Journal: International journal of obesity (2005)

Article Title: Integrating untargeted metabolomics, genetically informed causal inference, and pathway enrichment to define the obesity metabolome

doi: 10.1038/s41366-020-0603-x

Figure Lengend Snippet: (a) OE and MCDS metabolomics datasets, matched using PAIRUP-MS, were used to identify known and unknown metabolites associated with BMI. (b) Independent genetic instruments (G M ) for the BMI-associated metabolites were selected using OE and MCDS data, and then used to test for a metabolite-to-BMI (M → B) causal effect in UKB; in parallel, BMI genetic instrument (G B ), a polygenic risk score built using GIANT BMI-associated SNPs (G b ) and UKB effect estimate weights (β b ), was used to test for a BMI-to-metabolite (B → M) causal effect in OE and MCDS. (c) A subset of metabolites was categorized into “cause”, “effect”, and “bidirectional” groups based on the magnitude of the evidence according to the G M and G B IV effect estimate p -values, reflecting different types of causal connections between the metabolites and BMI. (d) Pathway analyses of the three metabolite groups were performed using metabolite set annotations generated using PAIRUP-MS and an independent dataset (BioAge). Y ~ X, regression of Y on X; U, unmeasured confounder; n , number of samples; m , number of metabolites; k , number of known (or shared known) metabolites; s , number of metabolite sets.

Article Snippet: The PAIRUP-MS pathway annotation method and BioAge metabolite data were used to generate metabolite set annotations as described previously ( – ), resulting in 690 metabolite sets in which each metabolite was assigned a numeric membership score in each set.

Techniques: Generated

Journal: International journal of obesity (2005)

Article Title: Integrating untargeted metabolomics, genetically informed causal inference, and pathway enrichment to define the obesity metabolome

doi: 10.1038/s41366-020-0603-x

Figure Lengend Snippet: Genetic associations for 100 shared known (top) or 477 matched (bottom) BMI-associated metabolites were consolidated to plot the best p -value for each SNP (i.e. only the p -value for the best associated metabolite was plotted for each SNP). Genome-wide significance threshold ( p < 5 × 10 −8 ) is marked by the orange lines. Genome-wide significant SNPs are plotted in red or blue, for shared known or matched metabolites, respectively. Lead SNPs of the most significant loci ( p < 1 × 10 −15 ) are annotated with nearest genes (within 5kb), along with the best associated known metabolites if applicable.

Article Snippet: The PAIRUP-MS pathway annotation method and BioAge metabolite data were used to generate metabolite set annotations as described previously ( – ), resulting in 690 metabolite sets in which each metabolite was assigned a numeric membership score in each set.

Techniques: Genome Wide

Journal: International journal of obesity (2005)

Article Title: Integrating untargeted metabolomics, genetically informed causal inference, and pathway enrichment to define the obesity metabolome

doi: 10.1038/s41366-020-0603-x

Figure Lengend Snippet: BMI-associated known metabolites classified into cause, effect, or bidirectional group based on their IV effect estimate p -values. Y ~ X, regression of Y on X; B, BMI; M, metabolite; covariate adjustment for B and M described in Methods ; β, effect size estimate; SNP, hg19 chromosome:position is shown; EA, effect allele (i.e. metabolite level-increasing allele). IV effect estimate p -values < 0.05 are in bold italic.

Article Snippet: The PAIRUP-MS pathway annotation method and BioAge metabolite data were used to generate metabolite set annotations as described previously ( – ), resulting in 690 metabolite sets in which each metabolite was assigned a numeric membership score in each set.

Techniques:

Journal: International journal of obesity (2005)

Article Title: Integrating untargeted metabolomics, genetically informed causal inference, and pathway enrichment to define the obesity metabolome

doi: 10.1038/s41366-020-0603-x

Figure Lengend Snippet: Euclidean distance-based hierarchical clustering was performed using metabolite membership ranks in the BioAge-based metabolite set annotations. Each column is a shared known (red label) or matched (black label) metabolite from the cause (yellow bar) or effect (blue bar) metabolite group. Each row is an enriched ( p < 0.05) metabolite set in pathway analysis in either the cause (yellow bar) or effect (blue bar) direction (with representative pathway name shown in label; see for full pathway list). Larger number in membership rank (darker red) indicates higher membership score. Dashed light blue boxes highlight the two major cause and effect clusters according to the clustering dendrograms.

Article Snippet: The PAIRUP-MS pathway annotation method and BioAge metabolite data were used to generate metabolite set annotations as described previously ( – ), resulting in 690 metabolite sets in which each metabolite was assigned a numeric membership score in each set.

Techniques:

Journal: Oxidative Medicine and Cellular Longevity

Article Title: Metabolomic Identification of Serum Exosome-Derived Biomarkers for Bipolar Disorder

doi: 10.1155/2022/5717445

Figure Lengend Snippet: Dysregulated serum exosomal metabolite coexpression modules in patients with BD. (a) Dendrogram showing metabolite coexpression modules defined in 72 samples (32 BD patients and 40 HC subjects). (b) Pearson's correlation coefficient between gender, age, disease status, disease severity, and module eigengene. (c, d) Coexpression hub node (metabolite) network plots for blue and turquoise modules. (e) KEGG enrichment pathways for the 20 hub nodes (metabolites) in blue and turquoise modules.

Article Snippet: Briefly, the public database of metabolite information and metabolomics data curation environment (MetWare) was performed to achieve qualitative analysis of the first- and second-order mass spectra.

Techniques:

Journal: Frontiers in Cellular and Infection Microbiology

Article Title: Analysis of the gut microbiota in children with gastroesophageal reflux disease using metagenomics and metabolomics

doi: 10.3389/fcimb.2023.1267192

Figure Lengend Snippet: Aberrant metabolic patterns in GERD patients (A) Differential metabolite cluster heatmaps; (B) Matchstick map of the differentially abundant metabolites; (C) Differentially abundant metabolite KEGG pathway annotation diagram; (D) Bubble map of the KEGG enrichment analysis.

Article Snippet: Principal component analysis (PCA) and partial least squares discriminant analysis (PLS-DA) were performed after the metabolite data were transformed by MetaX software.

Techniques:

Journal: Clinical and Translational Medicine

Article Title: Multi‐omics analysis revealed biomarkers for coronary atherosclerosis: Occurrence and development

doi: 10.1002/ctm2.70451

Figure Lengend Snippet: Changes in gut microbiome and fecal metabolites. (A) Scatter plot with a linear regression line and heatmap depicting Log2FC‐based alterations of the significantly different genus in the gut microbiome with different severity of coronary atherosclerosis, p < .05. Log2FC: log2 (fold change). (B) Differential analysis of the gut microbiota at the genus level. (C) Regression curve depicting the significantly different genus with different severity of coronary atherosclerosis, from control development into AS0, AS1, and AS2. (D) Changes and differences in fecal metabolites. (E) Correlation between gut microbiota and fecal metabolites (Spearman's correlation test, *: p < .05, **: p < .01).

Article Snippet: For the training cohort, we built a random forest machine learning model based on plasma metabolites, gut microbe, and clinical features (single/multi‐omics data) and discovered the key features associated with CA.

Techniques: Control

Journal: Clinical and Translational Medicine

Article Title: Multi‐omics analysis revealed biomarkers for coronary atherosclerosis: Occurrence and development

doi: 10.1002/ctm2.70451

Figure Lengend Snippet: Comparison of the plasma metabolites. (A) Venn diagram showing the overlaps among altered metabolites ( p < .05, compared Ctr vs. AS0, Ctr vs. AS1, and Ctr vs. AS2). (B) Heatmap showing significantly different metabolites in all AS groups compared Ctr. (C) Heatmap showing relative content of significantly different metabolites when comparing Ctr vs. AS0, Ctr vs. AS1, and Ctr vs. AS2, *: p < .05, **: p < .01. (D) Heatmap showing significantly different metabolites compared coronary atherosclerosis severity, Log2FC: log2 (fold change), *: p < .05, **: p < .01. (E) Enrichment pathway related to coronary atherosclerosis severity.

Article Snippet: For the training cohort, we built a random forest machine learning model based on plasma metabolites, gut microbe, and clinical features (single/multi‐omics data) and discovered the key features associated with CA.

Techniques: Comparison, Clinical Proteomics

Journal: Clinical and Translational Medicine

Article Title: Multi‐omics analysis revealed biomarkers for coronary atherosclerosis: Occurrence and development

doi: 10.1002/ctm2.70451

Figure Lengend Snippet: Accuracy score of random forest classification algorithm for predicting the class using omics data. (A) Single omics. (B) Multi‐omics combination of top features from each omics. (C) Receiver operating characteristic (ROC) curve for prediction of coronary atherosclerosis based on Top 5 feature from plasma metabolites. (D) ROC curve for prediction of coronary atherosclerosis based on combination of P (5), C (5), and G (5). (E) Top 15 features from the model with the highest accuracy, P (5) + C (5) + G (5).

Article Snippet: For the training cohort, we built a random forest machine learning model based on plasma metabolites, gut microbe, and clinical features (single/multi‐omics data) and discovered the key features associated with CA.

Techniques: Biomarker Discovery, Clinical Proteomics

Journal: Clinical and Translational Medicine

Article Title: Multi‐omics analysis revealed biomarkers for coronary atherosclerosis: Occurrence and development

doi: 10.1002/ctm2.70451

Figure Lengend Snippet: Target plasma metabolites identify high‐performing diagnostic and prognostic biomarkers for coronary atherosclerosis. Receiver operating characteristic (ROC) analysis to discriminate of Ctr, AS0, AS1, and AS2 groups. A: Ctr vs. AS0, AUC = .979. B: Ctr vs. AS1, AUC = .998. C: Ctr vs. AS2, AUC = 1.000. D: AS0 vs. AS1, AUC = .933. E: AS1 vs. AS2, AUC = .965. F: AS0 vs. AS2, AUC = 1.000.

Article Snippet: For the training cohort, we built a random forest machine learning model based on plasma metabolites, gut microbe, and clinical features (single/multi‐omics data) and discovered the key features associated with CA.

Techniques: Clinical Proteomics, Diagnostic Assay