4353 research outputs found
Sort by
Revisiting co-expression-based automated function prediction in yeast with neural networks and updated Gene Ontology annotations
Full text linkAutomated function prediction (AFP) is the process of predicting the function of genes or proteins with machine learning models trained on high-throughput biological data. Deep learning with neural networks has become the dominant machine learning architecture of contemporary AFP models. However, it is unclear what difference exists between neural networks and previous machine learning architectures for AFP. Therefore, we created a model of AFP in yeast using neural networks that is trained on gene co-expression data to predict Gene Ontology (GO) labels. When trained on the same input data, we found that our model outperforms two other experimentally-validated co-expression-based AFP models using other machine learning techniques (Bayesian networks and adaptive query-driven search) when predicting individual genes involved in mitochondrion organization. In particular, we found our neural network model better distinguished mis-annotated negatives in its training data. Finally, we quantified how differences in the gene expression data and Gene Ontology annotations affect the performance of our model across each of its predicted GO terms. Our results suggest that neural networks are more performant and robust to GO mis-annotations compared to other machine learning architectures for co-expression-based AFP of some biological processes
Building Athletic Community: Environmental Graphic Design
No full textTrinity University is nationally recognized for providing top-tier academic and athletic experiences. Under the University’s strategic plan, the use of space on campus is being reevaluated. Specifically, Trinity University Athletics wants to improve its users\u27 environmental experience. After conducting user research with Trinity University student-athletes and athletic alumni, four areas of improvement were highlighted: navigation, outdated signage, brand cohesion, and community representation. To address these issues, seven environmental graphic design tactics and additional recommendations will unify the athletic experience from student to alumni by connecting people to place. These tactics are supported by client research, audience research, user interviews, environmental graphic design research, and a site “communication design” analysis. Due to the interdisciplinary nature of the project, the deliverable is formatted as an Integrated Marketing Communications Plan based on research solutions surrounding the emerging field of Environmental Graphic Design
Ser Nhande’i va’e, da concepção aos primeiros passos: uma abordagem etnográfica sobre a permanência, o movimento e a palavra
No full textO evento do nascimento, assim como os ritos de passagem, são temas recorrentes na literatura etnográfica, com destaque aos resguardos, condutas e preceitos. Neste artigo, ao versar sobre o ato de nascer, refiro-me a alguns princípios referenciados na cosmologia guarani, na qual esse acontecimento é também um marcador – marco espaço/temporal – dos lugares habitados – isto é, ação de territorialização. Determinadas condutas pré e pós-parto são fundamentais para os Guarani Mbya e os Ava Guarani se socializarem com os demais viventes nas matas/terras que conformam seu território (Yvyrupa). A partir desse prisma, ressalto algumas propriedades e combinações de elementos/substâncias de origem animal e vegetal utilizadas na confecção dos adornos infantis, no preparo de alimentos e remédios (poã). Pela condição de vulnerabilidade da criança, esses e outros cuidados a acompanham, desde os presságios de sua concepção até seus primeiros passos – pisar a terra –, início de seu caminhar, momento da celebração da cerimônia do Nhemongarai, em que sua alma-palavra (nhe’ẽ) é revelada
The two hungers: food security, morality, and cash transfer policies for Canela Apanjekra people in Indigenous Central Brazil
No full textThis article examines how the Canela Apanjekra (Porquinhos Indigenous Land, Maranhão, Brazil) engage with conditional cash transfer policies aimed at combating hunger—especially the Bolsa Família Program—and how these policies are interpreted through Indigenous moral economy. Drawing on long-term ethnographic research (2011–2019) and a household survey conducted in 2019, it argues that “hunger” is not a unitary condition measurable solely through individual consumption, but a relational and ecological experience shaped by obligations to share, by shame/respect (paham), and by shifting conditions of subsistence. The analysis develops the notion of “two hungers”: in the village, hunger is commonly articulated as the absence of animal protein amid environmental degradation and fluctuating success in hunting and fishing; in town, hunger takes the form of food unavailability produced by dependence on cash, bureaucratic infrastructures, hostile interethnic relations, and the brokerage of “bosses” who control transport and access to benefits. By tracing how food, money, and goods circulate across village and city, the article shows how a policy designed to eradicate hunger may simultaneously generate new vulnerabilities. The conclusion proposes that food-security indicators and interventions must incorporate native conceptions of hunger and strengthen food sovereignty
Final Project Report: Atmospheric Water Generation
Full text linkThe Atmospheric Water Generation Team designed a system where a dehumidifier can extract moisture from the air and use the water for irrigation. This effort responds to the growing need for water conservation methods in South Texas, where there is a growing need to preserve freshwater sources. The final prototype will be powered using solar panels and a battery, and all water collected by the dehumidifier will be stored in an external tank. An Arduino microcontroller will manage when the system turns off and on under various conditions and times of day. The total budget for this project was $3200.
The design contains four major subsystems: the power supply, the water storage, the housing and the electronics subsystems. The first semester the team focused on identifying objectives for the design, doing the necessary research on how the design could complete the objectives, and doing calculations to ensure all components were compatible with each other. All components and electronics were chosen to be compatible with a HomeLabs Energy Star Dehumidifier, donated by the project sponsor Dr. Chris Victoria. By the end of the first semester, all subsystems were designed and components were ordered. During the second semester, the team focused on completing all the testing for subsystems and adjusting the final design as needed.
After discussing with the team advisor and sponsor, the team decided on how the final prototype would operate. The system would run every other day from 4 A.M. to 7:30 A.M. to allow the solar panels to charge the battery for two days. The system would shut off if the Arduino reads the battery voltage below 50V or if the external tank is full. After coding the operations within the microcontroller, the team focused on testing. All individual subsystem tests were completed successfully and confirmed the prototype meets the primary objectives set at the beginning of the project. However, there were numerous delays in testing due to weather delays, since the electronics could not be protected from the rain without housing. Because of these delays, the team was not able to run the 7-day full process test in its entirety. However, initial data does confirm that the subsystems work together, but the prototype needs more days to run to confirm the solar panels can charge the battery in two days to allow the system to run for 3.5 hours at night.
Looking ahead, several opportunities for future improvement and optimization have been identified. One potential enhancement is to connect the system directly to a home’s existing solar panel infrastructure. Doing so would reduce the need for dedicated solar components within the unit, making the system more cost-effective and efficient for homeowners already using solar energy. Additionally, further work could be done to refine the system’s housing. The housing within the prototype was limited to the equipment available in the Makerspace. Due to the team’s budget and available equipment, dimensions on the doors are slightly off and could allow water to get through the prototype. Overall, the design objectives were complete and the system works as intended. The design is easily scalable too. In the future if someone wants the system to run more frequently or for longer duration, scaling up on the solar panels and battery is all that is needed
Final Design Report: Smart Stethoscope
Full text linkThe Smart Stethoscope project integrates signal processing and artificial intelligence (AI) to improve the diagnostic accuracy of heart abnormalities, especially in remote environments where access to skilled professionals may be limited. Cardiovascular diseases remain a leading cause of mortality worldwide, highlighting a need for innovative solutions in healthcare. This initiative seeks to address that need by developing a machine learning model capable of classifying 42 different heart abnormalities with an accuracy of 90%, using a dataset composed of stethoscope recordings. The proof of concept of this project was carried out during the Fall semester, using a 2D-Convolutional Neural Network (CNN) approach on a dataset of 10 different heart abnormalities. During the Spring semester, the team seeked to simplify and optimize the proof of concept by replacing the 2D-CNN approach for two approaches: 1D-CNN and Wavelet Transforms. As per these approaches, the same dataset was used for their implementation in its full length, consisting of 42 unique heart conditions, each represented by a 10-15 second sample of continuous heartbeats. To better simulate a real-world dataset of heart sounds collected from a wide range of patients, this data was procedurally expanded using MATLAB. This dataset was splitted into 70% training data, 15% validation, and 15% testing data for both implementations.
The first model is a one-dimensional convolutional neural network (CNN) implemented in Python, trained on the full length of each of the 42 audio samples. The second model, implemented with MATLAB, is a feedforward neural network trained on wavelet-transformed embeddings of the same 42 samples. In terms of results, both approaches showcased promising results above the accuracy threshold set at the beginning of the project (90%). The CNN implementation achieved a 99.7% without overfitting with a size of 3 Mb. At the same time, the feedforward implementation yielded an accuracy of 96.14%, using three fully connected layers, and three dropout layers, which helped optimize the model. The results and implementation of this project showed that it was possible to classify many heart sounds, and it has the potential to be expanded for other applications such as ECG analysis. Overall, the work done in this project sets a basis for the creation of a valuable tool for improving heart condition diagnoses in areas with limited resources
Ultracold collisions of a neutral atom with a trapped ion in one dimension
No full textWe present a fully quantum mechanical description of a free Li atom scattering from a trapped Yb+ ion in one dimension. By reformulating the system in polar coordinates and employing the adiabatic representation, we extract a set of coupled adiabatic potentials representing the atom interacting with the ion in different trap states. In an approach similar to quantum defect theory, we leverage the vast difference in energy scale between the interaction, the trap, and the scattering energy to encapsulate the short-range atom-ion scattering behavior in a single phase parameter. The presence of trapped (Yb Li)+ molecular-ion states leads to a series of roughly evenly spaced resonances in the scattering cross section. The predicted distribution of resonances at low collision energies is at odds with the expectation of quantum chaos and the Bohigas-Giannoni-Schmit conjecture
Optimization of Soft-Sphere N-body Simulation through choice of Programming Language and Priority Queue
Full text linkIn this thesis I will describe my efforts to improve the performance of the current Rust softsphere N-body simulation. To do this I undertook two projects; validating the performance of our chosen programming language Rust along with comparing Rust to other languages, and implementing and testing a novel priority queue, the Parallel Bucket Queue, for object collision processing. In order to validate our programming language choice I benchmarked the parallel performance of C, C++, Go, Java, Julia, and Rust for N-body simulations. This benchmarking is based off the O(N2) simulation done each language in the Benchmark Game, with simulations modified to have larger number of objects simulated and be run in parallel. The benchmarking will show all compared languages have as similar performance and does not invalidate the choice to use Rust to implement the n-body simulation. Then I will describe the process of implementing a sequential bucket queue then adding parallel functionality to create a parallel bucket queue and comparing the performance of this novel queue to a sequential and parallel binary heap. The performance of a priority queue is done by having it play back and process collision events logs. Our testing shows that at large particle number counts that the parallel priority queue is almost four times as fast as the next fastest data structure
Using Biotinylated Fluorophores Attached to Streptavidin Adsorbed to Nanopatterned Gold Surfaces to Enhance Fret
No full textSurface plasmons generated on nanopatterned gold substrates can be used to enhance Förster Resonance Energy Transfer (FRET) between two fluorescent molecules by increasing the local density of optical states. Previously in the Steele lab, DNA with fluorescent molecules along its backbone were attached to the surface of these nanopatterned gold substrates in an effort to precisely space the fluorescent molecules and ascertain the specifications of FRET enhancement. However, this work continually gave results inconsistent with current understanding of fluorescence enhancement. It was hypothesized that these inconsistencies might be due to the fluorescent molecules on the DNA not having proper placement, either due to the DNA not standing upright or otherwise not behaving as we expected. To remedy these issues, this thesis explores using streptavidin and biotin to attach fluorescent molecules to gold nanopatterned surfaces. A procedure was developed to adsorb a single layer of streptavidin to both gold and glass to which a biotinylated target molecule can then be attached. Future work will investigate the feasibility of this method for FRET enhancement