Front Genet. 2025 May 2;16:1569290. doi: 10.3389/fgene.2025.1569290. eCollection 2025.

ABSTRACT

The Cancer Genome Atlas (TCGA) provides comprehensive genomic data across various cancer types. However, complex file naming conventions and the necessity of linking disparate data types to individual case IDs can be challenging for first-time users. While other tools have been introduced to facilitate TCGA data handling, they lack a straightforward combination of all required steps. To address this, we developed a streamlined pipeline using the Genomic Data Commons (GDC) portal's cart system for file selection and the GDC Data Transfer Tool for data downloads. We use the Sample Sheet provided by the GDC portal to replace the default 36-character opaque file IDs and filenames with human-readable case IDs. We developed a pipeline integrating customizable Python scripts in a Jupyter Notebook and a Snakemake pipeline for ID mapping along with automating data preprocessing tasks (https://github.com/alex-baumann-ur/TCGADownloadHelper). Our pipeline simplifies the data download process by modifying manifest files to focus on specific subsets, facilitating the handling of multimodal data sets related to single patients. The pipeline essentially reduced the effort required to preprocess data. Overall, this pipeline enables researchers to efficiently navigate the complexities of TCGA data extraction and preprocessing. By establishing a clear step-by-step approach, we provide a streamlined methodology that minimizes errors, enhances data usability, and supports the broader utilization of TCGA data in cancer research. It is particularly beneficial for researchers new to genomic data analysis, offering them a practical framework prior to conducting their TCGA studies.

PMID:40385985 | PMC:PMC12081331 | DOI:10.3389/fgene.2025.1569290

EMI Anim Environ. 2025 May 8:1-37. doi: 10.1080/29986990.2025.2503155. Online ahead of print.

ABSTRACT

Sapovirus (SaV) infections have been linked with moderate-to-severe acute gastroenteritis (AGE) in animals and humans and represent a significant risk to public health. SaVs from animals including pigs, chimpanzees, and rodents have been reported to be closely related with human SaVs, indicating the possibility of cross-species transmission. Divergent SaVs have been reported in various bat species across various continents including Asia, Europe, Oceania and Africa. However, little is known about the evolutionary history of SaVs across various bat species and their zoonotic potential. In this report, we describe the findings of a surveillance study across various bat species in Nigeria. Samples were pooled and subjected to metagenomics sequencing and analyses. Nine of 57 sample pools (containing 223 rectal swabs from five bat species) had SaV reads from which we assembled a total of four complete and three near-complete (having complete coding sequences) genomes. The bat SaV (BtSaV) strains from this study formed five distinct lineages of which four represented novel genogroups. BtSaV lineages clustered mainly according to bat families, which might suggest a likely virus-host-specific evolution. The BtSaV VP1 capsid protein structure prediction confirmed three main domains (S, P1, and P2) as reported for Human SaV (HuSaV). We found that the P2 subdomain of the VP1 protein contains a degree of homology to known immunoreactive epitopes suggesting these conserved regions may be valuable for diagnostics or medical countermeasure development. This study expands our understanding of reservoir hosts, provides information on the genetic diversity and continuous evolution of SaVs in bats.

PMID:40385501 | PMC:PMC12080456 | DOI:10.1080/29986990.2025.2503155

iScience. 2025 Apr 16;28(5):112446. doi: 10.1016/j.isci.2025.112446. eCollection 2025 May 16.

ABSTRACT

Focusing on the early stages of Alzheimer's disease (AD) holds great promise. However, the specific events in neural cells preceding AD onset remain elusive. To address this, we utilized human-induced pluripotent stem cells carrying APPswe mutation to explore the initial changes associated with AD progression. We observed enhanced neural activity and early neuronal differentiation in APPswe cerebral organoids cultured for one month. This phenomenon was also evident when neural progenitor cells (NPCs) were differentiated into neurons. Furthermore, transcriptomic analyses of NPCs and neurons confirmed altered expression of neurogenesis-related genes in APPswe NPCs. We also found that the upregulation of reactive oxygen species (ROS) is crucial for early neuronal differentiation in these cells. In addition, APPswe neurons remained immature after initial differentiation with increased susceptibility to toxicity, providing valuable insights into the premature exit from the neural progenitor state and the increased vulnerability of neural cells in AD.

PMID:40384927 | PMC:PMC12084003 | DOI:10.1016/j.isci.2025.112446

Plant J. 2025 May;122(4):e70139. doi: 10.1111/tpj.70139.

ABSTRACT

We examined the nature and evolution of three-dimensional (3D) genome conformation, including topologically associating domains (TADs), in five genomes within the genus Oryza. These included three varieties from subspecies within domesticated Asian rice O. sativa as well as their closely related wild relatives O. rufipogon and O. meridionalis. We used the high-resolution chromosome conformation capture technique Micro-C, which we modified for use in rice. Our analysis of rice TADs shows that TAD boundaries have high transcriptional activity, low methylation levels, low transposable element (TE) content, and increased gene density. We also find a significant correlation of expression levels for genes within TADs, suggesting that they do function as genomic domains with shared regulatory features. Our findings indicate that animal and plant TADs may share more commonalities than were initially thought, as evidenced by similar genetic and epigenetic signatures associated with TADs and boundaries. To examine 3D genome divergence, we employed a computer vision-based algorithm for the comparison of chromatin contact maps and complemented this analysis by assessing the evolutionary conservation of individual TADs and their boundaries. We conclude that overall chromatin organization is conserved in rice, and 3D structural divergence correlates with evolutionary distance between genomes. We also note that individual TADs are not well conserved, even at short evolutionary timescales.

PMID:40384625 | DOI:10.1111/tpj.70139

Clin Transl Sci. 2025 May;18(5):e70222. doi: 10.1111/cts.70222.

ABSTRACT

Rational prescribing in geriatrics represents an important ethical as well as socio-economic issue. The aim of this project was to analyze the drug-related problems (DRPs) among the Czech nursing home residents and increase public awareness of further possible employment of clinical pharmacists in social care. The project was designed as a multicenter observational study. A total of 16 nursing homes and 800 participants with an average age of 84.6 ± 7.3 years were included in the study. Of them, a DRP was noted in 93.3% of people. The total amount of DRPs identified was 2215, which means an average of 2.8 ± 1.6 DRPs per patient. The most common DRPs identified were 'overtreatment' (19.5%), 'undertreatment' (12.8%), inappropriate dose (10.6%), recommendations for laboratory monitoring (10.4%) and adverse effects (10.3%). Of different drug classes, BZDs (OR 16.6, 95% CI 1.0-270.2), PPIs (OR 2.5, 95% CI 1.1-5.6) and NSAIDs (OR 4.4, 95% CI 1.1-18.3) were identified to be most commonly associated with DRPs. The risk of DRP identification clearly increased with the number of drugs used, with seven drugs demonstrated as the best cut-off for predicting DRP identification (AUC 0.842, sensitivity 0.602; specificity 0.796). 'SENIOR' project has confirmed a high rate of excessive polypharmacy among nursing home residents in the Czech republic resulting in high risk of potential and manifested DRPs. The project emphasized the role of clinical pharmacists in optimizing safety and effectiveness of treatment among older nursing home residents.

PMID:40388195 | DOI:10.1111/cts.70222

Clin Res Cardiol. 2025 May 19. doi: 10.1007/s00392-025-02637-0. Online ahead of print.

ABSTRACT

AIMS: Concerns exist about the safety of amiodarone and dronedarone. We assessed the long-term outcome of both drugs for early rhythm control (ERC) in the EAST-AFNET 4 trial.

METHODS AND RESULTS: Patients randomized for ERC and treated with amiodarone or dronedarone were compared to other ERC-therapies. Patients receiving amiodarone or dronedarone at initial therapy (n = 653/1395) were older with more comorbidities and less paroxysmal atrial fibrillation (AF, 29%) compared to patients never receiving amiodarone or dronedarone (Amiodarone/Dronedaronenever, 43% paroxysmal AF). Patients treated with amiodarone had more often heart failure (HF, 42%) and persistent AF (40%) compared to patients treated with dronedarone (16% HF, 15% persistent AF) and Amiodarone/Dronedaronenever (25% HF, 22% persistent AF). 115/398 amiodarone-treated patients (6.7/100 patient-years) and 51/255 dronedarone-treated patients (4.2/100 patient-years) experienced a primary efficacy outcome (cardiovascular death, stroke, HF-hospitalization or acute coronary syndrome), while 98/398 (5.3/100 patient-years) and 43/255 (3.4/100 patient-years) experienced a primary safety outcome (death, stroke or serious adverse events related to rhythm-control therapy). Serious adverse events related to drug therapy were similar for amiodarone (1.4/100 patient-years), dronedarone (1.2/100 patient-years), and other ERC (0.8/100 patient-years). Dronedarone (hazard ratio (HR) 0.5; CI 0.28-0.91), age (HR 1.05; CI 1.03-1.07), coronary artery disease (HR 1.84; CI 1.38-2.46) and stable HF (HR 1.66; CI 1.28-2.16) were associated with efficacy outcome upon multivariate Cox regression. Age (HR 1.07; CI 1.05-1.09) and left ventricular hypertrophy (HR 1.94; CI 1.13-3.32) were associated with safety outcome.

CONCLUSION: Early rhythm control using amiodarone or dronedarone rarely led to drug-related serious adverse events in EAST-AFNET 4.

CLINICAL TRIAL REGISTRATION: ISRCTN04708680, NCT01288352, EudraCT2010-021258-20.

PMID:40387892 | DOI:10.1007/s00392-025-02637-0

Stem Cells Transl Med. 2025 May 19;14(5):szaf018. doi: 10.1093/stcltm/szaf018.

ABSTRACT

This study investigated the safety and efficacy of MR-MC-01, a mesenchymal stem cell therapy derived from human embryonic stem cells, in patients with interstitial cystitis (IC), particularly those with Hunner lesions unresponsive to pentosan polysulfate sodium (PPS). Conducted as a prospective, randomized, double-blind, placebo-controlled phase I/IIa clinical trial, it enrolled 22 patients, with six completing phase I and 16 participating in phase IIa. Phase I tested 2 doses (2.0 × 107 and 5.0 × 107 cells) to determine the maximum tolerated dose (MTD), revealing no dose-limiting toxicities and only mild adverse events such as transient hemorrhage and bladder pain. In phase IIa, 12 participants received the MTD of 5.0 × 107 cells, and 4 received placebo. Significant reductions in interstitial cystitis questionnaire (ICQ) and pain urgency frequency (PUF) scores were observed in the treatment group. Improvements were noted in nocturnal voiding frequency and Hunner lesion size, with 8 patients showing either a reduction or complete resolution of lesions after 6 months. The global response assessment (GRA) reported moderate to marked improvement in 41.67% of treated patients versus 25% in the placebo group. MR-MC-01 demonstrated no serious drug-related adverse events, highlighting its favorable safety profile. These findings suggest that MR-MC-01 not only alleviates symptoms but also promotes structural recovery in IC, making it a promising treatment option. Further large-scale, long-term studies are warranted to confirm these results and optimize therapeutic protocols. (Identifier: NCT04610359).

PMID:40387787 | DOI:10.1093/stcltm/szaf018

Exp Biol Med (Maywood). 2025 May 2;250:10374. doi: 10.3389/ebm.2025.10374. eCollection 2025.

ABSTRACT

Adverse drug events are harms associated with drug use, whether the drug is used correctly or incorrectly. Identifying adverse drug events is vital in pharmacovigilance to safeguard public health. Drug safety surveillance can be performed using unstructured data. A comprehensive and accurate list of drug names is essential for effective identification of adverse drug events. While there are numerous sources for drug names, RxNorm is widely recognized as a leading resource. However, its effectiveness for unstructured data analysis in drug safety surveillance has not been thoroughly assessed. To address this, we evaluated the drug names in RxNorm for their suitability in unstructured data analysis and developed a refined set of drug names. Initially, we removed duplicates, the names exceeding 199 characters, and those that only describe administrative details. Drug names with four or fewer characters were analyzed using 18,000 drug-related PubMed abstracts to remove names which rarely appear in unstructured data. The remaining names, which ranged from five to 199 characters, were further refined to exclude those that could lead to inaccurate drug counts in unstructured data analysis. We compared the efficiency and accuracy of the refined set with the original RxNorm set by testing both on the 18,000 drug-related PubMed abstracts. The results showed a decrease in both computational cost and the number of false drug names identified. Further analysis of the removed names revealed that most originated from only one of the 14 sources. Our findings suggest that the refined set can enhance drug identification in unstructured data analysis, thereby improving pharmacovigilance.

PMID:40386037 | PMC:PMC12083459 | DOI:10.3389/ebm.2025.10374

Cureus. 2025 Apr 18;17(4):e82530. doi: 10.7759/cureus.82530. eCollection 2025 Apr.

ABSTRACT

We report a case of a 64-year-old man with advanced non-small cell lung cancer (NSCLC) who developed peritoneal metastasis during systemic treatment. Initially diagnosed with lung adenocarcinoma with pleural dissemination and bone metastases, he received carboplatin, pemetrexed, and pembrolizumab, followed by docetaxel due to clinical progression. While primary lung lesions responded to docetaxel, the patient developed new-onset abdominal pain and ascites. Radiologic findings suggested peritoneal thickening, raising suspicion for either docetaxel-induced toxicity or disease progression. Given the rarity of peritoneal metastasis in NSCLC and concurrent treatment response elsewhere, drug-induced complications were initially considered. However, worsening symptoms and further imaging prompted cytological evaluation of ascitic fluid, which confirmed metastatic adenocarcinoma consistent with lung origin. This case highlights the diagnostic challenge of distinguishing treatment-related adverse events from disease progression, especially in patients presenting with nonspecific abdominal symptoms during therapy. Clinicians should maintain a high index of suspicion for uncommon metastatic sites when new symptoms arise, even in the setting of apparent response at the primary site.

PMID:40385873 | PMC:PMC12085952 | DOI:10.7759/cureus.82530

Ther Adv Drug Saf. 2025 May 16;16:20420986251339358. doi: 10.1177/20420986251339358. eCollection 2025.

ABSTRACT

BACKGROUND: Drug-drug interactions (DDI) are an important cause of adverse drug reactions (ADRs). Could large language models (LLMs) serve as valuable tools for pharmacovigilance specialists in detecting DDIs that lead to ADR notifications?

OBJECTIVE: To compare the performance of three LLMs (ChatGPT, Gemini, and Claude) in detecting and explaining clinically significant DDIs that have led to an ADR.

DESIGN: Observational cross-sectional study.

METHODS: We used the French National Pharmacovigilance Database to randomly extract Individual Case Safety Reports (ICSRs) of ADRs with DDI (positive controls) and ICSRs of ADRs without DDI (negative controls) registered in 2022. Interaction cases were classified by difficulty level (level-1 DDI being the easiest and level-2 DDI being the most difficult). We give each LLM (ChatGPT, Gemini, and Claude) the same prompt and case summary. Sensitivity, specificity, and F-measure were calculated for each LLM in detecting DDIs in the case summaries.

RESULTS: We assessed 82 ICSRs with DDIs and 22 ICSRs without DDIs. Among ICSRs with DDIs, 37 involved level-1 DDIs, and 45 involved level-2 DDIs. Correct responses were more frequent for level-1 DDIs than for level-2 DDIs. Regardless of difficulty level, ChatGPT detected 99% of DDI cases, and Claude and Gemini detected 95%. The percentage of correct answers to all DDI-related questions was 66% for ChatGPT, 68% for Claude, and 33% for Gemini. ChatGPT and Claude produced comparable results and outperformed Gemini (F-measure between 0.83 and 0.85 for ChatGPT and Claude and 0.63-0.68 for Gemini) to detect drugs involved in DDI. All exhibited low specificity (ChatGPT 0.68, Claude 0.64, and Gemini 0.36) and reported nonexistent DDIs for negative controls.

CONCLUSION: LLMs can detect DDIs leading to pharmacovigilance cases, but cannot reliably exclude DDIs in cases without interactions. Pharmacologists are crucial for assessing whether a DDI is implicated in an ADR.

PMID:40385316 | PMC:PMC12084699 | DOI:10.1177/20420986251339358

Drug Dev Res. 2025 May;86(3):e70101. doi: 10.1002/ddr.70101.

ABSTRACT

The global antibiotic resistance crisis demands innovative strategies targeting bacterial virulence rather than survival. Quorum sensing (QS), a key regulator of virulence and biofilm formation, offers a promising avenue to mitigate resistance by disarming pathogens without bactericidal pressure. This study investigates the repurposing of nitroimidazoles as anti-QS and anti-virulence agents at subminimum inhibitory concentrations (sub-MICs). In Silico analyses, including molecular docking and molecular dynamics (MD) simulations, were performed to investigate ligand-receptor interactions with structurally distinct Lux-type QS receptors and assess binding stability and conformational dynamics over time. In Vitro assays evaluated the effects of representative nitroimidazoles, metronidazole (MET) and secnidazole (SEC), on QS-controlled phenotypes, including violacein production in Chromobacterium violaceum and biofilm formation and protease activity in Pseudomonas aeruginosa, Acinetobacter baumannii, Salmonella enterica, and Proteus mirabilis. In Vivo efficacy was assessed using a murine infection model and HeLa cell invasion assays. Molecular docking revealed high-affinity binding to QS receptors, corroborating their mechanistic interference. Sub-MIC MET/SEC significantly suppressed violacein synthesis, biofilm biomass, and protease secretion in Gram-negative pathogens. Both compounds reduced bacterial invasiveness in HeLa cells and In Vivo protected mice from lethal P. aeruginosa infections. Crucially, nitroimidazoles attenuated virulence without affecting bacterial viability, preserving microbial ecology. These findings position nitroimidazoles as dual-function agents; antimicrobial at bactericidal doses and anti-virulence at sub-MICs. Their validated efficacy across In Silico, In Vitro, and In Vivo models underscores their potential as adjunctive therapies, bridging the gap between drug repurposing and next-generation anti-infective development.

PMID:40384051 | DOI:10.1002/ddr.70101

Integr Cancer Ther. 2025 Jan-Dec;24:15347354251339121. doi: 10.1177/15347354251339121. Epub 2025 May 18.

ABSTRACT

Chemotherapy-induced peripheral neuropathy (CIPN) has a markedly deleterious impact on a patient's quality of life. It manifests as pain, paresthesia, numbness, and weakness, particularly in the context of cisplatin (CDDP), a widely utilised chemotherapeutic agent renowned for its pronounced peripheral nerve toxicity. Trichosanthes kirilowii Maxim. (Cucurbitaceae, TK) and cucurbitacin D(CucD), its bioactive compound, have been demonstrated to possess anti-tumour, anti-inflammatory, and antioxidant properties. However, their potential to alleviate CIPN has not been fully exploredyet. The present study evaluated effectiveness of TK and CucD in mitigating CDDP-induced neuropathic pain using both cellular and animal models. CDDP, TK extracts (TKD and TKE), and CucD dose-dependently reduced viability and apoptosis of PC12 cells. Conversely, pre-treatment with TKD, TKE, and CucD exhibited significant protective effects against CDDP-induced cytotoxicity, preserving cell viability and morphology while enhancing neurite outgrowth. In vivo, administration of CDDP resulted in the development of mechanical allodynia and thermalhyperalgesia in rats. However, treatment with TKD and TKE led to a notable improvement in pain threshold and a reduction in hyperalgesia, while CucD demonstrated less pronounced effects. Although body weight was reduced in the CDDP-treated group, it was not significantly mitigated bytreatments. In conclusion, results of this study indicate that TKD, TKE, and CucD have the potential to alleviate CDDP-induced neuropathic pain by protecting against cell damage, promoting neuriteregeneration, and improving pain responses in animal models. Further investigation into TK and CucD as therapeutic options for managing CIPN is warranted.

PMID:40383960 | DOI:10.1177/15347354251339121

Obesity (Silver Spring). 2025 May 19. doi: 10.1002/oby.24300. Online ahead of print.

ABSTRACT

OBJECTIVE: The objective of this study is to identify whether the glucagon-like peptide-1 receptor (GLP1R) gene variant rs6923761G→A has an influence on semaglutide response in individuals with severe obesity.

METHODS: From March 2023 to July 2024, we prospectively genotyped 112 patients treated with semaglutide 2.4 mg weekly. All patients had been treated over 4 months for grade 3 obesity (BMI ≥ 40 kg/m2).

RESULTS: The frequency of the rs6923761 AA variant was 9 out of 112 patients (8%), GA was 42 out of 112 (37.5%), and GG was 61 out of 112 (54.5%). The mean weight loss kinetics was 1.64% (SD 0.78%) per month in homozygotes of variant A in comparison with a mean weight loss of 1.04% (SD 0.79%) per month in carriers of at least one G variant (p = 0.03). Multivariate analysis demonstrated that rs6923761G→A and sex were independent predictors of weight loss. The rate of weight loss in women homozygous for the A allele was more than double that observed in men carrying the G allele: mean (SD) 1.89% (0.75%) per month versus 0.7% (0.7%) per month (p = 0.0009). No woman homozygous for the A allele was a nonresponder, compared with 56% (21 out of 37) of the men carrying the G allele.

CONCLUSIONS: The rs6923761G→A gene variant and sex profoundly affect weight loss in response to semaglutide in patients with severe obesity.

PMID:40384505 | DOI:10.1002/oby.24300

Neurourol Urodyn. 2025 May 19. doi: 10.1002/nau.70057. Online ahead of print.

ABSTRACT

OBJECTIVE: We aimed to prospectively investigate whether bladder volume measured using deep learning artificial intelligence (AI) algorithms (AI-BV) is more accurate than that measured using conventional methods (C-BV) if using a portable ultrasound bladder scanner (PUBS).

PATIENTS AND METHODS: Patients who underwent filling cystometry because of lower urinary tract symptoms between January 2021 and July 2022 were enrolled. Every time the bladder was filled serially with normal saline from 0 mL to maximum cystometric capacity in 50 mL increments, C-BV was measured using PUBS. Ultrasound images obtained during this process were manually annotated to define the bladder contour, which was used to build a deep learning AI model. The true bladder volume (T-BV) for each bladder volume range was compared with C-BV and AI-BV for analysis.

RESULTS: We enrolled 250 patients (213 men and 37 women), and a deep learning AI model was established using 1912 bladder images. There was a significant difference between C-BV (205.5 ± 170.8 mL) and T-BV (190.5 ± 165.7 mL) (p = 0.001), but no significant difference between AI-BV (197.0 ± 161.1 mL) and T-BV (190.5 ± 165.7 mL) (p = 0.081). In bladder volume ranges of 101-150, 151-200, and 201-300 mL, there were significant differences in the percentage of volume differences between [C-BV and T-BV] and [AI-BV and T-BV] (p < 0.05), but no significant difference if converted to absolute values (p > 0.05). C-BV (R2 = 0.91, p < 0.001) and AI-BV (R2 = 0.90, p < 0.001) were highly correlated with T-BV. The mean difference between AI-BV and T-BV (6.5 ± 50.4) was significantly smaller than that between C-BV and T-BV (15.0 ± 50.9) (p = 0.001).

CONCLUSION: Following image pre-processing, deep learning AI-BV more accurately estimated true BV than conventional methods in this selected cohort on internal validation. Determination of the clinical relevance of these findings and performance in external cohorts requires further study.

TRIAL REGISTRATION: The clinical trial was conducted using an approved product for its approved indication, so approval from the Ministry of Food and Drug Safety (MFDS) was not required. Therefore, there is no clinical trial registration number.

PMID:40384598 | DOI:10.1002/nau.70057

Analyst. 2025 May 19. doi: 10.1039/d5an00253b. Online ahead of print.

ABSTRACT

Raman spectroscopy requires baseline correction to address fluorescence- and instrumentation-related distortions. The existing baseline correction methods can be broadly classified into traditional mathematical approaches and deep learning-based techniques. While traditional methods often require manual parameter tuning for different spectral datasets, deep learning methods offer greater adaptability and enhance automation. Recent research on deep learning-based baseline correction has primarily focused on optimizing existing methods or designing new network architectures to improve correction performance. This study proposes a novel deep learning network architecture to further enhance baseline correction effectiveness, building upon prior research. Experimental results demonstrate that the proposed method outperforms existing approaches by achieving superior correction accuracy, reducing computation time, and more effectively preserving peak intensity and shape.

PMID:40384579 | DOI:10.1039/d5an00253b

J Magn Reson Imaging. 2025 May 19. doi: 10.1002/jmri.29819. Online ahead of print.

ABSTRACT

BACKGROUND: Deep learning (DL) models for accurate renal tumor characterization may benefit from multi-center datasets for improved generalizability; however, data-sharing constraints necessitate privacy-preserving solutions like federated learning (FL).

PURPOSE: To assess the performance and reliability of FL for renal tumor segmentation and classification in multi-institutional MRI datasets.

STUDY TYPE: Retrospective multi-center study.

POPULATION: A total of 987 patients (403 female) from six hospitals were included for analysis. 73% (723/987) had malignant renal tumors, primarily clear cell carcinoma (n = 509). Patients were split into training (n = 785), validation (n = 104), and test (n = 99) sets, stratified across three simulated institutions.

FIELD STRENGTH/SEQUENCE: MRI was performed at 1.5 T and 3 T using T2-weighted imaging (T2WI) and contrast-enhanced T1-weighted imaging (CE-T1WI) sequences.

ASSESSMENT: FL and non-FL approaches used nnU-Net for tumor segmentation and ResNet for its classification. FL-trained models across three simulated institutional clients with central weight aggregation, while the non-FL approach used centralized training on the full dataset.

STATISTICAL TESTS: Segmentation was evaluated using Dice coefficients, and classification between malignant and benign lesions was assessed using accuracy, sensitivity, specificity, and area under the curves (AUCs). FL and non-FL performance was compared using the Wilcoxon test for segmentation Dice and Delong's test for AUC (p < 0.05).

RESULTS: No significant difference was observed between FL and non-FL models in segmentation (Dice: 0.43 vs. 0.45, p = 0.202) or classification (AUC: 0.69 vs. 0.64, p = 0.959) on the test set. For classification, no significant difference was observed between the models in accuracy (p = 0.912), sensitivity (p = 0.862), or specificity (p = 0.847) on the test set.

DATA CONCLUSION: FL demonstrated comparable performance to non-FL approaches in renal tumor segmentation and classification, supporting its potential as a privacy-preserving alternative for multi-institutional DL models.

EVIDENCE LEVEL: 4.

TECHNICAL EFFICACY: Stage 2.

PMID:40384349 | DOI:10.1002/jmri.29819

Med Phys. 2025 May 19. doi: 10.1002/mp.17895. Online ahead of print.

ABSTRACT

BACKGROUND: Microvascular invasion (MVI) is strongly associated with the prognosis of patients with hepatocellular carcinoma (HCC).

PURPOSE: To evaluate the value of Transformer models with Sonazoid contrast-enhanced ultrasound (CEUS) in the preoperative prediction of MVI.

METHODS: This retrospective study included 164 HCC patients. Deep learning features and radiomic features were extracted from arterial and Kupffer phase images, alongside the collection of clinicopathological parameters. Normality was assessed using the Shapiro-Wilk test. The Mann‒Whitney U-test and least absolute shrinkage and selection operator algorithm were applied to screen features. Transformer, radiomic, and clinical prediction models for MVI were constructed with logistic regression. Repeated random splits followed a 7:3 ratio, with model performance evaluated over 50 iterations. The area under the receiver operating characteristic curve (AUC, 95% confidence interval [CI]), sensitivity, specificity, accuracy, positive predictive value (PPV), negative predictive value (NPV), decision curve, and calibration curve were used to evaluate the performance of the models. The DeLong test was applied to compare performance between models. The Bonferroni method was used to control type I error rates arising from multiple comparisons. A two-sided p-value of < 0.05 was considered statistically significant.

RESULTS: In the training set, the diagnostic performance of the arterial-phase Transformer (AT) and Kupffer-phase Transformer (KT) models were better than that of the radiomic and clinical (Clin) models (p < 0.0001). In the validation set, both the AT and KT models outperformed the radiomic and Clin models in terms of diagnostic performance (p < 0.05). The AUC (95% CI) for the AT model was 0.821 (0.72-0.925) with an accuracy of 80.0%, and the KT model was 0.859 (0.766-0.977) with an accuracy of 70.0%. Logistic regression analysis indicated that tumor size (p = 0.016) and alpha-fetoprotein (AFP) (p = 0.046) were independent predictors of MVI.

CONCLUSIONS: Transformer models using Sonazoid CEUS have potential for effectively identifying MVI-positive patients preoperatively.

PMID:40384312 | DOI:10.1002/mp.17895

J Phys Chem Lett. 2025 May 18:5244-5251. doi: 10.1021/acs.jpclett.5c00592. Online ahead of print.

ABSTRACT

Machine learning (ML) approaches have become ubiquitous in the search for new materials in recent years. Bayesian optimization (BO) based on Gaussian processes (GPs) has become a widely recognized approach in material exploration. However, feature engineering has critical impacts on the efficiency of GP-based BO, because GPs cannot automatically generate descriptors. To address this limitation, this study applies deep kernel learning (DKL), which combines a neural network with a GP, to BO. The efficiency of the DKL model was comparable to or significantly better than that of a standard GP in a data set of 922 oxide data sets, covering band gaps, ionic dielectric constants, and effective masses of electrons, as well as in experimental data sets, the band gaps of 610 hybrid organic-inorganic perovskite alloys. When searching for the alloy with the highest Curie temperature among 4560 alloys, the standard GP outperformed the DKL model because a strongly correlated descriptor of the Curie temperature could be directly utilized. Additionally, DKL supports transfer learning, which further enhances its efficiency. Thus, we believe that BO based on DKL paves the way for exploring diverse material spaces more effectively than GPs.

PMID:40383929 | DOI:10.1021/acs.jpclett.5c00592

Protein Sci. 2025 Jun;34(6):e70155. doi: 10.1002/pro.70155.

ABSTRACT

The complex genotype-to-phenotype relationships in Mendelian diseases can be elucidated by mutation-induced disturbances to the networks of molecular interactions (interactomes) in human cells. Missense and nonsense mutations cause distinct perturbations within the human protein interactome, leading to functional and phenotypic effects with varying degrees of severity. Here, we structurally resolve the human protein interactome at atomic-level resolutions and perform structural and thermodynamic calculations to assess the biophysical implications of these mutations. We focus on a specific type of missense mutation, known as "quasi-null" mutations, which destabilize proteins and cause similar functional consequences (node removal) to nonsense mutations. We propose a "fold difference" quantification of deleteriousness, which measures the ratio between the fractions of node-removal mutations in datasets of Mendelian disease-causing and non-pathogenic mutations. We estimate the fold differences of node-removal mutations to range from 3 (for quasi-null mutations with folding ΔΔG ≥2 kcal/mol) to 20 (for nonsense mutations). We observe a strong positive correlation between biophysical destabilization and phenotypic deleteriousness, demonstrating that the deleteriousness of quasi-null mutations spans a continuous spectrum, with nonsense mutations at the extreme (highly deleterious) end. Our findings substantiate the disparity in phenotypic severity between missense and nonsense mutations and suggest that mutation-induced protein destabilization is indicative of the phenotypic outcomes of missense mutations. Our analyses of node-removal mutations allow for the potential identification of proteins whose removal or destabilization lead to harmful phenotypes, enabling the development of targeted therapeutic approaches, and enhancing comprehension of the intricate mechanisms governing genotype-to-phenotype relationships in clinically relevant diseases.

PMID:40384578 | DOI:10.1002/pro.70155

Nucleic Acids Res. 2025 May 19:gkaf406. doi: 10.1093/nar/gkaf406. Online ahead of print.

ABSTRACT

Base editing enables the high-throughput screening of genetic variants for phenotypic effects. Base editing screens require the design of single guide RNA (sgRNA) libraries to enable either gene- or variant-centric approaches. While computational tools supporting the design of sgRNAs exist, no solution offers versatile and scalable library design enabling all major use cases. Here, we introduce BEscreen, a comprehensive base editing guide design tool provided as a web server (bescreen.ostendorflab.org) and as a command line tool. BEscreen provides variant-, gene-, and region-centric modes to accommodate various screening approaches. The variant mode accepts genomic coordinates, amino acid changes, or rsIDs as input. The gene mode designs near-saturation libraries covering the entire coding sequence of given genes or transcripts, and the region mode designs all possible guides for given genomic regions. BEscreen enables selection of guides by biological consequence, it features comprehensive customization of base editor characteristics, and it offers optional annotation using Ensembl's Variant Effect Predictor. In sum, BEscreen is a highly versatile tool to design base editing screens for a wide range of use cases with seamless scalability from individual variants to large, near-saturation libraries.

PMID:40384567 | DOI:10.1093/nar/gkaf406