during the slow start phase of TCP either exit too early, before loss, or too late, after loss, especially on high-latency networks like satellite links. SEARCH is a new algorithm which attempts to improve performance over those links. It attempts to detect the optimal sending rate by detecting when more data is not making it to the receiver by comparing acknowledged bytes over time. We implemented the SEARCH algorithm within a kernel congestion control module for the FreeBSD operating system. We validated our implementation by testing its components incrementally, and evaluated performance compared to the existing slow start algorithms HyStart and HyStart++. We found our implementation exited slow start at a higher goodput without inducing loss more frequently than HyStart and HyStart++
Investigating the Phylogeography and Invasion History of Virile Crayfish (Faxonius virilis) in North America
No full textThe virile crayfish, Faxonius virilis, is widespread in the central U.S. and has spread across the eastern seaboard, threatening biodiversity and native freshwater ecosystems. The invasion history and phylogeography of virile crayfish are understudied. We sequenced the COI and 16S rRNA genes of F. virilis individuals collected from the species’ native and invasive ranges. We compiled unique haplotypes from individual crayfish samples into a phylogenetic tree and mapped the resulting clades to identify the invasion pattern of the crayfish’s invasive populations. Our data shows evidence for multiple distinct invasions, possibly originating from Illinois and Kansas. We identified multiple likely cryptic species, further supporting the cryptic diversity present in this species complex
Future Implications of Using Artemisia species as Anticancer Therapies
No full textResearch shows that Artemisia annua contains a compound, artemisinin, with potential to be used as a cancer therapy. This study tested three Artemisia species’ ability to reduce T-47D viable cell counts, their artemisinin content, and the predicted effects of climate change on their international distribution. The results suggested that A. annua has the greatest potential to be used as cancer therapy, possibly due to its artemisinin content, but growth of these species will likely be impacted by future climate change
Mobile Vans for children's eye exams: International practice and recommendations for setup in Armenia
No full textWe worked with the Armenian EyeCare Project (AECP) in expanding pediatric eye care to underserved areas by creating a design for a mobile eye screening van. By observing eye exams, interviewing professionals, and researching other international mobile healthcare units, we identified necessary design elements for mobile eye care. Using this information, we developed a practical van design to help the AECP deliver consistent and accessible eye care across the country
Responsible Futures — An Inter-University Model
No full textResponsible Futures (RF) is an accreditation program created in the United Kingdom to encourage universities to make investments in education for sustainable development. The objective of the research team was to work with the University of Worcester, UK, and Worcester Polytechnic Institute (WPI) to develop recommendations for both WPI and the Responsible Futures program to move towards a potential RF accreditation at WPI. For this purpose, surveys and interviews of faculty, staff, and students at each institution were conducted. Additionally, an existing data set from the University of Worcester, UK was reviewed. The data from each was used to assess gaps existing in sustainability education, as well as differences between the US and UK that may affect the Responsible Futures program’s success. The research finding indicated that WPI should increase investments in student-facing efforts, embedding sustainability in the curriculum, and whole-university initiatives, while Responsible Futures should continue to develop relationships with US institutions and adapt to the American cultural views on sustainability
Avalanches in the Indian Himalayas: Assessing Risk Perception, Preparedness, and the Role of Machine Learning
No full textAvalanches are hazards that occur in the mountainous regions of Himachal Pradesh. We were tasked by Pr. Gaurav Bhutani at IIT Mandi to explore how local risk perception influences avalanche preparedness and assess the potential use of machine learning to meet community needs in avalanche management. We interviewed community members in Keylong as well as the District Disaster Management Authorities in Kullu and Keylong. We discovered that community members are comfortable relying on their local knowledge to mitigate avalanches, leaving limited opportunities to implement machine-learning-based avalanche management methods
Towards Robust Robot Throwing: A Hybrid Framework of Physics and Learning
No full textRobotic throwing extends a manipulator’s functional workspace by leveraging object flight kinematics. However, accurately reaching distant, cluttered targets is notoriously sensitive to grasp pose, release velocity, aerodynamics, and real-world model error. Traditional projectile equations ignore these complexities, while purely data‑driven policies demand vast experience and often fail out of distribution. This thesis proposes a hybrid framework that fuses analytic physics with learning to achieve fast, precise, and generalizable robotic throwing. The system, implemented on a UR10 arm with an overhead Intel RealSense D455i camera, first predicts stable top‑down grasps given the object's Centre Of Mass from RGB‑D heightmaps. It then computes an analytic release velocity from closed‑form projectile equations and refines that estimate with two complementary learning modules: a \textbf{Residual‑Physics Network} that directly corrects model bias, and a \textbf{Physics‑Informed Neural Network (PINN)} that embeds the governing differential equations into its loss for data‑efficient trajectory prediction. All components are trained end‑to‑end with prioritized experience replay in both photo‑realistic PyBullet simulation and on a 10‑object YCB‑based physical benchmark spanning diverse masses, shapes, and textures. Extensive experiments compare four ablations—Physics-Only, Learning-Only, Residual, and PINN—across 150 training episodes, unseen-object generalisation and real-world transfer. These results demonstrate that coupling light‑weight analytic models with structure‑aware learning combines the interpretability of physics with the adaptability of data, enabling reliable robotic throwing in unstructured environments. Beyond tossing, the framework is readily extensible to other dynamic skills—such as pushing or kicking, where accurate long‑horizon physics reasoning is essential
School of Arts and Sciences Newsletter, Spring 2025
No full textEmail newsletter created by the School of Arts and Science