Mason Journals (George Mason Univ.)
Not a member yet
3256 research outputs found
Sort by
Characterization of a De Novo DNA-Binding Protein Targeting Five Operator Sites in the 933W Phage
No full textWe characterized a series of de novo proteins that each bind to five unique DNA sequences, offering insight into the minimal engineering requirements for this function. This was investigated via the 933W lambdoid phage associated with food poisoning from shiga toxin producing E. coli. When 933W infects E. coli, there are two possible life cycles: lytic and lysogenic. The genetic switch between these two pathways is controlled by a DNA-binding protein that recognizes five unique sequences. Activation of the lytic pathway via the Cro protein results in toxin production and host cell death. Activation of the lysogenic pathway via the CI protein integrates the genome into the host’s genome, where the virus is dormant and no toxin is produced. This project characterizes our efforts to engineer a de novo DNA-binding protein that outcompetes the native CI and Cro proteins at all five DNA sites. The proteins were designed with a recognition helix from the 933W CI protein, incorporated diagonally into a doubled four-helix bundle. This was named Double Positive Mutant Diagonal (DPMD). DPMD was expressed via recombinant DNA techniques, and its structure was verified by circular dichroism (CD). Surface Plasmon Resonance (SPR) was used to assess DPMD’s binding affinities to three sequences: 933WOR1, 933WOR2, and 933WOR3. DPMD binds to 933WOR1 with a KD of 1.76 nM, to 933WOR2 with a KD of 3.16 nM. Current work focuses on collecting SPR data for 933WOR3. Future experiments will compare variants with altered recognition helix placement
Thickness–dependent and Wavelength–dependent Dichroic Response of Layered Rhenium Sulfide
No full textAnisotropic two-dimensional (2D) materials, particularly transition metal dichalcogenides (TMDs), are promising candidates for next-generation optical sensors due to their unique responses to polarized light. Rhenium disulfide (ReS₂) possesses distinctive directional characteristics owing to its in-plane structural anisotropy. This anisotropy manifests in the differential absorption of light, which varies depending on the polarization direction. A thorough comprehension of the effects of flake thickness and incident wavelength on this anisotropic behavior continues to be an area of ongoing research. This study examines the sensitivity of ReS₂ flakes to light polarization through comprehensive absorption measurements across various polarization angles, controlled using a polarizer. The dichroic factor, defined as the ratio of maximum absorption (parallel to the crystal b-axis) to minimum absorption (perpendicular to the crystal b-axis), was subsequently calculated. The dichroic factor exhibits a decreasing trend as the flake thicknesses increase. Within the range of 9 to 197 nm, the dichroic factor decreased by approximately 54%, indicating that thinner layers are more susceptible to polarization manipulation. Angular absorption measurements conducted at diverse wavelengths indicate an enhanced dichroic factor at longer wavelengths. This phenomenon is attributed to more pronounced polarization dependence near band-edge excitation. These findings offer new insights into the wavelength and thickness-dependent anisotropy of ReS₂, supporting its potential application in polarization-sensitive and wavelength-tunable photodetectors
pH-Dependent Electrochemistry of Organic Electrode Materials in Aqueous Electrolytes
No full textCommercial rechargeable batteries predominantly use organic and non-aqueous electrolytes, which have significant potential in terms of redox voltage and energy density. However, aqueous electrolytes offer distinct advantages, such as reduced costs, decreased flammability, and enhanced environmental safety. In this study, we delve into the thermodynamic and electrochemical kinetics of an organic electrode material in aqueous electrolytes. Specifically, DAPT synthesized from perylene-3,4,9,10-tetracarboxylic dianhydride (PTCDA) and N1, N1-diphenylbenzene-1,4-diamine (DPA) was employed as the model system. The redox chemistry of DAPT involves ion-coupled electron transfer at the active redox carbonyl groups. A tunable parameter, pH, was employed to regulate the availability of protons within the reaction mechanism. Accordingly, glassy carbon electrodes coated with drop-casted DAPT were tested using cyclic voltammetry in sodium-phosphate buffers with pH values ranging from 2.0 to 12.0. The equilibrium potential (Eeq) dependence on pH reveals the proton-coupled electron transfer reaction mechanism, which operates with a 1:1 proton/electron ratio within the pH range of 2–7. At pH > 7, Eeq becomes independent of pH, indicating a transition to sodium ion-coupled electron transfer. The reaction kinetics determined from the peak separation (ΔEp) exhibit a strong dependence on pH, with a decreasing kinetics within pH 2–9 and a reversed trend at pH 9–12. These findings underscore the crucial roles of cation type and concentration in influencing the thermodynamic and electrochemical kinetics of organic electrode materials. This research is expected to advance the characterization of this material as a sustainable and safe candidate for aqueous electrolyte batteries
Evaluating AI Models in SWE Prediction
No full textSnow Water Equivalent (SWE) is an essential metric for analyzing water supply and climate in specific areas. Over the years, various AI models have been used to predict SWE. These AI models allow us to monitor SWE from a distance; however, the accuracy of each type of model is still up for debate. In this study, we built and tested 4 different types of AI models to measure their accuracy in SWE prediction: a Graph Neural Network (GNN), a Transformer, a Long Short-Term Memory (LSTM) model, and a Random Forest (RF) model. Currently, GNN, which uses graph nodes to generate predictions, and Transformer, which uses self-attention to determine the importance of variables, are considered to be among the most cutting-edge models for SWE prediction. LSTM, which uses memory to learn prediction patterns, and RF, which uses decision trees, are considered to be older models. The models were trained on historical data sets from 2018 to 2021, which included variables such as precipitation, air temperature, and humidity. Model accuracy was measured based on Mean Squared Error (MSE) and the R2 metric. Our current findings show that the GNN model had the lowest MSE and the highest R2 overall, showing the highest accuracy. This suggests that models that can represent data as graphs can be more effective for SWE prediction
Optimizing Dynamic Learning from Learned Hallucination (Dyna-LfLH) using a Transformer-based Motion Planner for Agile Navigation in Dynamic Environments
No full textAutonomous robot navigation in dense, dynamic environments is a fundamental challenge. Classical planners struggle with unpredictable obstacles due to computation limits, while learning-based methods are hampered by needing high-quality demonstrations for Imitation Learning or exploration inefficiencies in Reinforcement Learning. Self-supervised frameworks like Dynamic Learning from Learned Hallucination (Dyna-LfLH) offer a promising solution by generating training data without costly demonstrations. However, current frameworks are limited by less expressive motion planners. The original Dyna-LfLH framework utilized a two-layer Recurrent Neural Network (RNN) that struggles to model complex, long-range spatio-temporal dependencies. This work replaces the RNN planner with an autoregressive, Transformer-based architecture. Our model uses a multi-layer causal Transformer encoder with GELU activations to process sequences of command embeddings fused with features from a three-layer convolutional LiDAR encoder. This multi-head self-attention architecture enables a crucial shift from myopic single-command prediction to multi-step trajectory forecasting, providing foresight for smoother navigation. We trained our model on an Imitation Learning dataset of 53,000 dynamic and open-space augmentation (OSA) examples generated via the Dyna-LfLH algorithm. The model learns to predict a 3-step command trajectory of linear and angular velocities (v,ω) from historical 2D LiDAR scans, prior commands, and a target position vector, minimizing Mean Squared Error (MSE). The final model was evaluated on a Jackal robot in the DynaBARN simulation testbed, achieving an improvement in collision avoidance compared to the original RNN-based implementation. This research presents a Transformer-based framework for motion planning in dynamic, unpredictable environments, improving the safety and reliability of autonomous navigation
Data-Driven Prediction of Metal-Organic Dissolution and Adhesion Using Machine Learning
No full textNatural biomolecules, like phenolic compounds, are important to regulate metal ion dissolution and adhesion in various chemical and biological systems. However, there is a gap in predictive understanding of how environmental variables, organic ligands, and metal properties interact to influence these processes. This gap limits the production of accurate models for predicting metal-organic interactions, which are important in fields ranging from sustainable agriculture to biomaterials engineering.
This study seeks to fill this gap by compiling a dataset of 71 experimentally validated metal-organic-environmental interactions, 45 adhesive and 26 dissolution reactions, and training machine learning models to predict both outcomes. The molecular properties of metal ions, including ionic radius, polarizability, electron affinity, atomic number, and electron orbital configurations, were compiled from public databases like PubChem. 41 organic compounds were processed through Mordred to generate 1,614 structural and topological molecular descriptors per compound. Combined with environmental variables like pH, compound concentration, reaction temperature, and reaction speed, this information was merged into a single unified dataset for model development.
To support the predictive model, Random Forest and Support Vector Machine (SVM) models were trained with two main goals: to identify environmental conditions and molecules that maximize reactivity and produce favorable results, and to identify key molecular features and functional groups driving dissolution or adhesion. By learning from known reactions, the models reduce the need for trial-and-error experimentation and allow for improved compound selection. This work establishes a scalable, data-driven platform for simulating and guiding metal-organic reactivity in both natural and engineered environments. 
The Ebb and Flow of Ideas in Nation-Building: The Tumultuous Run of "The Times of Vietnam"
Full text linkNeilesh Bose, Editor, South Asian Migrations in Global History: Labor, Law, and Wayward Lives
Full text linkAnalyzing Prompt Engineering for Code Generation Accuracy with o4-Mini on CodeForces Problems
No full textGenerative AI tools such as OpenAI’s o4-mini are increasingly used to aid programmers in solving coding tasks, but their effectiveness often depends on the quality of the prompts, which can be improved through a process known as prompt engineering. While previous studies have explored prompt engineering in software development contexts, its impact on competitive programming tasks, which require more precise reasoning and problem-solving, remains underexplored. In this study, we evaluate the effect of four different prompt strategies—No Prompt, Zero-Shot Chain-of-Thought (CoT), Prompt Chaining, and Advanced CoT—on o4-mini’s performance for 90 advanced Codeforces problems rated 1600–2400. These problems were released after May 31st, 2024, the official knowledge cutoff date for o4-mini, to reduce overlap with the model’s training data. AI-generated code was submitted into the official Codeforces grader, and results were measured by acceptance rate. Among all methods, Advanced CoT achieved the highest acceptance rate at 56.67%, significantly outperforming No Prompt (44.44%) and Zero-Shot CoT (42.22%), and marginally outperforming Prompt Chaining (55.56%). McNemar’s test (significance level of 0.0166 with a Bonferroni Correction) confirmed that the improvement from Advanced CoT over No Prompt was statistically significant (p = 0.0139), while the comparison between Prompt Chaining and No Prompt approached significance (p = 0.0249). Together, these results strongly support the hypothesis that carefully structured prompts can lead to increased performance from LLMs on high difficulty, reasoning-intensive tasks. Specifically, achieving over 56% accuracy on recently released and likely unseen problems demonstrates o4-mini's impressive reasoning capabilities when guided by effective prompting, especially given its affordability at just $1.10 per million tokens. Future work could expand this experiment to a broader, more diverse range of problems and to various LLMs to compare their performance with o4-mini.
A Comparative Study of Gemini 2.5 Pro and Human Evaluative Metrics in Assessing Student Comprehension Across STEM and Humanities Topics through Conversation-based Assessments
No full textTo build Argumentation and Graph-based Open Student Modeling Software (ARGOS), an Intelligent Tutoring System (ITS) with a dynamic Open Student Model (OSM), the project utilized Gemini 2.5 Pro as the Large Language Model (LLM) to power the project. To validate the OSM, an experiment was conducted to evaluate whether Gemini can accurately assess a student's mastery through rubric-based grading. The 100-point rubric analyzed the following areas of student performance: accuracy, clear reasoning, conciseness and clarity, relevance, and effort. Eight simulated student conversations, four in algebra and four in history, were submitted line by line to Gemini, along with the system prompt and rubric. The system prompt instructed Gemini to return the student's mastery score for each turn at the very end of the conversation. To establish a baseline for comparison, four human researchers independently scored each conversation using the same rubric. The analysis revealed a strong Pearson correlation coefficient (r = 0.837) between Gemini's overall mastery scores and human averages, with a consistent positive or negative difference between turns. However, Gemini consistently scored students about 10 percent lower on average and showed more inconsistency in its scoring (SD = 14.66). While Gemini performed well on math problems with clearly defined steps, it struggled with open-ended humanities responses, often failing to recognize when a student partially understood a topic. Future versions of ARGOS will explore graph-based representations of student thinking to capture a more human-like and accurate analysis of a student's complete understanding of any topic