bioRxiv Science⌕ Search
Home/Archive/bioRxiv

bioRxiv · 10.1101/2025.06.12.659203

Quantifying Misuse of Color in Biological Research

Lin, T.-J.·Hillman, K.·Kim, L.·Garg, M.·Duong, M.·Landry, M. P.
View original source ↗Export metadata ↓

Abstract

Colors are widely used to report biological research data. Effective interpretation of colors relies on accurate, intuitive, and accessible colormaps that are properly labeled. Misuse of colormaps, such as using rainbow, mismatched, and other inaccessible colormaps, hinders reader interpretation. Despite their ubiquitous use to represent biological data, the prevalence of colormap misuse has not been quantified. Using established guidelines, we analyzed over 6,000 articles in 10 high-impact biological journals spanning 2020-2024 to quantify the prevalence of colormap use and misuse. Over 67% of articles contain at least one colormap, of which 81% contain some form of misuse. Among the over 11,000 acquired colormap-associated figures, 60% contain some form of misuse. Heatmaps, scatter plots, images, and image overlays are the most commonly used figures with colormaps. These statistics vary by journal but remain constant over time, demonstrating the urgent need to increase awareness and education about color-based data reporting guidelines.

Explore related subjects

scientific communication and educationPreprint

Keep this discovery

Explore connections, maps & timelines

BibTeXRIS

Lin, T.-J., Hillman, K., Kim, L., Garg, M., Duong, M., Landry, M. P.. 2025-06-15. Quantifying Misuse of Color in Biological Research. https://doi.org/10.1101/2025.06.12.659203

Cite the original work for its findings. Save a collection to share your selection of sources.

KEEP EXPLORING

Related preprints

bioRxiv2026

Evaluating Large Language Models as Tools to Navigate Researchers in Rapidly Evolving Research Landscapes: A Case Study in Cancer Drug Response Prediction

Large Language Models (LLMs) have emerged as promising tools for assisting researchers in automating and accelerating the synthesis of literature reviews. However, their reliability is a significant concern due to issues like factual inaccuracies and hallucinations. The key question is whether LLMs can reliably provide comprehensive, up-to-date overviews and analyses. This study evaluates the performance of three leading LLMs (OpenAI's ChatGPT, Google's Gemini, and DeepSeek) on the complex task of generating a comprehensive survey paper on deep learning for cancer Drug Response Prediction (DRP). By testing both standard and Deep Research (DR) / Deep Think (DT) modes of LLMs with prompts of varying detail, this paper assesses key academic dimensions, including reference management, content quality, and analytical depth. Key findings reveal that while DR modes of LLMs significantly improve reliability by eliminating hallucinations, performance variations exist across models and prompts. A trade-off between reference quantity and integration quality was observed, and even the best-performing models lacked the analytical depth of human experts, often requiring extensive human supervision. The study concludes that LLMs currently serve as powerful assistive tools but still cannot replace the critical validation and synthesis provided by human researchers. Choosing the best LLM to use depends on the task in hand, while several strategies can be implemented to improve the produced output.

scientific communication and education↗
bioRxiv2025

Attention, Emotion, and Authenticity: Eye-Tracking Evidence from AI vs. Human Visual Design

This study investigates how viewers perceive, attend to, and emotionally respond to AI-generated versus human-created visual content, integrating multimodal data from eye-tracking, facial-coding, and self-report surveys. The sample consisted of 136 undergraduate and graduate students enrolled in a graphic design program at a public university. Participants viewed a series of static and video stimuli produced either by human designers or artificial intelligence systems. Gaze behavior (fixation count, duration, and saccade length), emotional reliability (k-coefficient from RealEye facial-coding), and attitudinal evaluations were analyzed through both parametric and nonparametric statistical tests. The results reveal that human-made visuals elicited longer viewing durations (M = 7035 ms), higher fixation counts (M = 1.44), and broader spatial exploration, suggesting richer semantic and aesthetic engagement. In contrast, AI-generated images produced shorter but more focused attention patterns (M = 4945 ms) and higher but less stable emotional reactions (k = 0.16). The correlation between fixation metrics and affective responses was non-significant ({rho} = -0.015), indicating that cognitive attention and emotional resonance operate as distinct dimensions. Attitudinal data showed a 68.4% accuracy in attributing authorship, with AI visuals often misclassified as human-made reflection of perceptual authenticity bias. Participants described AI content as technically refined yet emotionally limited. These findings suggest that while AI imagery achieves perceptual salience, it still lacks the emotional intentionality and narrative coherence that characterize human creativity.

scientific communication and education↗
bioRxiv2025

Participatory development of innovation and implementation strategy - a practical approach

BackgroundHealthcare and academic institutions face growing challenges in strategic planning due to rapid advances in medicine and technology, alongside fiscal and workforce constraints that limit traditional consultation. Participatory approaches offer a way to integrate diverse stakeholder perspectives under these constraints, generating contextually relevant strategies that can indicate whether current directions are appropriate or whether priorities have been overlooked. MethodsA structured participatory workshop was conducted at the 10th Grampian Research Conference (June 2025). One hundred seventy-eight participants including National Health Service (NHS) staff, academics, industry partners, patients, and public contributors, engaged in 14 parallel roundtable discussions. Contributions were captured using posters and Post-it notes, collecting 148 written annotations. Data were analysed using thematic and content analysis, supplemented by strategic frameworks including Strengths, Weaknesses, Opportunities and Threats (SWOT/TOWS), and Easy Wins, to identify and prioritise actionable strategies. ResultsFive core themes emerged: (1) access to healthcare and services, (2) patient and public involvement and engagement, (3) digital health and service delivery innovation, (4) data access, integration, and governance, and (5) workforce development and culture. SWOT analysis identified strengths in telemedicine, interdisciplinary student training, and patient and public involvement, alongside weaknesses in fragmented data, referral tracking, and workforce pressures. TOWS matrix produced strategy-oriented recommendations such as AI-enabled scheduling, remote monitoring, and transparent referral systems. Easy Wins framework assessment highlighted immediate, low-cost improvements including identifiable NHS caller identification, automated text message reminders, updated informational videos and multilingual materials. ConclusionBy combining participatory outputs with structured strategy tools, this approach demonstrated a resource-efficient model for adaptive planning. The findings align with and extend current national health policy frameworks, offering a replicable approach for institutions aiming to obtain meaningful stakeholder engagement despite fiscal and temporal constraints.

scientific communication and education↗