Identification, analysis, and interpretation of a human serum metabolomics causal network in an observational study

WITHDRAWN: Broadcasters, receivers, functional groups of metabolites and the link to heart failure using polygenic factors

The full text of this preprint has been withdrawn, as it was submitted in error. Therefore, the authors do not wish this work to be cited as a reference. Questions should be directed to the corresponding author.

Broadcasters, receivers, functional groups of metabolites and the link to heart failure progression using polygenic factors

In a prospective study with records of heart failure (HF) incidence, we present metabolite profiling data from individuals without HF at baseline. We uncovered the interconnectivity of metabolites using data-driven and causal networks augmented with polygenic factors. Exploring the networks, we identified metabolite broadcasters, receivers, mediators, and subnetworks corresponding to functional classes of metabolites, and provided insights into the link between metabolomic architecture and regulation in health. We incorporated the network structure into the identification of metabolites associated with HF to control the effect of confounding metabolites. We identified metabolites associated with higher or lower risk of HF incidence, the associations that were not confounded by the other metabolites, such as glycine, ureidopropionic and glycocholic acids, and LPC 18:2. We revealed the underlying relationships of the findings. For example, asparagine directly influenced glycine, and both were inversely associated with HF. These two metabolites were influenced by polygenic factors and only essential amino acids which are not synthesized in the human body and come directly from the diet. Metabolites may play a critical role in linking genetic background and lifestyle factors to HF progression. Revealing the underlying connectivity of metabolites associated with HF strengthens the findings and facilitates a mechanistic understanding of HF progression.

Deep neural networks with knockoff features identify nonlinear causal relations and estimate effect sizes in complex biological systems

BACKGROUND

Learning the causal structure helps identify risk factors, disease mechanisms, and candidate therapeutics for complex diseases. However, although complex biological systems are characterized by nonlinear associations, existing bioinformatic methods of causal inference cannot identify the nonlinear relationships and estimate their effect size.

RESULTS

To overcome these limitations, we developed the first computational method that explicitly learns nonlinear causal relations and estimates the effect size using a deep neural network approach coupled with the knockoff framework, named causal directed acyclic graphs using deep learning variable selection (DAG-deepVASE). Using simulation data of diverse scenarios and identifying known and novel causal relations in molecular and clinical data of various diseases, we demonstrated that DAG-deepVASE consistently outperforms existing methods in identifying true and known causal relations. In the analyses, we also illustrate how identifying nonlinear causal relations and estimating their effect size help understand the complex disease pathobiology, which is not possible using other methods.

CONCLUSIONS

With these advantages, the application of DAG-deepVASE can help identify driver genes and therapeutic agents in biomedical studies and clinical trials.

Comprehensive characterization of putative genetic influences on plasma metabolome in a pediatric cohort

BACKGROUND

The human exposome is composed of diverse metabolites and small chemical compounds originated from endogenous and exogenous sources, respectively. Genetic and environmental factors influence metabolite levels, while the extent of genetic contributions across metabolic pathways is not yet known. Untargeted profiling of human metabolome using high-resolution mass spectrometry (HRMS) combined with genome-wide genotyping allows comprehensive identification of genetically influenced metabolites. As such previous studies of adults discovered and replicated genotype-metabotype associations. However, these associations have not been characterized in children.

RESULTS

We conducted the largest genome by metabolome-wide association study to date of children (N = 441) using 619,688 common genetic variants and 14,342 features measured by HRMS. Narrow-sense heritability (h²) estimates of plasma metabolite concentrations using genomic relatedness matrix restricted maximum likelihood (GREML) method showed a bimodal distribution with high h² (> 0.8) for 15.9% of features and low h² (< 0.2) for most of features (62.0%). The features with high h² were enriched for amino acid and nucleic acid metabolism, while carbohydrate and lipid concentrations showed low h². For each feature, a metabolite quantitative trait loci (mQTL) analysis was performed to identify genetic variants that were potentially associated with plasma levels. Fifty-four associations among 29 features and 43 genetic variants were identified at a genome-wide significance threshold p < 3.5 × 10^-12 (= 5 × 10^-8/14,342 features). Previously reported associations such as UGT1A1 and bilirubin; PYROXD2 and methyl lysine; and ACADS and butyrylcarnitine were successfully replicated in our pediatric cohort. We found potential candidates for novel associations including CSMD1 and a monostearyl alcohol triglyceride (m/z 781.7483, retention time (RT) 89.3 s); CALN1 and Tridecanol (m/z 283.2741, RT 27.6). A gene-level enrichment analysis using MAGMA revealed highly interconnected modules for dADP biosynthesis, sterol synthesis, and long-chain fatty acid transport in the gene-feature network.

CONCLUSION

Comprehensive profiling of plasma metabolome across age groups combined with genome-wide genotyping revealed a wide range of genetic influence on diverse chemical species and metabolic pathways. The developmental trajectory of a biological system is shaped by gene-environment interaction especially in early life. Therefore, continuous efforts on generating metabolomics data in diverse human tissue types across age groups are required to understand gene-environment interaction toward healthy aging trajectories.

A review of causal discovery methods for molecular network analysis

BACKGROUND

With the increasing availability and size of multi-omics datasets, investigating the casual relationships between molecular phenotypes has become an important aspect of exploring underlying biology andgenetics. There are an increasing number of methodlogies that have been developed and applied to moleular networks to investigate these causal interactions.

METHODS

We have introduced and reviewed the available methods for building large-scale causal molecular networks that have been developed and applied in the past decade.

RESULTS

In this review we have identified and summarized the existing methods for infering causality in large-scale causal molecular networks, and discussed important factors that will need to be considered in future research in this area.

CONCLUSION

Existing methods to infering causal molecular networks have their own strengths and limitations so there is no one best approach, and it is instead down to the discretion of the researcher. This review also to discusses some of the current limitations to biological interpretation of these networks, and important factors to consider for future studies on molecular networks.

Multivariate analysis of NMR-based metabolomic data

Nuclear magnetic resonance (NMR) spectroscopy allows for simultaneous detection of a wide range of metabolites and lipids. As metabolites act together in complex metabolic networks, they are often highly correlated, and optimal biological insight is achieved when using methods that take the correlation into account. For this reason, latent-variable-based methods, such as principal component analysis and partial least-squares discriminant analysis, are widely used in metabolomic studies. However, with increasing availability of larger population cohorts, and a shift from analysis of spectral data to using quantified metabolite levels, both more traditional statistical approaches and alternative machine learning methods have become more widely used. This review aims at providing an overview of the current state-of-the-art multivariate methods for the analysis of NMR-based metabolomic data as well as alternative methods, highlighting their strengths and limitations.

From classical mendelian randomization to causal networks for systematic integration of multi-omics

The number of studies with information at multiple biological levels of granularity, such as genomics, proteomics, and metabolomics, is increasing each year, and a biomedical questaion is how to systematically integrate these data to discover new biological mechanisms that have the potential to elucidate the processes of health and disease. Causal frameworks, such as Mendelian randomization (MR), provide a foundation to begin integrating data for new biological discoveries. Despite the growing number of MR applications in a wide variety of biomedical studies, there are few approaches for the systematic analysis of omic data. The large number and diverse types of molecular components involved in complex diseases interact through complex networks, and classical MR approaches targeting individual components do not consider the underlying relationships. In contrast, causal network models established in the principles of MR offer significant improvements to the classical MR framework for understanding omic data. Integration of these mostly distinct branches of statistics is a recent development, and we here review the current progress. To set the stage for causal network models, we review some recent progress in the classical MR framework. We then explain how to transition from the classical MR framework to causal networks. We discuss the identification of causal networks and evaluate the underlying assumptions. We also introduce some tests for sensitivity analysis and stability assessment of causal networks. We then review practical details to perform real data analysis and identify causal networks and highlight some of the utility of causal networks. The utilities with validated novel findings reveal the full potential of causal networks as a systems approach that will become necessary to integrate large-scale omic data.

Bounded Fuzzy Possibilistic Method Reveals Information about Lung Cancer through Analysis of Metabolomics

Learning methods, such as conventional clustering and classification, have been applied in diagnosing diseases to categorize samples based on their features. Going beyond clustering samples, membership degrees represent to what degree each sample belongs to a cluster. Variation of membership degrees in each cluster provides information about the cluster as a whole and each sample individually which enables us to have insights toward precision medicine. Membership degrees are measured more accurately through removing restrictions from clustering samples. Bounded Fuzzy Possibilistic Method (BFPM) introduces a membership function that keeps the search space flexible to cluster samples with higher accuracy. The method evaluates samples for their movement from one cluster to another. This technique allows us to find critical samples in advance those with the potential ability to belong to other clusters in the near future. BFPM was applied on metabolomics of individuals in a lung cancer case-control study. Metabolomics as proximal molecular signals to the actual disease processes may serve as strong biomarkers of current disease process. The goal is to know whether serum metabolites of a healthy human can be differentiated from those with lung cancer. Using BFPM, some differences were observed, the pathology data were evaluated, and critical samples were recognized.

Genome analysis and pleiotropy assessment using causal networks with loss of function mutation and metabolomics

BACKGROUND

Many genome-wide association studies have detected genomic regions associated with traits, yet understanding the functional causes of association often remains elusive. Utilizing systems approaches and focusing on intermediate molecular phenotypes might facilitate biologic understanding.

RESULTS

The availability of exome sequencing of two populations of African-Americans and European-Americans from the Atherosclerosis Risk in Communities study allowed us to investigate the effects of annotated loss-of-function (LoF) mutations on 122 serum metabolites. To assess the findings, we built metabolomic causal networks for each population separately and utilized structural equation modeling. We then validated our findings with a set of independent samples. By use of methods based on concepts of Mendelian randomization of genetic variants, we showed that some of the affected metabolites are risk predictors in the causal pathway of disease. For example, LoF mutations in the gene KIAA1755 were identified to elevate the levels of eicosapentaenoate (p-value = 5E-14), an essential fatty acid clinically identified to increase essential hypertension. We showed that this gene is in the pathway to triglycerides, where both triglycerides and essential hypertension are risk factors of metabolomic disorder and heart attack. We also identified that the gene CLDN17, harboring loss-of-function mutations, had pleiotropic actions on metabolites from amino acid and lipid pathways.

CONCLUSION

Using systems biology approaches for the analysis of metabolomics and genetic data, we integrated several biological processes, which lead to findings that may functionally connect genetic variants with complex diseases.

Plasma metabolites predict both insulin resistance and incident type 2 diabetes: a metabolomics approach within the Prevención con Dieta Mediterránea (PREDIMED) study

BACKGROUND

Insulin resistance is a complex metabolic disorder and is often associated with type 2 diabetes (T2D).

OBJECTIVES

The aim of this study was to test whether baseline metabolites can additionally improve the prediction of insulin resistance beyond classical risk factors. Furthermore, we examined whether a multimetabolite model predicting insulin resistance in nondiabetics can also predict incident T2D.

METHODS

We used a case-cohort study nested within the Prevención con Dieta Mediterránea (PREDIMED) trial in subsets of 700, 500, and 256 participants without T2D at baseline and 1 and 3 y. Fasting plasma metabolites were semiquantitatively profiled with liquid chromatography-tandem mass spectrometry. We assessed associations between metabolite concentrations and the homeostasis model of insulin resistance (HOMA-IR) through the use of elastic net regression analysis. We subsequently examined associations between the baseline HOMA-IR-related multimetabolite model and T2D incidence through the use of weighted Cox proportional hazard models.

RESULTS

We identified a set of baseline metabolites associated with HOMA-IR. One-year changes in metabolites were also significantly associated with HOMA-IR. The area under the curve was significantly greater for the model containing the classical risk factors and metabolites together compared with classical risk factors alone at baseline [0.81 (95% CI: 0.79, 0.84) compared with 0.69 (95% CI: 0.66, 0.73)] and during a 1-y period [0.69 (95% CI: 0.66, 0.72) compared with 0.57 (95% CI: 0.53, 0.62)]. The variance in HOMA-IR explained by the combination of metabolites and classical risk factors was also higher in all time periods. The estimated HRs for incident T2D in the multimetabolite score (model 3) predicting high HOMA-IR (median value or higher) or HOMA-IR (continuous) at baseline were 2.00 (95% CI: 1.58, 2.55) and 2.24 (95% CI: 1.72, 2.90), respectively, after adjustment for T2D risk factors.

CONCLUSIONS

The multimetabolite model identified in our study notably improved the predictive ability for HOMA-IR beyond classical risk factors and significantly predicted the risk of T2D.

Enhanced Molecular Appreciation of Psychiatric Disorders Through High-Dimensionality Data Acquisition and Analytics

The initial diagnosis, molecular investigation, treatment, and posttreatment care of major psychiatric disorders (schizophrenia and bipolar depression) are all still significantly hindered by the current inability to define these disorders in an explicit molecular signaling manner. High-dimensionality data analytics, using large datastreams from transcriptomic, proteomic, or metabolomic investigations, will likely advance both the appreciation of the molecular nature of major psychiatric disorders and simultaneously enhance our ability to more efficiently diagnose and treat these debilitating conditions. High-dimensionality data analysis in psychiatric research has been heterogeneous in aims and methods and limited by insufficient sample sizes, poorly defined case definitions, methodological inhomogeneity, and confounding results. All of these issues combine to constrain the conclusions that can be extracted from them. Here, we discuss possibilities for overcoming methodological challenges through the implementation of transcriptomic, proteomic, or metabolomics signatures in psychiatric diagnosis and offer an outlook for future investigations. To fulfill the promise of intelligent high-dimensionality data-based differential diagnosis in mental disease diagnosis and treatment, future research will need large, well-defined cohorts in combination with state-of-the-art technologies.

Redefining environmental exposure for disease etiology

Etiological studies of human exposures to environmental factors typically rely on low-throughput methods that target only a few hundred chemicals or mixtures. In this Perspectives article, I outline how environmental exposure can be defined by the blood exposome—the totality of chemicals circulating in blood. The blood exposome consists of chemicals derived from both endogenous and exogenous sources. Endogenous chemicals are represented by the human proteome and metabolome, which establish homeostatic networks of functional molecules. Exogenous chemicals arise from diet, vitamins, drugs, pathogens, microbiota, pollution, and lifestyle factors, and can be measured in blood as subsets of the proteome, metabolome, metals, macromolecular adducts, and foreign DNA and RNA. To conduct ‘exposome-wide association studies’, blood samples should be obtained prospectively from subjects—preferably at critical stages of life—and then analyzed in incident disease cases and matched controls to find discriminating exposures. Results from recent metabolomic investigations of archived blood illustrate our ability to discover potentially causal exposures with current technologies.

A comparison of methods for inferring causal relationships between genotype and phenotype using additional biological measurements

Genome wide association studies (GWAS) have been very successful over the last decade at identifying genetic variants associated with disease phenotypes. However, interpretation of the results obtained can be challenging. Incorporation of further relevant biological measurements (e.g. ‘omics’ data) measured in the same individuals for whom we have genotype and phenotype data may help us to learn more about the mechanism and pathways through which causal genetic variants affect disease. We review various methods for causal inference that can be used for assessing the relationships between genetic variables, other biological measures, and phenotypic outcome, and present a simulation study assessing the performance of the methods under different conditions. In general, the methods we considered did well at inferring the causal structure for data simulated under simple scenarios. However, the presence of an unknown and unmeasured common environmental effect could lead to spurious inferences, with the methods we considered displaying varying degrees of robustness to this confounder. The use of causal inference techniques to integrate omics and GWAS data has the potential to improve biological understanding of the pathways leading to disease. Our study demonstrates the suitability of various methods for performing causal inference under several biologically plausible scenarios.