IEEE Computer Graphics and Applications最新文献

A goal of data visualization is to advance the understanding of multiparameter, large-scale datasets. In astrophysics, scientists map celestial objects to understand the hierarchical structure of the universe. In biology, genetic sequences and biological characteristics uncover evolutionary relationships and patterns (e.g., variation within species and ecological associations). Our highly interdisciplinary project entitled "A Cosmic View of Life on Earth" adapts an immersive astrophysics visualization platform called OpenSpace to contextualize diverse biological data. Dimensionality reduction techniques harmonize biological information to create spatial representations in which data are interactively explored on flat screens and planetarium domes. Visualizations are enriched with geographic metadata, 3-D scans of specimens, and species-specific sonifications (e.g., bird songs). The "Cosmic View" project eases the dissemination of stories related to biological domains (e.g., insects, birds, mammals, and human migrations) and facilitates scientific discovery.

In mass spectrometry-based proteomics, experts usually project data onto a single set of reference sequences, overlooking the influence of common haplotypes (combinations of genetic variants inherited together from a parent). We recently introduced ProHap, a tool for generating customized protein haplotype databases. Here, we present ProHap Explorer, a visualization interface designed to investigate the influence of common haplotypes on the human proteome. It enables users to explore haplotypes, their effects on protein sequences, and the identification of noncanonical peptides in public mass spectrometry datasets. The design builds on well-established representations in biological sequence analysis, ensuring familiarity for domain experts while integrating novel interactive elements tailored to proteogenomic data exploration. User interviews with proteomics experts confirmed the tool's utility, highlighting its ability to reveal whether haplotypes affect proteins of interest. By facilitating the intuitive exploration of proteogenomic variation, ProHap Explorer supports research in personalized medicine and the development of targeted therapies.

Visualization artifacts have long served as anchors for collaboration and knowledge transfer in data analysis. While effective for human-human collaboration, little is known about their role in capturing and externalizing knowledge when working with large language models (LLMs). Despite the growing role of LLMs in analytics, their linear text-based workflows limit the ability to structure artifacts into useful and traceable representations of the analytical process. We argue that dynamic visual representations of evolving analysis-organizing artifacts and provenance into semantic structures, such as idea development and shifts in inquiry-are critical for effective human-LLM workflows. We demonstrate the current opportunities and limitations of using LLMs to track, structure, and visualize analytic processes, and propose a research agenda to leverage rapid advances in LLM capabilities. Our goal is to present a compelling argument for maximizing the role of visualization as a catalyst for more structured, transparent, and insightful human-LLM analytical interactions.

Circular genome visualizations are essential for exploring structural variants and gene regulation. However, existing tools often require complex scripting and manual configuration, making the process time-consuming, error-prone, and difficult to learn. To address these challenges, we introduce AuraGenome, a large language model (LLM)-powered framework for rapid, reusable, and scalable generation of multilayered circular genome visualizations. AuraGenome combines a semantic-driven multiagent workflow with an interactive visual analytics system. The workflow employs seven specialized LLM-driven agents, each assigned distinct roles, such as intent recognition, layout planning, and code generation, to transform raw genomic data into tailored visualizations. The system supports multiple coordinated views tailored for genomic data, offering ring, radial, and chord-based layouts to represent multilayered circular genome visualizations. In addition to enabling interactions and configuration reuse, the system supports real-time refinement and high-quality report export. We validate its effectiveness through two case studies and a comprehensive user study. AuraGenome is available at https://github.com/Darius18/AuraGenome.

A central goal in neuropharmacological research is to alter brain function by targeting genes whose expression is specific to the corresponding brain circuit. Identifying such genes in large spatially resolved transcriptomics data requires the expertise of bioinformaticians for handling data complexity and to perform statistical tests. This time-consuming process is often decoupled from the routine workflow of neuroscientists, inhibiting fast target discovery. Here, we present a visual analytics approach to mining expression data in the context of meso-scale brain circuits for potential target genes tailored to domain experts with limited technical background. We support several workflows for interactive definition and refinement of circuits in the human or mouse brain, and combine spatial indexing with an alternative formulation of sample variance to enable differential gene expression analysis in arbitrary brain circuits at runtime. A user study highlights the usefulness, benefits, and future potential of our work.

Understanding how students learn visualization skills is becoming increasingly crucial as generative AI transforms technical education. We present a systematic study examining how structured exposure to large language models via Observable's AI Assist platform impacts data visualization education through a multiphase investigation across two universities. Our mixed-methods approach with 65 graduate students (32 data science and 33 computer science) revealed that structured generative AI exposure following constructivist learning principles enabled sustained engagement and tool adoption while maintaining pedagogical rigor. Through a structured multiphase study incorporating preassessments, intervention observations, detailed assignment reflections, and postintervention evaluation within the academic term constraints, we identified specific patterns in how students integrate generative AI into their visualization workflows. The results from our mixed-methods analysis suggest potential strategies for adapting visualization education to an AI-augmented future while preserving essential learning outcomes. We contribute practical frameworks for integrating generative AI tools into visualization curricula and evidence-based insights on scaffolding student learning with AI assistance, with initial evidence of sustained impact over a three-week period following instruction.

The pandemic had broad reaching impacts on how we do many things, including the way that we design and implement visualizations. In this article, we reflect on how visualization design changed in an emergency response. Based on these reflections, we present modifications to design methodologies for visualizations to accommodate an emergency response and its working conditions.

Reservoir computing is a form of machine learning particularly suited for time-series analysis, including forecasting predictions. We take an implementation of quantum reservoir computing that was initially designed to generate variants of musical scores and adapt it to create levels of Super Mario Bros. Motivated by our analysis of these levels, we develop a new Roblox obstacle course game (known as an "obby") where the courses can be generated in real time on superconducting qubit hardware and investigate some of the constraints placed by such real-time generation.

Recent advancements have enabled tissue samples to be profiled at the unprecedented level of detail of a single cell. Analysis of these data has enabled discoveries that are relevant to understanding disease and developing therapeutics. Large-scale profiling efforts are underway, which aim to generate "atlas" resources that catalog cellular archetypes, including biomarkers and spatial locations. While the problem of cellular data visualization is not new, the size, resolution, and heterogeneity of single-cell atlas datasets present challenges and opportunities. We survey the usage of visualization to interpret single-cell atlas datasets by assessing over 1800 figure panels from 45 biological publications. We intend for this report to serve as a foundational resource for the visualization community as atlas-scale single-cell datasets are emerging rapidly with aims of advancing our understanding of biological function in health and disease.

Protein folding is the process by which proteins go from a linear chain of amino acids to a 3-D structure that determines their biological function. Although recent advances in protein 3-D structure prediction can directly determine the folded protein's final shape, the process by which this happens is complex and not very well understood. Part of the study of protein folding focuses on the analysis of their "energy landscape," defined by the molecule's energy as a function of its structure. The data are mostly obtained through atomic-level computer simulations and are very high-dimensional, making them difficult to interpret. Visualization can be a powerful tool to support researchers studying the energy landscape of proteins; however, we noticed that they are not widely adopted by the scientific community. We present the main methods currently used and the challenges they face, as well as future opportunities for visualization in this field.