AI Models Identify H3K18ac as New Marker of Active Gene Enhancers
Researchers used explainable AI models to predict DNA enhancers and identified H3K18ac as an important epigenetic marker of active enhancers, validated through experimental perturbations and CRISPR-based rewriting. The study addresses a significant gap in understanding enhancer activity, which regulates gene transcription but remains incompletely mapped across human cells and tissues. This work could improve disease research and therapeutic development by enabling more accurate enhancer identification across different biological contexts.
A research team developed multiple AI models—including Convolutional Neural Networks, XGBoost, Logistic Regression, and an explainable type2 Fuzzy Logic system—to predict enhancers (non-coding DNA regions that regulate gene transcription) across human and mouse cell lines. While all models showed high accuracy on their training data, the type2-FLS model and partially the CNN and LR models generalized well to unseen cell lines, suggesting robust performance. The explainable AI approach identified H3K18ac as an important enhancer marker alongside novel putative enhancers with epigenetic signatures matching experimentally validated ones. The researchers validated some predictions through global epigenetic perturbations and directed enhancer rewriting using CRISPRi technology. Notably, just seven epigenetic marks in humans and five in mice proved sufficient to annotate enhancers without sacrificing accuracy, effectively deciphering the epigenetic code of mammalian enhancers.
What's missing
The study's own limitations and open questions are not detailed in the abstract provided. Specific information about the size of the validation dataset, the number of novel enhancers identified, and the biological significance of H3K18ac compared to previously known enhancer marks would strengthen interpretation of the findings.
What different sources said
- bioRxivCenter
Explainable AI identifies H3K18ac as a new marker of active enhancers
Related
Urban Spider Populations Show Increased Body Size but Reduced Body Condition, Study Finds
A study of European garden spiders across rural-urban gradients in Belgium found that urbanization is associated with larger body sizes but reduced abdominal area (indicating lower body condition and reproductive investment). The research examined how multiple traits—including size, coloration, and thermoregulation—respond to urbanization at different spatial scales. Understanding how urban environments affect spider morphology and physiology is important for predicting how ectothermic species adapt to cities and the ecological consequences of urbanization.
Researchers Propose 'Generativism' as New Learning Theory for Generative AI Era
Computer science researchers have published a framework called 'Generativism' that proposes a new learning theory designed specifically for educational environments where generative AI is present. The framework argues that existing learning theories—behaviorism, cognitivism, constructivism, and connectivism—have conceptual limitations when applied to AI-assisted learning. The proposal matters because it could reshape how educators design instruction, assessment, and skill development as generative AI becomes increasingly integrated into learning.
New Method for Detecting AI Hallucinations in Real-Time Using Statistical Change-Point Theory
Researchers have developed a new approach to detect when AI language models begin producing hallucinations (false information) by framing the problem as a statistical change-point detection task. The method uses a learned CUSUM (Cumulative Sum Control Chart) algorithm that can identify hallucination onset in 11-13 tokens, compared to 31 tokens for baseline methods. This matters because real-time hallucination detection is critical for deploying AI systems safely, and the research reveals fundamental limits on how quickly such detection is theoretically possible.