TreeKernel: interpretable kernel machine tests for interactions between -omics and clinical predictors with applications to metabolomics and COPD phenotypes.

BACKGROUND: In this paper, we are interested in interactions between a high-dimensional -omics dataset and clinical covariates. The goal is to evaluate the relationship between a phenotype of interest and a high-dimensional omics pathway, where the effect of the omics data depends on subjects' clinical covariates (age, sex, smoking status, etc.). For instance, metabolic pathways can vary greatly between sexes which may also change the relationship between certain metabolic pathways and a clinical phenotype of interest. We propose partitioning the clinical covariate space and performing a kernel association test within those partitions. To illustrate this idea, we focus on hierarchical partitions of the clinical covariate space and kernel tests on metabolic pathways. RESULTS: We see that our proposed method outperforms competing methods in most simulation scenarios. It can identify different relationships among clinical groups with higher power in most scenarios while maintaining a proper Type I error rate. The simulation studies also show a robustness to the grouping structure within the clinical space. We also apply the method to the COPDGene study and find several clinically meaningful interactions between metabolic pathways, the clinical space, and lung function. CONCLUSION: TreeKernel provides a simple and interpretable process for testing for relationships between high-dimensional omics data and clinical outcomes in the presence of interactions within clinical cohorts. The method is broadly applicable to many studies.

Bilirubin-associated single nucleotide polymorphism (SNP) and respiratory health outcomes: a mendelian randomization study.

BACKGROUND: Observational studies have shown an association between higher bilirubin levels and improved respiratory health outcomes. Targeting higher bilirubin levels has been proposed as a novel therapeutic strategy in COPD. However, bilirubin levels are influenced by multiple intrinsic and extrinsic factors, and these observational studies are prone to confounding. Genetic analyses are one approach to overcoming residual confounding in observational studies. OBJECTIVES: To test associations between a genetic determinant of bilirubin levels and respiratory health outcomes. METHODS: COPDGene participants underwent genotyping at the baseline visit. We confirmed established associations between homozygosity for rs6742078 and higher bilirubin, and between higher bilirubin and decreased risk of acute respiratory events within this cohort. For our primary analysis, we used negative binomial regression to test associations between homozygosity for rs6742078 and rate of acute respiratory events. RESULTS: 8,727 participants (n = 6,228 non-Hispanic white and 2,499 African American) were included. Higher bilirubin was associated with decreased rate of acute respiratory events [incidence rate ratio (IRR) 0.85, 95% CI 0.75 to 0.96 per SD increase in bilirubin intensity]. We did not find significant associations between homozygosity for rs6742078 and acute respiratory events (IRR 0.94, 95% CI 0.70 to 1.25 for non-Hispanic white and 1.09, 95% CI 0.91 to 1.31 for African American participants). CONCLUSIONS: A genetic determinant of higher bilirubin levels was not associated with better respiratory health outcomes. These results do not support targeting higher bilirubin levels as a therapeutic strategy in COPD.

PaIRKAT: A pathway integrated regression-based kernel association test with applications to metabolomics and COPD phenotypes.

High-throughput data such as metabolomics, genomics, transcriptomics, and proteomics have become familiar data types within the "-omics" family. For this work, we focus on subsets that interact with one another and represent these "pathways" as graphs. Observed pathways often have disjoint components, i.e., nodes or sets of nodes (metabolites, etc.) not connected to any other within the pathway, which notably lessens testing power. In this paper we propose the Pathway Integrated Regression-based Kernel Association Test (PaIRKAT), a new kernel machine regression method for incorporating known pathway information into the semi-parametric kernel regression framework. This work extends previous kernel machine approaches. This paper also contributes an application of a graph kernel regularization method for overcoming disconnected pathways. By incorporating a regularized or "smoothed" graph into a score test, PaIRKAT can provide more powerful tests for associations between biological pathways and phenotypes of interest and will be helpful in identifying novel pathways for targeted clinical research. We evaluate this method through several simulation studies and an application to real metabolomics data from the COPDGene study. Our simulation studies illustrate the robustness of this method to incorrect and incomplete pathway knowledge, and the real data analysis shows meaningful improvements of testing power in pathways. PaIRKAT was developed for application to metabolomic pathway data, but the techniques are easily generalizable to other data sources with a graph-like structure.

Soluble receptor for advanced glycation end products (sRAGE) as a biomarker of COPD.

BACKGROUND: Soluble receptor for advanced glycation end products (sRAGE) is a proposed emphysema and airflow obstruction biomarker; however, previous publications have shown inconsistent associations and only one study has investigate the association between sRAGE and emphysema. No cohorts have examined the association between sRAGE and progressive decline of lung function. There have also been no evaluation of assay compatibility, receiver operating characteristics, and little examination of the effect of genetic variability in non-white population. This manuscript addresses these deficiencies and introduces novel data from Pittsburgh COPD SCCOR and as well as novel work on airflow obstruction. A meta-analysis is used to quantify sRAGE associations with clinical phenotypes. METHODS: sRAGE was measured in four independent longitudinal cohorts on different analytic assays: COPDGene (n = 1443); SPIROMICS (n = 1623); ECLIPSE (n = 2349); Pittsburgh COPD SCCOR (n = 399). We constructed adjusted linear mixed models to determine associations of sRAGE with baseline and follow up forced expiratory volume at one second (FEV1) and emphysema by quantitative high-resolution CT lung density at the 15th percentile (adjusted for total lung capacity). RESULTS: Lower plasma or serum sRAGE values were associated with a COPD diagnosis (P < 0.001), reduced FEV1 (P < 0.001), and emphysema severity (P < 0.001). In an inverse-variance weighted meta-analysis, one SD lower log10-transformed sRAGE was associated with 105 ± 22 mL lower FEV1 and 4.14 ± 0.55 g/L lower adjusted lung density. After adjusting for covariates, lower sRAGE at baseline was associated with greater FEV1 decline and emphysema progression only in the ECLIPSE cohort. Non-Hispanic white subjects carrying the rs2070600 minor allele (A) and non-Hispanic African Americans carrying the rs2071288 minor allele (A) had lower sRAGE measurements compare to those with the major allele, but their emphysema-sRAGE regression slopes were similar. CONCLUSIONS: Lower blood sRAGE is associated with more severe airflow obstruction and emphysema, but associations with progression are inconsistent in the cohorts analyzed. In these cohorts, genotype influenced sRAGE measurements and strengthened variance modelling. Thus, genotype should be included in sRAGE evaluations.

Comparison of Proteomic Assessment Methods in Multiple Cohort Studies.

Novel proteomics platforms, such as the aptamer-based SOMAscan platform, can quantify large numbers of proteins efficiently and cost-effectively and are rapidly growing in popularity. However, comparisons to conventional immunoassays remain underexplored, leaving investigators unsure when cross-assay comparisons are appropriate. The correlation of results from immunoassays with relative protein quantification is explored by SOMAscan. For 63 proteins assessed in two chronic obstructive pulmonary disease (COPD) cohorts, subpopulations and intermediate outcome measures in COPD Study (SPIROMICS), and COPDGene, using myriad rules based medicine multiplex immunoassays and SOMAscan, Spearman correlation coefficients range from -0.13 to 0.97, with a median correlation coefficient of ≈0.5 and consistent results across cohorts. A similar range is observed for immunoassays in the population-based Multi-Ethnic Study of Atherosclerosis and for other assays in COPDGene and SPIROMICS. Comparisons of relative quantification from the antibody-based Olink platform and SOMAscan in a small cohort of myocardial infarction patients also show a wide correlation range. Finally, cis pQTL data, mass spectrometry aptamer confirmation, and other publicly available data are integrated to assess relationships with observed correlations. Correlation between proteomics assays shows a wide range and should be carefully considered when comparing and meta-analyzing proteomics data across assays and studies.

An open source knowledge graph ecosystem for the life sciences.

Translational research requires data at multiple scales of biological organization. Advancements in sequencing and multi-omics technologies have increased the availability of these data, but researchers face significant integration challenges. Knowledge graphs (KGs) are used to model complex phenomena, and methods exist to construct them automatically. However, tackling complex biomedical integration problems requires flexibility in the way knowledge is modeled. Moreover, existing KG construction methods provide robust tooling at the cost of fixed or limited choices among knowledge representation models. PheKnowLator (Phenotype Knowledge Translator) is a semantic ecosystem for automating the FAIR (Findable, Accessible, Interoperable, and Reusable) construction of ontologically grounded KGs with fully customizable knowledge representation. The ecosystem includes KG construction resources (e.g., data preparation APIs), analysis tools (e.g., SPARQL endpoint resources and abstraction algorithms), and benchmarks (e.g., prebuilt KGs). We evaluated the ecosystem by systematically comparing it to existing open-source KG construction methods and by analyzing its computational performance when used to construct 12 different large-scale KGs. With flexible knowledge representation, PheKnowLator enables fully customizable KGs without compromising performance or usability.

Systemic Markers of Lung Function and Forced Expiratory Volume in 1 Second Decline across Diverse Cohorts.

Rationale: Chronic obstructive pulmonary disease (COPD) is a complex disease characterized by airway obstruction and accelerated lung function decline. Our understanding of systemic protein biomarkers associated with COPD remains incomplete. Objectives: To determine what proteins and pathways are associated with impaired pulmonary function in a diverse population. Methods: We studied 6,722 participants across six cohort studies with both aptamer-based proteomic and spirometry data (4,566 predominantly White participants in a discovery analysis and 2,156 African American cohort participants in a validation). In linear regression models, we examined protein associations with baseline forced expiratory volume in 1 second (FEV1) and FEV1/forced vital capacity (FVC). In linear mixed effects models, we investigated the associations of baseline protein levels with rate of FEV1 decline (ml/yr) in 2,777 participants with up to 7 years of follow-up spirometry. Results: We identified 254 proteins associated with FEV1 in our discovery analyses, with 80 proteins validated in the Jackson Heart Study. Novel validated protein associations include kallistatin serine protease inhibitor, growth differentiation factor 2, and tumor necrosis factor-like weak inducer of apoptosis (discovery ß = 0.0561, Q = 4.05 × 10-10; ß = 0.0421, Q = 1.12 × 10-3; and ß = 0.0358, Q = 1.67 × 10-3, respectively). In longitudinal analyses within cohorts with follow-up spirometry, we identified 15 proteins associated with FEV1 decline (Q < 0.05), including elafin leukocyte elastase inhibitor and mucin-associated TFF2 (trefoil factor 2; ß = -4.3 ml/yr, Q = 0.049; ß = -6.1 ml/yr, Q = 0.032, respectively). Pathways and processes highlighted by our study include aberrant extracellular matrix remodeling, enhanced innate immune response, dysregulation of angiogenesis, and coagulation. Conclusions: In this study, we identify and validate novel biomarkers and pathways associated with lung function traits in a racially diverse population. In addition, we identify novel protein markers associated with FEV1 decline. Several protein findings are supported by previously reported genetic signals, highlighting the plausibility of certain biologic pathways. These novel proteins might represent markers for risk stratification, as well as novel molecular targets for treatment of COPD.

Molecular cartooning with knowledge graphs.

Molecular "cartoons," such as pathway diagrams, provide a visual summary of biomedical research results and hypotheses. Their ubiquitous appearance within the literature indicates their universal application in mechanistic communication. A recent survey of pathway diagrams identified 64,643 pathway figures published between 1995 and 2019 with 1,112,551 mentions of 13,464 unique human genes participating in a wide variety of biological processes. Researchers generally create these diagrams using generic diagram editing software that does not itself embody any biomedical knowledge. Biomedical knowledge graphs (KGs) integrate and represent knowledge in a semantically consistent way, systematically capturing biomedical knowledge similar to that in molecular cartoons. KGs have the potential to provide context and precise details useful in drawing such figures. However, KGs cannot generally be translated directly into figures. They include substantial material irrelevant to the scientific point of a given figure and are often more detailed than is appropriate. How could KGs be used to facilitate the creation of molecular diagrams? Here we present a new approach towards cartoon image creation that utilizes the semantic structure of knowledge graphs to aid the production of molecular diagrams. We introduce a set of "semantic graphical actions" that select and transform the relational information between heterogeneous entities (e.g., genes, proteins, pathways, diseases) in a KG to produce diagram schematics that meet the scientific communication needs of the user. These semantic actions search, select, filter, transform, group, arrange, connect and extract relevant subgraphs from KGs based on meaning in biological terms, e.g., a protein upstream of a target in a pathway. To demonstrate the utility of this approach, we show how semantic graphical actions on KGs could have been used to produce three existing pathway diagrams in diverse biomedical domains: Down Syndrome, COVID-19, and neuroinflammation. Our focus is on recapitulating the semantic content of the figures, not the layout, glyphs, or other aesthetic aspects. Our results suggest that the use of KGs and semantic graphical actions to produce biomedical diagrams will reduce the effort required and improve the quality of this visual form of scientific communication.

Metabolomic Profiling Reveals Sex Specific Associations with Chronic Obstructive Pulmonary Disease and Emphysema.

Susceptibility and progression of lung disease, as well as response to treatment, often differ by sex, yet the metabolic mechanisms driving these sex-specific differences are still poorly understood. Women with chronic obstructive pulmonary disease (COPD) have less emphysema and more small airway disease on average than men, though these differences become less pronounced with more severe airflow limitation. While small studies of targeted metabolites have identified compounds differing by sex and COPD status, the sex-specific effect of COPD on systemic metabolism has yet to be interrogated. Significant sex differences were observed in 9 of the 11 modules identified in COPDGene. Sex-specific associations by COPD status and emphysema were observed in 3 modules for each phenotype. Sex stratified individual metabolite associations with COPD demonstrated male-specific associations in sphingomyelins and female-specific associations in acyl carnitines and phosphatidylethanolamines. There was high preservation of module assignments in SPIROMICS (SubPopulations and InteRmediate Outcome Measures In COPD Study) and similar female-specific shift in acyl carnitines. Several COPD associated metabolites differed by sex. Acyl carnitines and sphingomyelins demonstrate sex-specific abundances and may represent important metabolic signatures of sex differences in COPD. Accurately characterizing the sex-specific molecular differences in COPD is vital for personalized diagnostics and therapeutics.

Multi-omics subtyping pipeline for chronic obstructive pulmonary disease.

Chronic Obstructive Pulmonary Disease (COPD) is the third leading cause of mortality in the United States; however, COPD has heterogeneous clinical phenotypes. This is the first large scale attempt which uses transcriptomics, proteomics, and metabolomics (multi-omics) to determine whether there are molecularly defined clusters with distinct clinical phenotypes that may underlie the clinical heterogeneity. Subjects included 3,278 subjects from the COPDGene cohort with at least one of the following profiles: whole blood transcriptomes (2,650 subjects); plasma proteomes (1,013 subjects); and plasma metabolomes (1,136 subjects). 489 subjects had all three contemporaneous -omics profiles. Autoencoder embeddings were performed individually for each -omics dataset. Embeddings underwent subspace clustering using MineClus, either individually by -omics or combined, followed by recursive feature selection based on Support Vector Machines. Clusters were tested for associations with clinical variables. Optimal single -omics clustering typically resulted in two clusters. Although there was overlap for individual -omics cluster membership, each -omics cluster tended to be defined by unique molecular pathways. For example, prominent molecular features of the metabolome-based clustering included sphingomyelin, while key molecular features of the transcriptome-based clusters were related to immune and bacterial responses. We also found that when we integrated the -omics data at a later stage, we identified subtypes that varied based on age, severity of disease, in addition to diffusing capacity of the lungs for carbon monoxide, and precent on atrial fibrillation. In contrast, when we integrated the -omics data at an earlier stage by treating all data sets equally, there were no clinical differences between subtypes. Similar to clinical clustering, which has revealed multiple heterogenous clinical phenotypes, we show that transcriptomics, proteomics, and metabolomics tend to define clusters of COPD patients with different clinical characteristics. Thus, integrating these different -omics data sets affords additional insight into the molecular nature of COPD and its heterogeneity.

Identifying Protein-metabolite Networks Associated with COPD Phenotypes.

Chronic obstructive pulmonary disease (COPD) is a disease in which airflow obstruction in the lung makes it difficult for patients to breathe. Although COPD occurs predominantly in smokers, there are still deficits in our understanding of the additional risk factors in smokers. To gain a deeper understanding of the COPD molecular signatures, we used Sparse Multiple Canonical Correlation Network (SmCCNet), a recently developed tool that uses sparse multiple canonical correlation analysis, to integrate proteomic and metabolomic data from the blood of 1008 participants of the COPDGene study to identify novel protein-metabolite networks associated with lung function and emphysema. Our aim was to integrate -omic data through SmCCNet to build interpretable networks that could assist in the discovery of novel biomarkers that may have been overlooked in alternative biomarker discovery methods. We found a protein-metabolite network consisting of 13 proteins and 7 metabolites which had a -0.34 correlation (p-value = 2.5 × 10-28) to lung function. We also found a network of 13 proteins and 10 metabolites that had a -0.27 correlation (p-value = 2.6 × 10-17) to percent emphysema. Protein-metabolite networks can provide additional information on the progression of COPD that complements single biomarker or single -omic analyses.

IGSF3 mutation identified in patient with severe COPD alters cell function and motility.

Cigarette smoking (CS) and genetic susceptibility determine the risk for development, progression, and severity of chronic obstructive pulmonary diseases (COPD). We posited that an incidental balanced reciprocal chromosomal translocation was linked to a patient's risk of severe COPD. We determined that 46,XX,t(1;4)(p13.1;q34.3) caused a breakpoint in the immunoglobulin superfamily member 3 (IGSF3) gene, with markedly decreased expression. Examination of COPDGene cohort identified 14 IGSF3 SNPs, of which rs1414272 and rs12066192 were directly and rs6703791 inversely associated with COPD severity, including COPD exacerbations. We confirmed that IGSF3 is a tetraspanin-interacting protein that colocalized with CD9 and integrin B1 in tetraspanin-enriched domains. IGSF3-deficient patient-derived lymphoblastoids exhibited multiple alterations in gene expression, especially in the unfolded protein response and ceramide pathways. IGSF3-deficient lymphoblastoids had high ceramide and sphingosine-1 phosphate but low glycosphingolipids and ganglioside levels, and they were less apoptotic and more adherent, with marked changes in multiple TNFRSF molecules. Similarly, IGSF3 knockdown increased ceramide in lung structural cells, rendering them more adherent, with impaired wound repair and weakened barrier function. These findings suggest that, by maintaining sphingolipid and membrane receptor homeostasis, IGSF3 is required for cell mobility-mediated lung injury repair. IGSF3 deficiency may increase susceptibility to CS-induced lung injury in COPD.

Plasma Metabolomic Signatures of Chronic Obstructive Pulmonary Disease and the Impact of Genetic Variants on Phenotype-Driven Modules.

Background: Small studies have recently suggested that there are specific plasma metabolic signatures in chronic obstructive pulmonary disease (COPD), but there have been no large comprehensive study of metabolomic signatures in COPD that also integrate genetic variants. Materials and Methods: Fresh frozen plasma from 957 non-Hispanic white subjects in COPDGene was used to quantify 995 metabolites with Metabolon's global metabolomics platform. Metabolite associations with five COPD phenotypes (chronic bronchitis, exacerbation frequency, percent emphysema, post-bronchodilator forced expiratory volume at one second [FEV1]/forced vital capacity [FVC], and FEV1 percent predicted) were assessed. A metabolome-wide association study was performed to find genetic associations with metabolite levels. Significantly associated single-nucleotide polymorphisms were tested for replication with independent metabolomic platforms and independent cohorts. COPD phenotype-driven modules were identified in network analysis integrated with genetic associations to assess gene-metabolite-phenotype interactions. Results: Of metabolites tested, 147 (14.8%) were significantly associated with at least 1 COPD phenotype. Associations with airflow obstruction were enriched for diacylglycerols and branched chain amino acids. Genetic associations were observed with 109 (11%) metabolites, 72 (66%) of which replicated in an independent cohort. For 20 metabolites, more than 20% of variance was explained by genetics. A sparse network of COPD phenotype-driven modules was identified, often containing metabolites missed in previous testing. Of the 26 COPD phenotype-driven modules, 6 contained metabolites with significant met-QTLs, although little module variance was explained by genetics. Conclusion: A dysregulation of systemic metabolism was predominantly found in COPD phenotypes characterized by airflow obstruction, where we identified robust heritable effects on individual metabolite abundances. However, network analysis, which increased the statistical power to detect associations missed previously in classic regression analyses, revealed that the genetic influence on COPD phenotype-driven metabolomic modules was modest when compared with clinical and environmental factors.

Bronchoalveolar Lavage Fluid from COPD Patients Reveals More Compounds Associated with Disease than Matched Plasma.

Smoking causes chronic obstructive pulmonary disease (COPD). Though recent studies identified a COPD metabolomic signature in blood, no large studies examine the metabolome in bronchoalveolar lavage (BAL) fluid, a more direct representation of lung cell metabolism. We performed untargeted liquid chromatography-mass spectrometry (LC-MS) on BAL and matched plasma from 115 subjects from the SPIROMICS cohort. Regression was performed with COPD phenotypes as the outcome and metabolites as the predictor, adjusted for clinical covariates and false discovery rate. Weighted gene co-expression network analysis (WGCNA) grouped metabolites into modules which were then associated with phenotypes. K-means clustering grouped similar subjects. We detected 7939 and 10,561 compounds in BAL and paired plasma samples, respectively. FEV1/FVC (Forced Expiratory Volume in One Second/Forced Vital Capacity) ratio, emphysema, FEV1 % predicted, and COPD exacerbations associated with 1230, 792, eight, and one BAL compounds, respectively. Only two plasma compounds associated with a COPD phenotype (emphysema). Three BAL co-expression modules associated with FEV1/FVC and emphysema. K-means BAL metabolomic signature clustering identified two groups, one with more airway obstruction (34% of subjects, median FEV1/FVC 0.67), one with less (66% of subjects, median FEV1/FVC 0.77; p < 2 × 10-4). Associations between metabolites and COPD phenotypes are more robustly represented in BAL compared to plasma.