Multi-scale variational autoencoder for imputation of missing values in untargeted metabolomics using whole-genome sequencing data.

BACKGROUND: Missing data is a common challenge in mass spectrometry-based metabolomics, which can lead to biased and incomplete analyses. The integration of whole-genome sequencing (WGS) data with metabolomics data has emerged as a promising approach to enhance the accuracy of data imputation in metabolomics studies. METHOD: In this study, we propose a novel method that leverages the information from WGS data and reference metabolites to impute unknown metabolites. Our approach utilizes a multi-scale variational autoencoder to jointly model the burden score, polygenetic risk score (PGS), and linkage disequilibrium (LD) pruned single nucleotide polymorphisms (SNPs) for feature extraction and missing metabolomics data imputation. By learning the latent representations of both omics data, our method can effectively impute missing metabolomics values based on genomic information. RESULTS: We evaluate the performance of our method on empirical metabolomics datasets with missing values and demonstrate its superiority compared to conventional imputation techniques. Using 35 template metabolites derived burden scores, PGS and LD-pruned SNPs, the proposed methods achieved R2-scores > 0.01 for 71.55 % of metabolites. CONCLUSION: The integration of WGS data in metabolomics imputation not only improves data completeness but also enhances downstream analyses, paving the way for more comprehensive and accurate investigations of metabolic pathways and disease associations. Our findings offer valuable insights into the potential benefits of utilizing WGS data for metabolomics data imputation and underscore the importance of leveraging multi-modal data integration in precision medicine research.

Dissecting the shared genetic architecture between anti-Müllerian hormone and age at menopause based on genome-wide association study.

BACKGROUND: While the phenotypic association between anti-Müllerian hormone (AMH) and age at menopause has been widely studied, the role of AMH in predicting the age at menopause is currently controversial, and the genetic architecture or causal relationships underlying these two traits is not well understood. AIM: We aimed to explore the shared genetic architecture between AMH and age at menopause, to identify shared pleiotropic loci and genes, and to investigate causal association and potential causal mediators. STUDY DESIGN: Using summary statistics from publicly available genome-wide association studies on AMH (N=7,049) and age at menopause (N=201,323) in Europeans, we investigated the global genetic architecture between AMH and age at menopause through linkage disequilibrium score regression. We employed pleiotropic analysis under composite null hypothesis (PLACO), Functional Mapping and Annotation of Genetic Associations (FUMA), Multimarker analysis of GenoMic annotation (MAGMA), and colocalization analysis to identify loci and genes with pleiotropic effects. Tissue enrichment analysis based on GTEx data was conducted using the Linkage Disequilibrium Score for the specific expression of genes analysis (LDSC-SEG). Functional genes that were shared were additionally identified through summary data-based Mendelian randomization (SMR). The relationship between AMH and age at menopause was examined through two-sample Mendelian randomization (MR), and potential mediators were further explored using colocalization and metabolite-mediated analysis. RESULTS: A positive genetic association (correlation coefficient = 0.88, P = 1.33 × 10-5) was observed between AMH and age at menopause. By using PLACO and FUMA, 42 significant pleiotropic loci were identified that were associated with AMH and age at menopause, and ten of these (rs10734411, rs61913600, rs2277339, rs75770066, rs28416520, rs9796, rs11668344, rs403727, rs6011452, and rs62237617) had colocalized loci. Additionally, 245 significant pleiotropic genes were identified by MAGMA. Genetic associations between AMH and age at menopause were markedly concentrated in various tissues including whole blood, brain, heart, liver, muscle, pancreas, and kidneys. Further, SMR analysis revealed nine genes that may have a causative effect on both AMH and age at menopause. A potential causal effect of age at menopause on AMH was suggested by two-sample MR analysis, with very-low-density lipoprotein identified as a potential mediator. CONCLUSIONS: Our study revealed a shared genetic architecture between AMH and age at menopause, providing a basis for experimental investigations and individual therapies to enhance reproductive outcomes. Furthermore, our findings emphasized that relying solely on AMH is not sufficient for accurately predicting the age at menopause, and a combination of other factors needs to be considered. Exploring new therapeutics aimed at delaying at the onset of menopause holds promise, particularly when targeting shared genes based on their shared genetic architecture.

MIMOSA: a resource consisting of improved methylome prediction models increases power to identify DNA methylation-phenotype associations.

Although DNA methylation (DNAm) has been implicated in the pathogenesis of numerous complex diseases, from cancer to cardiovascular disease to autoimmune disease, the exact methylation sites that play key roles in these processes remain elusive. One strategy to identify putative causal CpG sites and enhance disease etiology understanding is to conduct methylome-wide association studies (MWASs), in which predicted DNA methylation that is associated with complex diseases can be identified. However, current MWAS models are primarily trained using the data from single studies, thereby limiting the methylation prediction accuracy and the power of subsequent association studies. Here, we introduce a new resource, MWAS Imputing Methylome Obliging Summary-level mQTLs and Associated LD matrices (MIMOSA), a set of models that substantially improve the prediction accuracy of DNA methylation and subsequent MWAS power through the use of a large summary-level mQTL dataset provided by the Genetics of DNA Methylation Consortium (GoDMC). Through the analyses of GWAS (genome-wide association study) summary statistics for 28 complex traits and diseases, we demonstrate that MIMOSA considerably increases the accuracy of DNA methylation prediction in whole blood, crafts fruitful prediction models for low heritability CpG sites, and determines markedly more CpG site-phenotype associations than preceding methods. Finally, we use MIMOSA to conduct a case study on high cholesterol, pinpointing 146 putatively causal CpG sites.

A Staged Approach using Machine Learning and Uncertainty Quantification to Predict the Risk of Hip Fracture.

Hip fractures present a significant healthcare challenge, especially within aging populations, where they are often caused by falls. These fractures lead to substantial morbidity and mortality, emphasizing the need for timely surgical intervention. Despite advancements in medical care, hip fractures impose a significant burden on individuals and healthcare systems. This paper focuses on the prediction of hip fracture risk in older and middle-aged adults, where falls and compromised bone quality are predominant factors. We propose a novel staged model that combines advanced imaging and clinical data to improve predictive performance. By using convolutional neural networks (CNNs) to extract features from hip DXA images, along with clinical variables, shape measurements, and texture features, our method provides a comprehensive framework for assessing fracture risk. The study cohort included 547 patients, with 94 experiencing hip fracture. A staged machine learning-based model was developed using two ensemble models: Ensemble 1 (clinical variables only) and Ensemble 2 (clinical variables and DXA imaging features). This staged approach used uncertainty quantification from Ensemble 1 to decide if DXA features are necessary for further prediction. Ensemble 2 exhibited the highest performance, achieving an Area Under the Curve (AUC) of 0.9541, an accuracy of 0.9195, a sensitivity of 0.8078, and a specificity of 0.9427. The staged model also performed well, with an AUC of 0.8486, an accuracy of 0.8611, a sensitivity of 0.5578, and a specificity of 0.9249, outperforming Ensemble 1, which had an AUC of 0.5549, an accuracy of 0.7239, a sensitivity of 0.1956, and a specificity of 0.8343. Furthermore, the staged model suggested that 54.49% of patients did not require DXA scanning. It effectively balanced accuracy and specificity, offering a robust solution when DXA data acquisition is not always feasible. Statistical tests confirmed significant differences between the models, highlighting the advantages of the advanced modeling strategies. Our staged approach offers a cost-effective holistic view of patients' health. It could identify individuals at risk with a high accuracy but reduce the unnecessary DXA scanning. Our approach has great promise to guide interventions to prevent hip fractures with reduced cost and radiation.

Developing and comparing deep learning and machine learning algorithms for osteoporosis risk prediction.

Introduction: Osteoporosis, characterized by low bone mineral density (BMD), is an increasingly serious public health issue. So far, several traditional regression models and machine learning (ML) algorithms have been proposed for predicting osteoporosis risk. However, these models have shown relatively low accuracy in clinical implementation. Recently proposed deep learning (DL) approaches, such as deep neural network (DNN), which can discover knowledge from complex hidden interactions, offer a new opportunity to improve predictive performance. In this study, we aimed to assess whether DNN can achieve a better performance in osteoporosis risk prediction. Methods: By utilizing hip BMD and extensive demographic and routine clinical data of 8,134 subjects with age more than 40 from the Louisiana Osteoporosis Study (LOS), we developed and constructed a novel DNN framework for predicting osteoporosis risk and compared its performance in osteoporosis risk prediction with four conventional ML models, namely random forest (RF), artificial neural network (ANN), k-nearest neighbor (KNN), and support vector machine (SVM), as well as a traditional regression model termed osteoporosis self-assessment tool (OST). Model performance was assessed by area under 'receiver operating curve' (AUC) and accuracy. Results: By using 16 discriminative variables, we observed that the DNN approach achieved the best predictive performance (AUC = 0.848) in classifying osteoporosis (hip BMD T-score ≤ -1.0) and non-osteoporosis risk (hip BMD T-score > -1.0) subjects, compared to the other approaches. Feature importance analysis showed that the top 10 most important variables identified by the DNN model were weight, age, gender, grip strength, height, beer drinking, diastolic pressure, alcohol drinking, smoke years, and economic level. Furthermore, we performed subsampling analysis to assess the effects of varying number of sample size and variables on the predictive performance of these tested models. Notably, we observed that the DNN model performed equally well (AUC = 0.846) even by utilizing only the top 10 most important variables for osteoporosis risk prediction. Meanwhile, the DNN model can still achieve a high predictive performance (AUC = 0.826) when sample size was reduced to 50% of the original dataset. Conclusion: In conclusion, we developed a novel DNN model which was considered to be an effective algorithm for early diagnosis and intervention of osteoporosis in the aging population.

Spatial Dissection of the Distinct Cellular Responses to Normal Aging and Alzheimer's Disease in Human Prefrontal Cortex at Single-Nucleus Resolution.

Aging significantly elevates the risk for Alzheimer's disease (AD), contributing to the accumulation of AD pathologies, such as amyloid-ß (Aß), inflammation, and oxidative stress. The human prefrontal cortex (PFC) is highly vulnerable to the impacts of both aging and AD. Unveiling and understanding the molecular alterations in PFC associated with normal aging (NA) and AD is essential for elucidating the mechanisms of AD progression and developing novel therapeutics for this devastating disease. In this study, for the first time, we employed a cutting-edge spatial transcriptome platform, STOmics® SpaTial Enhanced Resolution Omics-sequencing (Stereo-seq), to generate the first comprehensive, subcellular resolution spatial transcriptome atlas of the human PFC from six AD cases at various neuropathological stages and six age, sex, and ethnicity matched controls. Our analyses revealed distinct transcriptional alterations across six neocortex layers, highlighted the AD-associated disruptions in laminar architecture, and identified changes in layer-to-layer interactions as AD progresses. Further, throughout the progression from NA to various stages of AD, we discovered specific genes that were significantly upregulated in neurons experiencing high stress and in nearby non-neuronal cells, compared to cells distant from the source of stress. Notably, the cell-cell interactions between the neurons under the high stress and adjacent glial cells that promote Aß clearance and neuroprotection were diminished in AD in response to stressors compared to NA. Through cell-type specific gene co-expression analysis, we identified three modules in excitatory and inhibitory neurons associated with neuronal protection, protein dephosphorylation, and negative regulation of Aß plaque formation. These modules negatively correlated with AD progression, indicating a reduced capacity for toxic substance clearance in AD subject samples. Moreover, we have discovered a novel transcription factor, ZNF460, that regulates all three modules, establishing it as a potential new therapeutic target for AD. Overall, utilizing the latest spatial transcriptome platform, our study developed the first transcriptome-wide atlas with subcellular resolution for assessing the molecular alterations in the human PFC due to AD. This atlas sheds light on the potential mechanisms underlying the progression from NA to AD.

omicsMIC: a comprehensive benchmarking platform for robust comparison of imputation methods in mass spectrometry-based omics data.

Mass spectrometry is a powerful and widely used tool for generating proteomics, lipidomics and metabolomics profiles, which is pivotal for elucidating biological processes and identifying biomarkers. However, missing values in mass spectrometry-based omics data may pose a critical challenge for the comprehensive identification of biomarkers and elucidation of the biological processes underlying human complex disorders. To alleviate this issue, various imputation methods for mass spectrometry-based omics data have been developed. However, a comprehensive comparison of these imputation methods is still lacking, and researchers are frequently confronted with a multitude of options without a clear rationale for method selection. To address this pressing need, we developed omicsMIC (mass spectrometry-based omics with Missing values Imputation methods Comparison platform), an interactive platform that provides researchers with a versatile framework to evaluate the performance of 28 diverse imputation methods. omicsMIC offers a nuanced perspective, acknowledging the inherent heterogeneity in biological data and the unique attributes of each dataset. Our platform empowers researchers to make data-driven decisions in imputation method selection based on real-time visualizations of the outcomes associated with different imputation strategies. The comprehensive benchmarking and versatility of omicsMIC make it a valuable tool for the scientific community engaged in mass spectrometry-based omics research. omicsMIC is freely available at https://github.com/WQLin8/omicsMIC.

Identification of the serum metabolites associated with cow milk consumption in Chinese Peri-/Postmenopausal women.

Cow milk consumption (CMC) and downstream alterations of serum metabolites are commonly considered important factors regulating human health status. Foods may lead to metabolic changes directly or indirectly through remodelling gut microbiota (GM). We sought to identify the metabolic alterations in Chinese Peri-/Postmenopausal women with habitual CMC and explore if the GM mediates the CMC-metabolite associations. 346 Chinese Peri-/Postmenopausal women participants were recruited in this study. Fixed effects regression and partial least squares discriminant analysis (PLS-DA) were applied to reveal alterations of serum metabolic features in different CMC groups. Spearman correlation coefficient was computed to detect metabolome-metagenome association. 36 CMC-associated metabolites including palmitic acid (FA(16:0)), 7alpha-hydroxy-4-cholesterin-3-one (7alphaC4), citrulline were identified by both fixed effects regression (FDR < 0.05) and PLS-DA (VIP score > 2). Some significant metabolite-GM associations were observed, including FA(16:0) with gut species Bacteroides ovatus, Bacteroides sp.D2. These findings would further prompt our understanding of the effect of cow milk on human health.

A Demographic-Conditioned Variational Autoencoder for fMRI Distribution Sampling and Removal of Confounds.

Objective: fMRI and derived measures such as functional connectivity (FC) have been used to predict brain age, general fluid intelligence, psychiatric disease status, and preclinical neurodegenerative disease. However, it is not always clear that all demographic confounds, such as age, sex, and race, have been removed from fMRI data. Additionally, many fMRI datasets are restricted to authorized researchers, making dissemination of these valuable data sources challenging. Methods: We create a variational autoencoder (VAE)-based model, DemoVAE, to decorrelate fMRI features from demographics and generate high-quality synthetic fMRI data based on user-supplied demographics. We train and validate our model using two large, widely used datasets, the Philadelphia Neurodevelopmental Cohort (PNC) and Bipolar and Schizophrenia Network for Intermediate Phenotypes (BSNIP). Results: We find that DemoVAE recapitulates group differences in fMRI data while capturing the full breadth of individual variations. Significantly, we also find that most clinical and computerized battery fields that are correlated with fMRI data are not correlated with DemoVAE latents. An exception are several fields related to schizophrenia medication and symptom severity. Conclusion: Our model generates fMRI data that captures the full distribution of FC better than traditional VAE or GAN models. We also find that most prediction using fMRI data is dependent on correlation with, and prediction of, demographics. Significance: Our DemoVAE model allows for generation of high quality synthetic data conditioned on subject demographics as well as the removal of the confounding effects of demographics. We identify that FC-based prediction tasks are highly influenced by demographic confounds.

A Demographic-Conditioned Variational Autoencoder for fMRI Distribution Sampling and Removal of Confounds.

Objective: fMRI and derived measures such as functional connectivity (FC) have been used to predict brain age, general fluid intelligence, psychiatric disease status, and preclinical neurodegenerative disease. However, it is not always clear that all demographic confounds, such as age, sex, and race, have been removed from fMRI data. Additionally, many fMRI datasets are restricted to authorized researchers, making dissemination of these valuable data sources challenging. Methods: We create a variational autoencoder (VAE)-based model, DemoVAE, to decorrelate fMRI features from demographics and generate high-quality synthetic fMRI data based on user-supplied demographics. We train and validate our model using two large, widely used datasets, the Philadelphia Neurodevel-opmental Cohort (PNC) and Bipolar and Schizophrenia Network for Intermediate Phenotypes (BSNIP). Results: We find that DemoVAE recapitulates group differences in fMRI data while capturing the full breadth of individual variations. Significantly, we also find that most clinical and computerized battery fields that are correlated with fMRI data are not correlated with DemoVAE latents. An exception are several fields related to schizophrenia medication and symptom severity. Conclusion: Our model generates fMRI data that captures the full distribution of FC better than traditional VAE or GAN models. We also find that most prediction using fMRI data is dependent on correlation with, and prediction of, demographics. Significance: Our DemoVAE model allows for generation of high quality synthetic data conditioned on subject demographics as well as the removal of the confounding effects of demographics. We identify that FC-based prediction tasks are highly influenced by demographic confounds.

Mendelian randomization and transcriptome analysis identified immune-related biomarkers for osteoarthritis.

Background: The immune microenvironment assumes a significant role in the pathogenesis of osteoarthritis (OA). However, the current biomarkers for the diagnosis and treatment of OA are not satisfactory. Our study aims to identify new OA immune-related biomarkers to direct the prevention and treatment of OA using multi-omics data. Methods: The discovery dataset integrated the GSE89408 and GSE143514 datasets to identify biomarkers that were significantly associated with the OA immune microenvironment through multiple machine learning methods and weighted gene co-expression network analysis (WGCNA). The identified signature genes were confirmed using two independent validation datasets. We also performed a two-sample mendelian randomization (MR) study to generate causal relationships between biomarkers and OA using OA genome-wide association study (GWAS) summary data (cases n = 24,955, controls n = 378,169). Inverse-variance weighting (IVW) method was used as the main method of causal estimates. Sensitivity analyses were performed to assess the robustness and reliability of the IVW results. Results: Three signature genes (FCER1G, HLA-DMB, and HHLA-DPA1) associated with the OA immune microenvironment were identified as having good diagnostic performances, which can be used as biomarkers. MR results showed increased levels of FCER1G (OR = 1.118, 95% CI 1.031-1.212, P = 0.041), HLA-DMB (OR = 1.057, 95% CI 1.045 -1.069, P = 1.11E-21) and HLA-DPA1 (OR = 1.030, 95% CI 1.005-1.056, P = 0.017) were causally and positively associated with the risk of developing OA. Conclusion: The present study identified the 3 potential immune-related biomarkers for OA, providing new perspectives for the prevention and treatment of OA. The MR study provides genetic support for the causal effects of the 3 biomarkers with OA and may provide new insights into the molecular mechanisms leading to the development of OA.

A Versatile Strategy for the Controlled Synthesis of Atomically Precise Palladium Nanoclusters.

The study of the structures, applications, and structure-property relationships of atomically precise metal nanoclusters relies heavily on their controlled synthesis. Although great progress has been made in the controlled synthesis of Group 11 (Cu, Ag, Au) metal nanoclusters, the preparation of Pd nanoclusters remains a grand challenge. Herein, a new, simple, and versatile synthetic strategy for the controlled synthesis of Pd nanoclusters is reported with tailorable structures and functions. The synthesis strategy involves the controllable transformations of Pd4(CO)4(CH3COO)4 in air, allowing the discovery of a family of Pd nanoclusters with well-defined structure and high yield. For example, by treating the Pd4(CO)4(CH3COO)4 with 2,2-dipyridine ligands, two clusters of Pd4 and Pd10 whose metal framework describes the growth of vertex-sharing tetrahedra have been selectively isolated. Interestingly, chiral Pd4 nanoclusters can be gained by virtue of customized chiral pyridine-imine ligands, thus representing a pioneering example to shed light on the hierarchical chiral nanostructures of Pd. This synthetic methodology also tolerates a wide variety of ligands and affords phosphine-ligated Pd nanoclusters in a simple way. It is believed that the successful exploration of the synthetic strategy would simulate the research enthusiasm on both the synthesis and application of atomically precise Pd nanoclusters.

Somatomotor-visual resting state functional connectivity increases after 2 years in the UK Biobank longitudinal cohort.

Purpose: Functional magnetic resonance imaging (fMRI) and functional connectivity (FC) have been used to follow aging in both children and older adults. Robust changes have been observed in children, in which high connectivity among all brain regions changes to a more modular structure with maturation. We examine FC changes in older adults after 2 years of aging in the UK Biobank (UKB) longitudinal cohort. Approach: We process fMRI connectivity data using the Power264 atlas and then test whether the average internetwork FC changes in the 2722-subject longitudinal cohort are statistically significant using a Bonferroni-corrected t-test. We also compare the ability of Power264 and UKB-provided, independent component analysis (ICA)-based FC to determine which of a longitudinal scan pair is older. Finally, we investigate cross-sectional FC changes as well as differences due to differing scanner tasks in the UKB, Philadelphia Neurodevelopmental Cohort, and Alzheimer's Disease Neuroimaging Initiative datasets. Results: We find a 6.8% average increase in somatomotor network (SMT)-visual network (VIS) connectivity from younger to older scans (corrected p<10-15) that occurs in male, female, older subject (>65 years old), and younger subject (<55 years old) groups. Among all internetwork connections, the average SMT-VIS connectivity is the best predictor of relative scan age. Using the full FC and a training set of 2000 subjects, one is able to predict which scan is older 82.5% of the time using either the full Power264 FC or the UKB-provided ICA-based FC. Conclusions: We conclude that SMT-VIS connectivity increases with age in the UKB longitudinal cohort and that resting state FC increases with age in the UKB cross-sectional cohort.

Unveiling the link between physical activity levels and dementia risk: Insights from the UK Biobank study.

BACKGROUND: There is limited information on the mixture effect and weights of light physical activity (LPA), moderate physical activity (MPA), and vigorous physical activity (VPA) on dementia risk. METHODS: A prospective cohort study was conducted based on the UK Biobank dataset. We included participants aged at least 45 years old without dementia at baseline between 2006-2010. The weighted quantile sum regression was used to explore the mixture effect and weights of three types of physical activity on dementia risk. RESULTS: This study includes 354,123 participants, with a mean baseline age of 58.0-year-old and 52.4 % of female participants. During a median follow-up time of 12.5 years, 5,136 cases of dementia were observed. The mixture effect of LPA, MPA, and VPA on dementia was statistically significant (ß: -0.0924, 95 % Confidence Interval (CI): (-0.1402, -0.0446), P < 0.001), with VPA (weight: 0.7922) contributing most to a lower dementia risk, followed by MPA (0.1939). For Alzheimer's disease, MPA contributed the most (0.8555); for vascular dementia, VPA contributed the most (0.6271). CONCLUSION: For Alzheimer's disease, MPA was identified as the most influential factor, while VPA stood out as the most impactful for vascular dementia.

Proteome-wide association study and functional validation identify novel protein markers for pancreatic ductal adenocarcinoma.

Pancreatic ductal adenocarcinoma (PDAC) remains a lethal malignancy, largely due to the paucity of reliable biomarkers for early detection and therapeutic targeting. Existing blood protein biomarkers for PDAC often suffer from replicability issues, arising from inherent limitations such as unmeasured confounding factors in conventional epidemiologic study designs. To circumvent these limitations, we use genetic instruments to identify proteins with genetically predicted levels to be associated with PDAC risk. Leveraging genome and plasma proteome data from the INTERVAL study, we established and validated models to predict protein levels using genetic variants. By examining 8,275 PDAC cases and 6,723 controls, we identified 40 associated proteins, of which 16 are novel. Functionally validating these candidates by focusing on 2 selected novel protein-encoding genes, GOLM1 and B4GALT1, we demonstrated their pivotal roles in driving PDAC cell proliferation, migration, and invasion. Furthermore, we also identified potential drug repurposing opportunities for treating PDAC. SIGNIFICANCE: PDAC is a notoriously difficult-to-treat malignancy, and our limited understanding of causal protein markers hampers progress in developing effective early detection strategies and treatments. Our study identifies novel causal proteins using genetic instruments and subsequently functionally validates selected novel proteins. This dual approach enhances our understanding of PDAC etiology and potentially opens new avenues for therapeutic interventions.

Beyond Amyloid: A Machine Learning-Driven Approach Reveals Properties of Potent GSK-3ß Inhibitors Targeting Neurofibrillary Tangles.

Current treatments for Alzheimer's disease (AD) focus on slowing memory and cognitive decline, but none offer curative outcomes. This study aims to explore and curate the common properties of active, drug-like molecules that modulate glycogen synthase kinase 3ß (GSK-3ß), a well-documented kinase with increased activity in tau hyperphosphorylation and neurofibrillary tangles-hallmarks of AD pathology. Leveraging quantitative structure-activity relationship (QSAR) data from the PubChem and ChEMBL databases, we employed seven machine learning models: logistic regression (LogR), k-nearest neighbors (KNN), random forest (RF), support vector machine (SVM), extreme gradient boosting (XGB), neural networks (NNs), and ensemble majority voting. Our goal was to correctly predict active and inactive compounds that inhibit GSK-3ß activity and identify their key properties. Among the six individual models, the NN demonstrated the highest performance with a 79% AUC-ROC on unbalanced external validation data, while the SVM model was superior in accurately classifying the compounds. The SVM and RF models surpassed NN in terms of Kappa values, and the ensemble majority voting model demonstrated slightly better accuracy to the NN on the external validation data. Feature importance analysis revealed that hydrogen bonds, phenol groups, and specific electronic characteristics are important features of molecular descriptors that positively correlate with active GSK-3ß inhibition. Conversely, structural features like imidazole rings, sulfides, and methoxy groups showed a negative correlation. Our study highlights the significance of structural, electronic, and physicochemical descriptors in screening active candidates against GSK-3ß. These predictive features could prove useful in therapeutic strategies to understand the important properties of GSK-3ß candidate inhibitors that may potentially benefit non-amyloid-based AD treatments targeting neurofibrillary tangles.

Oligogenic basis of premature ovarian insufficiency: an observational study.

BACKGROUND: The etiology of premature ovarian insufficiency, that is, the loss of ovarian activity before 40 years of age, is complex. Studies suggest that genetic factors are involved in 20-25% of cases. The aim of this study was to explore the oligogenic basis of premature ovarian insufficiency. RESULTS: Whole-exome sequencing of 93 patients with POI and whole-genome sequencing of 465 controls were performed. In the gene-burden analysis, multiple genetic variants, including those associated with DNA damage repair and meiosis, were more common in participants with premature ovarian insufficiency than in controls. The ORVAL-platform analysis confirmed the pathogenicity of the RAD52 and MSH6 combination. CONCLUSIONS: The results of this study indicate that oligogenic inheritance is an important cause of premature ovarian insufficiency and provide insights into the biological mechanisms underlying premature ovarian insufficiency.

Associations between genetically predicted plasma protein levels and Alzheimer's disease risk: a study using genetic prediction models.

BACKGROUND: Specific peripheral proteins have been implicated to play an important role in the development of Alzheimer's disease (AD). However, the roles of additional novel protein biomarkers in AD etiology remains elusive. The availability of large-scale AD GWAS and plasma proteomic data provide the resources needed for the identification of causally relevant circulating proteins that may serve as risk factors for AD and potential therapeutic targets. METHODS: We established and validated genetic prediction models for protein levels in plasma as instruments to investigate the associations between genetically predicted protein levels and AD risk. We studied 71,880 (proxy) cases and 383,378 (proxy) controls of European descent. RESULTS: We identified 69 proteins with genetically predicted concentrations showing associations with AD risk. The drugs almitrine and ciclopirox targeting ATP1A1 were suggested to have a potential for being repositioned for AD treatment. CONCLUSIONS: Our study provides additional insights into the underlying mechanisms of AD and potential therapeutic strategies.

A new hip fracture risk index derived from FEA-computed proximal femur fracture loads and energies-to-failure.

Hip fracture risk assessment is an important but challenging task. Quantitative CT-based patient-specific finite element (FE) analysis (FEA) incorporates bone geometry and bone density in the proximal femur. We developed a global FEA-computed fracture risk index to increase the prediction accuracy of hip fracture incidence. PURPOSE: Quantitative CT-based patient-specific finite element (FE) analysis (FEA) incorporates bone geometry and bone density in the proximal femur to compute the force (fracture load) and energy necessary to break the proximal femur in a particular loading condition. The fracture loads and energies-to-failure are individually associated with incident hip fracture, and provide different structural information about the proximal femur. METHODS: We used principal component analysis (PCA) to develop a global FEA-computed fracture risk index that incorporates the FEA-computed yield and ultimate failure loads and energies-to-failure in four loading conditions of 110 hip fracture subjects and 235 age- and sex-matched control subjects from the AGES-Reykjavik study. Using a logistic regression model, we compared the prediction performance for hip fracture based on the stratified resampling. RESULTS: We referred the first principal component (PC1) of the FE parameters as the global FEA-computed fracture risk index, which was the significant predictor of hip fracture (p-value < 0.001). The area under the receiver operating characteristic curve (AUC) using PC1 (0.776) was higher than that using all FE parameters combined (0.737) in the males (p-value < 0.001). CONCLUSIONS: The global FEA-computed fracture risk index increased hip fracture risk prediction accuracy in males.

CLCLSA: Cross-omics linked embedding with contrastive learning and self attention for integration with incomplete multi-omics data.

Integration of heterogeneous and high-dimensional multi-omics data is becoming increasingly important in understanding etiology of complex genetic diseases. Each omics technique only provides a limited view of the underlying biological process and integrating heterogeneous omics layers simultaneously would lead to a more comprehensive and detailed understanding of diseases and phenotypes. However, one obstacle faced when performing multi-omics data integration is the existence of unpaired multi-omics data due to instrument sensitivity and cost. Studies may fail if certain aspects of the subjects are missing or incomplete. In this paper, we propose a deep learning method for multi-omics integration with incomplete data by Cross-omics Linked unified embedding with Contrastive Learning and Self Attention (CLCLSA). Utilizing complete multi-omics data as supervision, the model employs cross-omics autoencoders to learn the feature representation across different types of biological data. The multi-omics contrastive learning is employed, which maximizes the mutual information between different types of omics. In addition, the feature-level self-attention and omics-level self-attention are employed to dynamically identify the most informative features for multi-omics data integration. Finally, a Softmax classifier is employed to perform multi-omics data classification. Extensive experiments were conducted on four public multi-omics datasets. The experimental results indicate that our proposed CLCLSA produces promising results in multi-omics data classification using both complete and incomplete multi-omics data.