3821 papers
This study presents a standardized dataset of 3,607 length-at-maturity (L50) estimates for 48 Northeast Atlantic marine fish species, derived from approximately 1.6 million individual records across 21 fishery-independent surveys to enable high-resolution analysis of maturation schedules across populations, sexes, and time.
This study demonstrates that after matching neuroblastoma patients on conventional clinical variables, favorable and unfavorable outcomes are distinguished not by unstructured heterogeneity but by a robust, shared bidirectional regulatory axis in transcription-factor activity that separates cell-cycle programs from inflammatory and mesenchymal states, suggesting a potential target for therapeutic intervention.
This study demonstrates that acquisition duration in calcium imaging can be optimized as an inverse design problem to determine the minimum time required for specific tasks, revealing that protocols tailored to different indicators and conditions achieve peak performance at specific durations beyond which longer acquisition actually degrades inference accuracy.
This study utilizes a software model based on longitudinal chromatic aberration to demonstrate how prolonged near work induces myopia through visual feedback mechanisms, while also validating existing theories on lens correction and proposing a novel strategy of marginally blurring blue light on digital screens to inhibit myopia onset and progression.
This study identifies endothelial RANK signaling as a critical driver of bone marrow inflammaging by inducing IL-1β expression, which leads to senescent cell accumulation, hematopoietic stem cell dysfunction, and myeloid-biased hematopoiesis, thereby offering a potential therapeutic target for age-related diseases.
This study demonstrates that the Parkinson's disease-associated LRRK2 G2019S mutation exacerbates intestinal inflammation and enteric α-synuclein accumulation by driving excessive neutrophil extracellular trap formation through a kinase-dependent RAB10–myeloperoxidase signaling axis.
MitoSeqGen is a novel constrained generative transformer trained on a large vertebrate mitochondrial dataset that outperforms existing codon-optimization models by guaranteeing 100% protein correctness under the distinct mitochondrial genetic code while achieving superior similarity to natural codon sequences.
This systematic review of 21 studies (2020–2025) evaluates 54 machine learning models for host-based infectious disease diagnostics, finding that while Random Forest and neural networks are the most common algorithms, no single model is universally optimal, and future clinical translation requires standardized reporting, external validation, and context-aware algorithm selection.
DeepCNet is a novel multimodal deep learning framework that leverages single-cell multiome data to accurately predict cell type-specific gene expression and infer regulatory promoter-enhancer interactions, thereby enabling the construction of transcription factor gene regulatory networks and the identification of condition-specific regulatory alterations in diseases like Alzheimer's.
This study of pandemic-born children in Spain finds that while prenatal SARS-CoV-2 exposure itself does not uniformly predict language delays at 18 months, a pro-inflammatory IL-6/IL-10 imbalance within exposed pregnancies is associated with lower performance in specific language domains like difficult verbs and vocalizations, particularly among girls.