A multiethnic whole genome sequencing study to identify novel loci for bone mineral density

Genomic structural variations link multiple genes to bone mineral density in a multi-ethnic cohort study: Louisiana osteoporosis study

Osteoporosis, characterized by low BMD, is a highly heritable metabolic bone disorder. Although single nucleotide variations (SNVs) have been extensively studied, they explain only a fraction of BMD heritability. Although genomic structural variations (SVs) are large-scale genomic alterations that contribute to genetic diversity in shaping phenotypic variations, the role of SVs in osteoporosis susceptibility remains poorly understood. This study aims to identify and prioritize genes that harbor BMD-related SVs. We performed whole genome sequencing on 4982 subjects from the Louisiana Osteoporosis Study. To obtain high-confidence SVs, the detection of SVs was performed using an ensemble approach. The SVs were tested for association with BMD variation at the hip (HIP), femoral neck (FNK), and lumbar spine (SPN), respectively. Additionally, we conducted co-occurrence analysis using multi-omics approaches to prioritize the identified genes based on their functional importance. Stratification was employed to explore the sex- and ethnicity-specific effects. We identified significant SV-BMD associations: 125 for FNK-BMD, 99 for SPN-BMD, and 83 for HIP-BMD. We observed SVs that were commonly associated with both FNK and HIP BMDs in our combined and stratified analyses. These SVs explain 13.3% to 19.1% of BMD variation. Novel bone-related genes emerged, including LINC02370, ZNF family genes, and ZDHHC family genes. Additionally, FMN2, carrying BMD-related deletions, showed associations with FNK or HIP BMDs, with sex-specific effects. The co-occurrence analysis prioritized an RNA gene LINC00494 and ZNF family genes positively associated with BMDs at different skeletal sites. Two potential causal genes, IBSP and SPP1, for osteoporosis were also identified. Our study uncovers new insights into genetic factors influencing BMD through SV analysis. We highlight BMD-related SVs, revealing a mix of shared and specific genetic influences across skeletal sites and gender or ethnicity. These findings suggest potential roles in osteoporosis pathophysiology, opening avenues for further research and therapeutic targets.

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.

An atlas of genetic effects on the monocyte methylome across European and African populations

Elucidating the genetic architecture of DNA methylation (DNAm) is crucial for decoding the etiology of complex diseases. However, current epigenomic studies often suffer from incomplete coverage of methylation sites and the use of tissues containing heterogeneous cell populations. To address these challenges, we present a comprehensive human methylome atlas based on deep whole-genome bisulfite sequencing (WGBS) and whole-genome sequencing (WGS) of purified monocytes from 298 European Americans (EA) and 160 African Americans (AA) in the Louisiana Osteoporosis Study. Our atlas enables the analysis of over 25 million DNAm sites. We identified 1,383,250 and 1,721,167 methylation quantitative trait loci (meQTLs) in cis -regions for EA and AA populations, respectively, with 880,108 sites shared between ancestries. While cis -meQTLs exhibited population-specific patterns, primarily due to differences in minor allele frequencies, shared cis -meQTLs showed high concordance across ancestries. Notably, cis -heritability estimates revealed significantly higher mean values in the AA population (0.09) compared to the EA population (0.04). Furthermore, we developed population-specific DNAm imputation models using Elastic Net, enabling methylome-wide association studies (MWAS) for 1,976,046 and 2,657,581 methylation sites in EA and AA, respectively. The performance of our MWAS models was validated through a systematic multi-ancestry analysis of 41 complex traits from the Million Veteran Program. Our findings bridge the gap between genomics and the monocyte methylome, uncovering novel methylation-phenotype associations and their transferability across diverse ancestries. The identified meQTLs, MWAS models, and data resources are freely available at www.gcbhub.org and https://osf.io/gct57/ .

Variability in performance of genetic-enhanced DXA-BMD prediction models across diverse ethnic and geographic populations: A risk prediction study

BACKGROUND

Osteoporosis is a major global health issue, weakening bones and increasing fracture risk. Dual-energy X-ray absorptiometry (DXA) is the standard for measuring bone mineral density (BMD) and diagnosing osteoporosis, but its costliness and complexity impede widespread screening adoption. Predictive modeling using genetic and clinical data offers a cost-effective alternative for assessing osteoporosis and fracture risk. This study aims to develop BMD prediction models using data from the UK Biobank (UKBB) and test their performance across different ethnic and geographical populations.

METHODS AND FINDINGS

We developed BMD prediction models for the femoral neck (FNK) and lumbar spine (SPN) using both genetic variants and clinical factors (such as sex, age, height, and weight), within 17,964 British white individuals from UKBB. Models based on regression with least absolute shrinkage and selection operator (LASSO), selected based on the coefficient of determination (R2) from a model selection subset of 5,973 individuals from British white population. These models were tested on 5 UKBB test sets and 12 independent cohorts of diverse ancestries, totaling over 15,000 individuals. Furthermore, we assessed the correlation of predicted BMDs with fragility fractures risk in 10 years in a case-control set of 287,183 European white participants without DXA-BMDs in the UKBB. With single-nucleotide polymorphism (SNP) inclusion thresholds at 5×10-6 and 5×10-7, the prediction models for FNK-BMD and SPN-BMD achieved the highest R2 of 27.70% with a 95% confidence interval (CI) of [27.56%, 27.84%] and 48.28% (95% CI [48.23%, 48.34%]), respectively. Adding genetic factors improved predictions slightly, explaining an additional 2.3% variation for FNK-BMD and 3% for SPN-BMD over clinical factors alone. Survival analysis revealed that the predicted FNK-BMD and SPN-BMD were significantly associated with fragility fracture risk in the European white population (P < 0.001). The hazard ratios (HRs) of the predicted FNK-BMD and SPN-BMD were 0.83 (95% CI [0.79, 0.88], corresponding to a 1.44% difference in 10-year absolute risk) and 0.72 (95% CI [0.68, 0.76], corresponding to a 1.64% difference in 10-year absolute risk), respectively, indicating that for every increase of one standard deviation in BMD, the fracture risk will decrease by 17% and 28%, respectively. However, the model's performance declined in other ethnic groups and independent cohorts. The limitations of this study include differences in clinical factors distribution and the use of only SNPs as genetic factors.

CONCLUSIONS

In this study, we observed that combining genetic and clinical factors improves BMD prediction compared to clinical factors alone. Adjusting inclusion thresholds for genetic variants (e.g., 5×10-6 or 5×10-7) rather than solely considering genome-wide association study (GWAS)-significant variants can enhance the model's explanatory power. The study highlights the need for training models on diverse populations to improve predictive performance across various ethnic and geographical groups.

Shared genetic effect of kidney function on bipolar and major depressive disorders: a large-scale genome-wide cross-trait analysis

BACKGROUND

Epidemiological studies have revealed a significant association between impaired kidney function and certain mental disorders, particularly bipolar disorder (BIP) and major depressive disorder (MDD). However, the evidence regarding shared genetics and causality is limited due to residual confounding and reverse causation.

METHODS

In this study, we conducted a large-scale genome-wide cross-trait association study to investigate the genetic overlap between 5 kidney function biomarkers (eGFRcrea, eGFRcys, blood urea nitrogen (BUN), serum urate, and UACR) and 2 mental disorders (MDD, BIP). Summary-level data of European ancestry were extracted from UK Biobank, Chronic Kidney Disease Genetics Consortium, and Psychiatric Genomics Consortium.

RESULTS

Using LD score regression, we found moderate but significant genetic correlations between kidney function biomarker traits on BIP and MDD. Cross-trait meta-analysis identified 1 to 19 independent significant loci that were found shared among 10 pairs of 5 kidney function biomarkers traits and 2 mental disorders. Among them, 3 novel genes: SUFU, IBSP, and PTPRJ, were also identified in transcriptome-wide association study analysis (TWAS), most of which were observed in the nervous and digestive systems (FDR < 0.05). Pathway analysis showed the immune system could play a role between kidney function biomarkers and mental disorders. Bidirectional mendelian randomization analysis suggested a potential causal relationship of kidney function biomarkers on BIP and MDD.

CONCLUSIONS

In conclusion, the study demonstrated that both BIP and MDD shared genetic architecture with kidney function biomarkers, providing new insights into their genetic architectures and suggesting that larger GWASs are warranted.

GWAS-informed data integration and non-coding CRISPRi screen illuminate genetic etiology of bone mineral density

Over 1,100 independent signals have been identified with genome-wide association studies (GWAS) for bone mineral density (BMD), a key risk factor for mortality-increasing fragility fractures; however, the effector gene(s) for most remain unknown. Informed by a variant-to-gene mapping strategy implicating 89 non-coding elements predicted to regulate osteoblast gene expression at BMD GWAS loci, we executed a single-cell CRISPRi screen in human fetal osteoblast 1.19 cells (hFOBs). The BMD relevance of hFOBs was supported by heritability enrichment from cross-cell type stratified LD-score regression involving 98 cell types grouped into 15 tissues. 24 genes showed perturbation in the screen, with four (ARID5B, CC2D1B, EIF4G2, and NCOA3) exhibiting consistent effects upon siRNA knockdown on three measures of osteoblast maturation and mineralization. Lastly, additional heritability enrichments, genetic correlations, and multi-trait fine-mapping revealed that many BMD GWAS signals are pleiotropic and likely mediate their effects via non-bone tissues that warrant attention in future screens.

Multi-View Variational Autoencoder for Missing Value Imputation in Untargeted Metabolomics

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-view 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 R^2-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.

Multi-view information fusion using multi-view variational autoencoder to predict proximal femoral fracture load

Background

Hip fracture occurs when an applied force exceeds the force that the proximal femur can support (the fracture load or "strength") and can have devastating consequences with poor functional outcomes. Proximal femoral strengths for specific loading conditions can be computed by subject-specific finite element analysis (FEA) using quantitative computerized tomography (QCT) images. However, the radiation and availability of QCT limit its clinical usability. Alternative low-dose and widely available measurements, such as dual energy X-ray absorptiometry (DXA) and genetic factors, would be preferable for bone strength assessment. The aim of this paper is to design a deep learning-based model to predict proximal femoral strength using multi-view information fusion.

Results

We developed new models using multi-view variational autoencoder (MVAE) for feature representation learning and a product of expert (PoE) model for multi-view information fusion. We applied the proposed models to an in-house Louisiana Osteoporosis Study (LOS) cohort with 931 male subjects, including 345 African Americans and 586 Caucasians. We performed genome-wide association studies (GWAS) to select 256 genetic variants with the lowest p-values for each proximal femoral strength and integrated whole genome sequence (WGS) features and DXA-derived imaging features to predict proximal femoral strength. The best prediction model for fall fracture load was acquired by integrating WGS features and DXA-derived imaging features. The designed models achieved the mean absolute percentage error of 18.04%, 6.84% and 7.95% for predicting proximal femoral fracture loads using linear models of fall loading, nonlinear models of fall loading, and nonlinear models of stance loading, respectively.

Conclusion

The proposed models are capable of predicting proximal femoral strength using WGS features and DXA-derived imaging features. Though this tool is not a substitute for predicting FEA using QCT images, it would make improved assessment of hip fracture risk more widely available while avoiding the increased radiation exposure from QCT.

Discovery of potential biomarkers for osteoporosis diagnosis by individual omics and multi-omics technologies

INTRODUCTION

Global aging has made osteoporosis an increasingly serious public health problem. Osteoporotic fractures seriously affect the quality of life of patients and increase disability and mortality rates. Early diagnosis is important for timely intervention. The continuous development of individual- and multi-omics methods is helpful for the exploration and discovery of biomarkers for the diagnosis of osteoporosis.

AREAS COVERED

In this review, we first introduce the epidemiological status of osteoporosis and then describe the pathogenesis of osteoporosis. Furthermore, the latest progress in individual- and multi-omics technologies for exploring biomarkers for osteoporosis diagnosis is summarized. Moreover, we clarify the advantages and disadvantages of the application of osteoporosis biomarkers obtained using the omics method. Finally, we put forward valuable views on the future research direction of diagnostic biomarkers of osteoporosis.

EXPERT OPINION

Omics methods undoubtedly provide greatly contribute to the exploration of diagnostic biomarkers of osteoporosis; however, in the future, the clinical validity and clinical utility of the obtained potential biomarkers should be thoroughly examined. In addition, the improvement and optimization of the detection methods for different types of biomarkers and standardization of the detection process guarantee the reliability and accuracy of the detection results.

FHL5 Controls Vascular Disease-Associated Gene Programs in Smooth Muscle Cells

BACKGROUND

Genome-wide association studies have identified hundreds of loci associated with common vascular diseases, such as coronary artery disease, myocardial infarction, and hypertension. However, the lack of mechanistic insights for many GWAS loci limits their translation into the clinic. Among these loci with unknown functions is UFL1-four-and-a-half LIM (LIN-11, Isl-1, MEC-3) domain 5 (FHL5; chr6q16.1), which reached genome-wide significance in a recent coronary artery disease/ myocardial infarction GWAS meta-analysis. UFL1-FHL5 is also associated with several vascular diseases, consistent with the widespread pleiotropy observed for GWAS loci.

METHODS

We apply a multimodal approach leveraging statistical fine-mapping, epigenomic profiling, and ex vivo analysis of human coronary artery tissues to implicate FHL5 as the top candidate causal gene. We unravel the molecular mechanisms of the cross-phenotype genetic associations through in vitro functional analyses and epigenomic profiling experiments in coronary artery smooth muscle cells.

RESULTS

We prioritized FHL5 as the top candidate causal gene at the UFL1-FHL5 locus through expression quantitative trait locus colocalization methods. FHL5 gene expression was enriched in the smooth muscle cells and pericyte population in human artery tissues with coexpression network analyses supporting a functional role in regulating smooth muscle cell contraction. Unexpectedly, under procalcifying conditions, FHL5 overexpression promoted vascular calcification and dysregulated processes related to extracellular matrix organization and calcium handling. Lastly, by mapping FHL5 binding sites and inferring FHL5 target gene function using artery tissue gene regulatory network analyses, we highlight regulatory interactions between FHL5 and downstream coronary artery disease/myocardial infarction loci, such as FOXL1 and FN1 that have roles in vascular remodeling.

CONCLUSIONS

Taken together, these studies provide mechanistic insights into the pleiotropic genetic associations of UFL1-FHL5. We show that FHL5 mediates vascular disease risk through transcriptional regulation of downstream vascular remodeling gene programs. These transacting mechanisms may explain a portion of the heritable risk for complex vascular diseases.

Single-cell RNA sequencing reveals in vivo osteoimmunology interactions between the immune and skeletal systems

BACKGROUND

While osteoimmunology interactions between the immune and skeletal systems are known to play an important role in osteoblast development, differentiation and bone metabolism related disease like osteoporosis, such interactions in either bone microenvironment or peripheral circulation in vivo at the single-cell resolution have not yet been characterized.

METHODS

We explored the osteoimmunology communications between immune cells and osteoblastic lineage cells (OBCs) by performing CellphoneDB and CellChat analyses with single-cell RNA sequencing (scRNA-seq) data from human femoral head. We also explored the osteoimmunology effects of immune cells in peripheral circulation on skeletal phenotypes. We used a scRNA-seq dataset of peripheral blood monocytes (PBMs) to perform deconvolution analysis. Then weighted gene co-expression network analysis (WGCNA) was used to identify monocyte subtype-specific subnetworks. We next used cell-specific network (CSN) and the least absolute shrinkage and selection operator (LASSO) to analyze the correlation of a gene subnetwork identified by WGCNA with bone mineral density (BMD).

RESULTS

We constructed immune cell and OBC communication networks and further identified L-R genes, such as JAG1 and NOTCH1/2, with ossification related functions. We also found a Mono4 related subnetwork that may relate to BMD variation in both older males and postmenopausal female subjects.

CONCLUSIONS

This is the first study to identify numerous ligand-receptor pairs that likely mediate signals between immune cells and osteoblastic lineage cells. This establishes a foundation to reveal advanced and in-depth osteoimmunology interactions to better understand the relationship between local bone microenvironment and immune cells in peripheral blood and the impact on bone phenotypes.

An autoencoder-based deep learning method for genotype imputation

Genotype imputation has a wide range of applications in genome-wide association study (GWAS), including increasing the statistical power of association tests, discovering trait-associated loci in meta-analyses, and prioritizing causal variants with fine-mapping. In recent years, deep learning (DL) based methods, such as sparse convolutional denoising autoencoder (SCDA), have been developed for genotype imputation. However, it remains a challenging task to optimize the learning process in DL-based methods to achieve high imputation accuracy. To address this challenge, we have developed a convolutional autoencoder (AE) model for genotype imputation and implemented a customized training loop by modifying the training process with a single batch loss rather than the average loss over batches. This modified AE imputation model was evaluated using a yeast dataset, the human leukocyte antigen (HLA) data from the 1,000 Genomes Project (1KGP), and our in-house genotype data from the Louisiana Osteoporosis Study (LOS). Our modified AE imputation model has achieved comparable or better performance than the existing SCDA model in terms of evaluation metrics such as the concordance rate (CR), the Hellinger score, the scaled Euclidean norm (SEN) score, and the imputation quality score (IQS) in all three datasets. Taking the imputation results from the HLA data as an example, the AE model achieved an average CR of 0.9468 and 0.9459, Hellinger score of 0.9765 and 0.9518, SEN score of 0.9977 and 0.9953, and IQS of 0.9515 and 0.9044 at missing ratios of 10% and 20%, respectively. As for the results of LOS data, it achieved an average CR of 0.9005, Hellinger score of 0.9384, SEN score of 0.9940, and IQS of 0.8681 at the missing ratio of 20%. In summary, our proposed method for genotype imputation has a great potential to increase the statistical power of GWAS and improve downstream post-GWAS analyses.