Biodata Mining最新文献

The endothelial tube formation assay is an established in vitro model for evaluating angiogenesis. Although widely used, quantification of angiogenic behavior in such assays remains semi-empirical and often lacks spatial, topological, and structural context. Here, we present a graph-theoretic framework to quantify network morphology, temporal dynamics, and spatial heterogeneity in tube formation assays. We simulated two distinct angiogenic network morphologies using human umbilical vein endothelial cells (HUVECs) seeded at two densities and imaged at 2, 4, and 18 h post-seeding. Skeletonized images were converted to mathematical graphs from which 11 graph-based metrics were extracted. This framework captured both morphological differences and temporal progression. Sparse networks exhibited significantly higher average node degree (p = 0.00079), clustering coefficient (p = 0.00109), and tortuosity (p = 0.0171), whereas dense networks showed greater node and edges counts (p = 0.00109). Over time, networks evolved from fragmented forms at 2 h to integrated structures at 18 h, as reflected by increased largest component size (p = 0.00216), connectivity index (p = 0.00216), and efficiency (p = 0.0152). ROC AUC analysis revealed that metrics such as average degree (AUC = 0.98) and clustering coefficient (AUC = 0.96) effectively distinguished between sparse and dense morphologies, while component-based metrics perfectly separated 2- and 18-hour networks (AUC = 1.00). Radial zone analysis revealed that vascular distribution becomes more compartmentalized over time, with increasing standard deviation and coefficient of variation. This approach provides a sensitive and scalable method for quantifying angiogenic dynamics, offering insight into both therapeutic efficacy and disease-related vascular remodeling.

Traditional radiomics approaches focus on single-region feature extraction, limiting their ability to capture complex inter-regional relationships crucial for understanding pathophysiological mechanisms in complex diseases. This study introduces a novel cross-regional radiomics framework that systematically extracts relationship-based features between anatomically and functionally connected brain regions. We analyzed T1-weighted magnetic resonance imaging (MRI) data from 140 early-stage Parkinson's disease patients (70 tremor-dominant, 70 postural instability gait difficulty) from the Parkinson's Progression Markers Initiative (PPMI) database across multiple imaging centers. Eight bilateral motor circuit regions (putamen, caudate nucleus, globus pallidus, substantia nigra) were segmented using standardized atlases. Two feature sets were developed: 48 traditional single-region of interest (ROI) features and 60 novel motor-circuit features capturing cross-regional ratios, asymmetry indices, volumetric relationships, and shape distributions. Six feature engineering scenarios were evaluated using center-based 5-fold cross-validation with six machine learning classifiers to ensure robust generalization across different imaging centers. Motor-circuit features demonstrated superior performance compared to single-ROI features across enhanced preprocessing scenarios. Peak performance was achieved with area under the curve (AUC) of 0.821 ± 0.117 versus 0.650 ± 0.220 for single-ROI features (p = 0.0012, Cohen's d = 0.665). Cross-regional ratios, particularly putamen-substantia nigra relationships, dominated the most discriminative features. Motor-circuit features showed superior generalization across multi-center data and better clinical utility through decision curve analysis and calibration curves. The proposed cross-regional radiomics framework significantly outperforms traditional single-region approaches for Parkinson's disease motor subtype classification. This methodology provides a foundation for advancing radiomics applications in complex diseases where inter-regional connectivity patterns are fundamental to pathophysiology.

Yersinia enterocolitica infection can manifest as self-limiting gastroenteritis and may lead to more severe conditions, such as mesenteric lymphadenitis, reactive arthritis, or rare systemic infections. Fluoroquinolones and third-generation cephalosporins are the most effective treatment options but tetracyclines and co-trimoxazole effectiveness may vary based on resistance patterns. To explore new therapeutic options in case of antibiotic resistance, we initially mined drug targets from the Yersinia enterocolitica proteome using a subtractive proteomics approach. Subsequently, we repurposed FDA approved & Traditional Chinese Medicinal (TCM) compounds against its cell wall synthesis mechanism by targeting DD-transpeptidase. DrugRep screening prioritized FDA-approved hits (Digitoxin, Irinotecan, Acetyldigitoxin; ≤ -9.4 kcal/mol) and TCM hits (Vaccarin, Narirutin, Hinokiflavone; ≤ -9.5 kcal/mol). Machine learning-based validation identified Hinokiflavone and Acetyldigitoxin as most potent binders. Molecular dynamics simulations (100 ns) revealed RMSD values < 1 nm for all complexes, indicating stable binding. ADMET profiling predicted all compounds as non-allergenic and TCM compounds having poor absorption. SBE-β-cyclodextrin coupling with FormulationAI showed improved compound solubility and oral bioavailability. InflamNat predicted strong anti-inflammatory potential for Hinokiflavone, highlighting its dual role in antibacterial and host-directed immunomodulatory activity. These computational insights mark an initial step in drug discovery, prompting comprehensive testing of prioritized compounds against Yersinia enterocolitica.

Background: The clinical potential of single nucleotide polymorphisms (SNPs) in prostate cancer (PCa) diagnosis has been extensively explored using conventional statistical and machine learning approaches. However, the predictive power and interpretability of these methods remain inadequate for clinical translation, primarily due to limited generalization across high-dimensional SNP datasets. This study addresses the contested diagnostic utility of SNPs by integrating interpretable feature selection with deep learning to enhance both classification performance and biological relevance.

Methods: We propose an interpretable deep feature selection framework designed to enhance both the classification performance and biological relevance of SNP markers in distinguishing between benign and malignant prostate cancer samples. This study specifically investigates the debated diagnostic value of SNPs in PCa classification by integrating feature selection with deep learning to uncover actionable insights. Specifically, our framework comprises four key components: (1) Heuristic feature reduction, which eliminates irrelevant SNPs during gradient computation for training deep neural networks (DNNs); (2) Iterative SNP subset optimization, aiming at maximizing classification AUC during model training; (3) Gradient variance minimization, mitigating instability caused by limited sample sizes; and (4) Nonlinear interaction modeling, which extracts high-level SNP interactions through hierarchical representations.

Results: Evaluated on the PLCO, BPC3, and MEC-AA datasets, our method achieved mean AUC scores of 0.747, 0.751, and 0.559, respectively, demonstrating statistically significant improvements (p < 0.05, a paired t-test) over existing approaches. Notably, the lower AUC for MEC-AA may reflect inherent population-specific complexities, as this dataset focuses on African American men, a group historically underrepresented in genomic studies. For interpretability, our framework identified 345, 373, and 437 consensus SNP markers across the PLCO, BPC3, and MEC-AA cohorts, respectively. Key SNPs were further validated against prior research on PCa racial disparities: rs10086908 and rs2273669 (PLCO); rs12284087, rs902774, rs9364554, and rs7611694 (BPC3); and rs3123078 and rs1447295 (MEC-AA) exhibited strong concordance with established loci linked to ethnic-specific risk profiles. For instance, rs1447295 on chromosome 8q24, recurrently associated with African ancestry, underscores the method's ability to recover population-relevant variants.

Conclusion: By synergizing interpretable feature selection with deep learning, this work advances the translation of SNP-based biomarkers into clinically actionable tools while clarifying their contested diagnostic role in PCa.

Molecular recognition features (MoRFs) can facilitate specific protein-protein interactions by undergoing disorder-to-order transitions when binding to their protein partners. Thus, it is essential to accurately predict MoRFs. In this paper, we propose an innovative MoRFs prediction method, named MoRFs_TransFuse, based on multimodal feature fusion and a lightweight Transformer network. To construct high-quality biological features, MoRFs_TransFuse innovatively integrates physicochemical properties, evolutionary features, and pre-trained model embeddings, while retaining optimal feature combinations through multi-window extraction and Random Forest secondary screening. In terms of architecture, MoRFs_TransFuse overcomes the limitations of modeling long-range dependencies by using a self-attention mechanism to accurately capture long-range residue associations in protein sequences. Comparative experiments on benchmark datasets show that MoRFs_TransFuse significantly outperforms existing single component and combined component predictors. Additionally, the lightweight design greatly improves computational efficiency while ensuring prediction accuracy.

Sepsis is a critical medical condition characterized by a highly variable and rapidly evolving clinical course, often necessitating early intervention and tailored treatment plans to improve patient outcomes. Due to its complexity and heterogeneity, understanding the progression of sepsis across different patient populations remains a significant challenge. In this study, we exploit a sophisticated analytical framework based on k-means multivariate longitudinal clustering to capture the diverse trajectories of sepsis. We do so by analyzing multiple clinical parameters tracked over time, providing a nuanced view of disease progression. By incorporating Dynamic Time Warping (DTW) as the distance metric, the proposed method effectively accounts for temporal misalignments and variability in the rate of disease progression, an essential capability given the unpredictable and heterogeneous nature of sepsis. This integration enhances the model's ability to detect distinct temporal patterns and phenotypic subgroups that may remain undetected using conventional analytical approaches. By leveraging sepsis-related electronic health records (EHRs), which provide rich time-series data on laboratory results along with patient demographics and underlying health conditions, the proposed method reveals distinct sepsis phenotypes that reflect variations in disease progression. We perform several experiments varying the number of clusters and clinical variable combinations, evaluating the clustering performances using Silhouette score, Caliski-Harabasz Index, and Davies-Bouldin Index, as reference quality metrics. Our results confirm the prognostic role of the Thrombin-Antigen complex and the Prothrombin Time-International Normalized Ratio for septic patients. Furthermore, to evaluate the relevance of subjects' stratification, the Adjusted Rand Index metric is used to quantify the survival prediction capability of our longitudinal clustering method, considering the 28-day death feature as the target variable. The same metric demonstrates that our proposal outperforms other longitudinal clustering algorithms available in the literature.

Background: Early 30-day readmission remains a significant burden on the socioeconomic and healthcare system in the context of decompensated cirrhosis. Early recognition and accurate identification are crucial. However, current evidence is elusive and traditional scores concerning liver disease severity are lacking specificity and sensitivity. We sought to construct and validate an explainable machine learning (ML)-based prediction model, and evaluate its prognostic implementation in patients readmitted due to acute episodes. The prediction model for discovery and validation was based on a two-center prospective investigation. Our discovery sample, comprising 636 patients with cirrhosis, was divided into a training set and a test set, with an additional cohort of 150 patients serving as an external validation. Eleven ML methods were performed to establish an indicative model based on a variety of easily accessible and obtainable variables from the electronic health record. The area under the ROC curve (AUC), alongside several evaluation parameters, was used for comparison regarding predictive performance. Considering feature importance and final model explanation, we adopted the SHapley Additive exPlanation method for ranking. Furthermore, prognostic implementation was verified by subgrouping according to the final model and clinical outcomes during follow-up.

Results: Among all 11 ML algorithms, the random forest (RF) algorithm represented the best discriminatory capability. Processing feature reduction generated a final 7-feature RF model with explainability based on the importance ranking. Our constructed model was of moderately accurate prediction pertaining to internal and external validations, with respective AUCs of 0.853 and 0.838, which was further transformed into an online tool to facilitate daily practice. Patients positively adjudged by the prediction model had aggravated underlying disease severity and poor psychophysiologic reservation.

Conclusions: The final explainable ML model was capable of predicting early readmission and was closely connected with adverse outcomes in individual patients experiencing decompensated cirrhosis. Notably, it allayed the "black-box" concerns inherent to ML techniques with an indirect interpretation.

Background: Disrupted cholesterol homeostasis may accelerate cognitive aging. This study investigated the relationship between serum HDL-C levels and cognitive function, utilizing cross-sectional data and Mendelian randomization (MR) analysis.

Methods: A cross-sectional study was conducted using data from the National Health and Nutrition Examination Survey (NHANES) 2011-2014, including 19,931 participants. Among them, 2,777 individuals aged 60 years and older with complete HDL-C levels and cognitive function data were included. Cognitive function was assessed using tests such as the Consortium to Establish a Registry for Alzheimer's Disease Immediate and Delayed Recall, the Animal Fluency Test, and the Digit Symbol Substitution Test. Additionally, MR analysis was employed to assess the causal relationship between genetically predicted HDL-C and dementia.

Results: Gender-stratified analyses revealed sex-specific patterns in the relationship between HDL-C and cognitive function. In fully adjusted linear models, men showed consistently positive associations across all cognitive domains, including delayed recall (β = 0.10, 95% CI 0.04-0.17, p < 0.001), immediate recall (β = 0.06, 95% CI 0.00-0.12, p = 0.047), verbal fluency (β = 0.20, 95% CI 0.14-0.26, p < 0.001), processing speed (β = 0.09, 95% CI 0.05-0.14, p < 0.001), and overall composite score (β = 0.45, 95% CI 0.29-0.62, p < 0.001). In women, these associations were attenuated or non-significant for immediate recall, delayed recall, and composite cognition, suggesting non-linearity. Further concentration-response analyses revealed a linear positive association in men and an inverted U-shaped relationship in women. MR analyses indicated a protective association between genetically predicted HDL-C and Alzheimer's disease risk (OR = 0.51, 95% CI 0.29-0.89, p = 0.019). However, sensitivity analyses revealed attenuation after MR-PRESSO outlier correction (β=-0.013, p = 0.756), and inconsistent estimates across methods, with significant heterogeneity (Q-test p < 0.001) and evidence of pleiotropy. In multivariable analysis, adjusting for LDL-C and TG, IVW (β = 0.290, p = 0.048) and Lasso regression (β = 0.752, p = 0.008) indicated weak positive correlations. However, MR-Egger (β = 0.752, p = 0.008) revealed potential pleiotropic interference (intercept p = 0.050).

Conclusions: Our findings suggest that maintaining optimal serum HDL-C levels may help preserve cognitive function in older adults. Notably, sex-specific associations were observed, warranting further investigation into underlying mechanisms.

Background: The immune response to SARS-CoV-2 varies greatly among individuals yielding highly varying severity levels among the patients. While there are various methods to spot severity associated biomarkers in COVID-19 patients, we investigated highly mutated regions, or mutation hotspots, within the SARS-CoV-2 genome that correlate with patient severity levels. SARS-CoV-2 mutation hotspots were searched in the GISAID database using a density based clustering algorithm, Mutclust, that searches for loci with high mutation density and diversity.

Results: Using Mutclust, 477 mutation hotspots were searched in the SARS-CoV-2 genome, of which 28 showed significant association with severity levels in a multi-omics COVID-19 cohort comprised of 387 infected patients. The patients were further stratified into moderate and severe patient groups based on the 28 severity related mutation hotspots that showed distinctive cytokine and gene expression levels in both cytokine profile and single-cell RNA-seq samples. The effect of the SARS-CoV-2 mutation hotspots on human genes was further investigated by network propagation analysis, where two mutation hotspots specific to the severe group showed association with NK cell activity. One of them showed to decrease the affinity between the viral epitope of the hotspot region and its binding HLA when compared to the non-mutated epitope.

Conclusion: Genes related to the immunological function of NK cells, especially the NK cell receptor and co-activating receptor genes, were significantly dysregulated in the severe patient group in both cytokine and single-cell levels. Collectively, mutation hotspots associated with severity and their related NK cell associated gene expression regulation were identified.

Class imbalance poses a serious challenge in biomedical machine learning, particularly in genomics, where datasets are characterized by extremely high dimensionality and very limited sample sizes. In such settings, standard classifiers tend to favor the majority class, leading to biased predictions - an especially problematic issue in clinical diagnostics where rare conditions must not be overlooked. In this study, we introduce a Kernel Density Estimation (KDE)-based oversampling approach to rebalance imbalanced genomic datasets by generating synthetic minority class samples. Unlike conventional methods such as SMOTE, KDE estimates the global probability distribution of the minority class and resamples accordingly, avoiding local interpolation pitfalls. We evaluate our method on 15 real-world genomic datasets using three classifiers -Naïve Bayes, Decision Trees, and Random Forests- and compare it to SMOTE and baseline training. Experimental results demonstrate that KDE oversampling consistently improves classification performance, especially in metrics robust to imbalance, such as AUC of the IMCP curve. Notably, KDE achieves superior results in tree-based models while dramatically simplifying the sampling process. This approach offers a statistically grounded and effective solution for balancing genomic datasets, with strong potential for improving fairness and accuracy in high-stakes medical decision-making.