❌

Reading view

Multi-Omics and Functional Analyses Identify let-7b-3p as a Negative Regulator of EMT in Lung Adenocarcinoma

J Biochem Mol Toxicol. 2026 Feb;40(2):e70700. doi: 10.1002/jbt.70700.

ABSTRACT

Lung adenocarcinoma (LUAD) is the most common subtype of non-small cell lung cancer (NSCLC), and its malignant progression involves complex molecular mechanisms. While microRNAs (miRNAs) play a crucial regulatory role in LUAD development, their specific mechanisms remain unclear. This study used bioinformatics analysis to identify key miRNA-mRNA interaction axes in LUAD, revealing that let-7b-3p was significantly downregulated. Functional analyses demonstrated that let-7b-3p regulates LUAD cell proliferation, migration, and invasion by targeting High Mobility Group AT-Hook 2 (HMGA2) and Lin-28 Homolog A (LIN28A). Dual-luciferase reporter assays confirmed that let-7b-3p directly binds to HMGA2 and LIN28A, suppressing their expression. Furthermore, Western blot and immunofluorescence (IF) assays showed that let-7b-3p inhibits the Wnt/TGF-β signaling pathway and epithelial-mesenchymal transition (EMT) via the HMGA2-LIN28A axis. In vivo, experiments using a nude mouse model further demonstrated that let-7b-3p overexpression significantly suppressed LUAD tumor growth and lung metastasis while reducing the expression of EMT-related molecules. Importantly, this study is the first to reveal the inhibitory role of let-7b-3p in LUAD through the HMGA2-LIN28A axis in regulating the Wnt/TGF-β signaling pathway and EMT. These findings highlight the originality of this work and underscore the potential clinical translational value of targeting let-7b-3p or the HMGA2-LIN28A axis as novel therapeutic strategies for LUAD.

PMID:41586577 | DOI:10.1002/jbt.70700

  •  

A machine learning-derived polygenic risk score reveals that healthy lifestyle counteracts obesity-related mortality

npj Digital Medicine, Published online: 06 January 2026; doi:10.1038/s41746-025-02314-6

A machine learning-derived polygenic risk score reveals that healthy lifestyle counteracts obesity-related mortality
  •  

GuardFed: A Trustworthy Federated Learning Framework Against Dual-Facet Attacks

arXiv:2511.09294v1 Announce Type: cross Abstract: Federated learning (FL) enables privacy-preserving collaborative model training but remains vulnerable to adversarial behaviors that compromise model utility or fairness across sensitive groups. While extensive studies have examined attacks targeting either objective, strategies that simultaneously degrade both utility and fairness remain largely unexplored. To bridge this gap, we introduce the Dual-Facet Attack (DFA), a novel threat model that concurrently undermines predictive accuracy and group fairness. Two variants, Synchronous DFA (S-DFA) and Split DFA (Sp-DFA), are further proposed to capture distinct real-world collusion scenarios. Experimental results show that existing robust FL defenses, including hybrid aggregation schemes, fail to resist DFAs effectively. To counter these threats, we propose GuardFed, a self-adaptive defense framework that maintains a fairness-aware reference model using a small amount of clean server data augmented with synthetic samples. In each training round, GuardFed computes a dual-perspective trust score for every client by jointly evaluating its utility deviation and fairness degradation, thereby enabling selective aggregation of trustworthy updates. Extensive experiments on real-world datasets demonstrate that GuardFed consistently preserves both accuracy and fairness under diverse non-IID and adversarial conditions, achieving state-of-the-art performance compared with existing robust FL methods.
  •  

A full life cycle biological clock based on routine clinical data and its impact in health and diseases

Nature Medicine, Published online: 27 October 2025; doi:10.1038/s41591-025-04006-w

The biological clock model LifeClock predicts biological age across all life stages from routine clinical data, revealing distinct pediatric and adult disease risk patterns.
  •  

Long-read RNA sequencing dataset of human pancreatic cancer cell lines

Sci Data. 2025 Oct 20;12(1):1653. doi: 10.1038/s41597-025-05939-0.

ABSTRACT

Long-read RNA sequencing (RNA-seq) technologies have revolutionized transcriptomic research by enabling the sequencing of full-length RNA molecules, thus providing a more accurate characterization of complex transcript isoforms than traditional short-read approaches. In this study, we present a high-coverage long-read transcriptome dataset generated using Oxford Nanopore Technologies' PromethION platform from ten human pancreatic cancer cell lines, with two biological replicates per line. The dataset comprises approximately 189.8 million reads across 20 samples, providing a valuable resource for studying transcript structures in pancreatic cancer. We perform systematic quality assessments, including read length, base quality, and gene body coverage, and report high reproducibility between replicates. Processed files, including transcript annotations in GTF, FASTA, and BED formats, are publicly available to facilitate reuse. This resource supports a wide range of downstream applications such as isoform discovery, transcriptome annotation, and integration with other omics data, offering a foundation for further exploration of transcriptomic complexity in cancer biology.

PMID:41115920 | PMC:PMC12537988 | DOI:10.1038/s41597-025-05939-0

  •  

Application of Behavioral Science in Digital Therapeutics for Individuals With Prediabetes: Scoping Review

Background: Digital therapeutics are increasingly used to manage prediabetes due to their accessibility and potential for personalization. Their success depends heavily on applying behavioral science and integrating theoretical models into digital platforms. However, there has not been a comprehensive account of how behavioral science has been used in digital therapeutics for individuals with prediabetes. Objective: This scoping review aimed to examine the use of behavioral theories and techniques in digital therapeutic interventions for individuals with prediabetes, and to identify opportunities to optimize theory-driven and technology-supported strategies. Methods: A scoping review was conducted following the Arksey and O’Malley framework and guided by the PRISMA-ScR (Preferred Reporting Items for Systematic Reviews and Meta-Analyses extension for Scoping Reviews) checklist. We systematically searched PubMed, Embase, Web of Science, the Cochrane Library, Scopus, CNKI, and VIP Database for Chinese Technical Periodicals for studies published up to March 10, 2025. Eligible studies included adults (≥18 years) with prediabetes, as defined by the American Diabetes Association, and examined digital therapeutic interventions informed by behavioral science. All study designs were eligible; included studies were screened, and key characteristics were charted. Results: Of the 21 included studies, 17 were randomized controlled trials. The most frequently used behavioral theories were social cognitive theory, theory of planned behavior, and transtheoretical model; however, 11 studies applied behavior change techniques without explicitly stating a theoretical framework. In terms of delivery, digital modalities often comprised smartphone apps (14/21, 67%), human coaching (13/21, 62%), messaging tools (9/21, 43%), wearable devices (9/21, 43%), and web platforms (3/21, 14%). About behavior change techniques, the most frequently used were self-monitoring of behavior (19/21), instruction on performing the behavior (16/21), goal setting (15/21), information about health consequences (15/21), and unspecified social support (11/21). Across studies, outcomes were typically assessed for metabolic and body composition (19/21), glycemic control metrics (17/21), cardiovascular risk and physiological function metrics (16/21), behavioral and cognitive intervention indicators (11/21), and, less frequently, comprehensive health outcome measures (2/21). Conclusions: Behavioral science plays a crucial role in developing effective digital therapeutics for individuals with prediabetes. However, greater clarity in theory selection, better integration between models and digital functions, and more culturally inclusive research are needed to improve the scalability and impact of these interventions.
  •  

Multi-omics analysis identifies UBA family as potential pan-cancer biomarkers for tumor prognosis and immune microenvironment infiltration

Front Immunol. 2025 Feb 17;16:1510503. doi: 10.3389/fimmu.2025.1510503. eCollection 2025.

ABSTRACT

BACKGROUND: UBA1 and UBA6 are classic ubiquitin-activating E1 enzymes, which participate in the ubiquitination degradation of intracellular proteins and are closely related to the occurrence and development of various diseases and tumors. However, at present, comprehensive analysis has not been used to study the role of UBA family in cancers.

METHODS: We extracted the relevant data of cancer patients from the TCGA database and studied the relationship between the expression patterns of UBA family and the survival rate, and stage of patients in pan-cancer, especially breast cancer (BRCA), colorectal cancer (COAD), renal cancer (KIRC) and lung adenocarcinoma (LUAD). In addition, we also evaluated their impact on immune infiltration using TISIDB database and R packages.

RESULTS: UBA1 and UBA6 are highly expressed in most cancer types, which may be associated with poor prognosis of patients. This study also investigated their expression had a closely tie with clinical stages in some specific tumors. Furthermore, this study also demonstrated that these genes were closely related to immune score, immune subtypes and tumor infiltrating immune cells.

CONCLUSIONS: Our study demonstrated that the differential expression of the UBA family, along with their associated survival landscape and immune infiltration across various cancer types, holds potential as biomarkers linked to cancer immune infiltration. This finding offers a novel perspective for informing the direction of cancer treatment strategies.

PMID:40046044 | PMC:PMC11880792 | DOI:10.3389/fimmu.2025.1510503

  •  
❌