A Multi-Omics Framework for Decoding Disease Mechanisms: Insights From Methylmalonic Aciduria

Abstract

The diverse perspectives offered by multi-omics data analysis can aid in identifying the most relevant molecular pathways involved in disease processes, and findings in one layer can substantiate findings in other layers of information. Integrating data from multiple omics sources is becoming increasingly important to improve disease diagnosis and treatment, especially for conditions with complex and poorly understood underlying pathomechanisms. Methylmalonic aciduria (MMA), an inherited metabolic disorder, serves as an illustrative example of such a disease with poorly understood pathogenesis for which published multi-omics data are readily available. Reusing these FAIR data, obtained from the multi-omics digitization of 230 individuals (210 patients with MMA and 20 controls), we pursued advanced data integration and analysis strategies to integrate different levels of biological information, combining genomic, transcriptomic, proteomic, and metabolomic profiling with biochemical and clinical data, with the aim of elucidating molecular perturbations in individuals affected by MMA. The analysis of protein-quantitative trait loci highlighted the importance of glutathione metabolism in the pathogenesis of MMA. This finding was supported by correlation network analyses that integrated proteomics and metabolomics data, alongside gene set enrichment and transcription factor analyses based on disease severity from transcriptomic data. The correlation network analysis also revealed that lysosomal function is compromised in patients with MMA, which is critical for maintaining metabolic balance. Our research introduces a comprehensive data analysis framework that effectively addresses the challenge of prioritizing disruptions in molecular pathways by accumulating evidence from multiple omics levels.

Keywords: correlation network analysis; gene set enrichment analysis; methylmalonic aciduria; module analysis; multi-omics data integration; pQTL; quantitative trait loci; transcription factor enrichment analysis.

Graphical abstract

Fig. 1

Workflow-Chart. The flowchart depicts our analytical workflow for identifying potential contributors and pathways involved in the pathophysiology of MMA. Multi-omics data were previously generated at the genome, transcriptome, proteome, and metabolome levels. These data modalities were integrated through pQTL mapping and enrichment analysis, in conjunction with correlation network and gene set enrichment and transcription factor analysis.

Fig. 2

Cis-pQTL distribution in patients with methylmalonic aciduria. A, distribution of gene regions (upper panel) and gene types (lower panel) among cis-pQTL variants. B, SNP density plot across chromosomes showing the number of SNPs in 1 Mb windows: each horizontal bar corresponds to a human chromosome, the x-axis indicates chromosome length in megabases (Mb), and the color blocks denote the count of SNPs per 1 Mb window, with a gradient from green to red where darker shades represent a higher SNP density in that genomic region. C, Circos plots of cis-pQTL variants: pQTL variants are listed according to their chromosomal location. Proteins with genome-wide significant pQTLs are listed in the right semicircle (blue color). D, significantly enriched pathways from the pathway enrichment analysis conducted on all proteins in (C). The Sankey plot on the left displays the enriched pathways and their related proteins. The y-axis depicts enriched pathways. The dot size in the bubble plot on the right indicates the number of protein hits, the color of the dots corresponds to the p-value.

Fig. 3

Protein co-expression modules in patients with methylmalonic aciduria.A, schematic showing the number and type of proteomics samples used in our module co-expression analysis. B, Bubble heatmap illustrating the results of the module enrichment analysis. Circle size and color both represent the normalized enrichment score, as determined by CEMiTool. Only modules with an adjusted p-value <0.05 between conditions are shown. C, the top five significantly enriched pathways from the pathway enrichment analysis performed on all modules identified in B are shown. The height of each bar corresponds to the -log10(p) value. D, the enrichment network for proteins in module M3, where each node represents an enriched term. Node sizes correspond to the number of proteins linked to each term, with the top 20 enriched terms highlighted. Different colors distinguish various clusters, and terms within the same cluster are positioned near each other, showing their functional similarity. E, Bar chart of the top 20 enriched pathways in the module M5 colored by p-value. F and G, visualization of protein interactions in the co-expression modules M3 and M5, red if derived from the interaction file (STRING database), blue if proteins are module hubs, and green if both conditions apply. Node size reflects the degree of connection strength as determined by CEMiTool. Pathways discussed in the manuscript are indicated in bold.

Fig. 4

Metabolites co-expression modules in patients with methylmalonic aciduria. A, Left side: schematic drawing showing the number and type of metabolomics samples used in our module co-expression analysis. Right side: bubble heatmap depicting the results of module enrichment analysis, highlighting module activities in affected individuals compared to controls. Circle size and color both represent the normalized enrichment score, as determined by CEMiTool. Only modules that are significantly different (adjusted p < 0.05) across conditions are shown. B, significantly enriched pathways from the pathway enrichment analysis conducted on module 4 (M4) identified in (A). The Sankey plot on the left displays the enriched pathways and their related metabolites. The dot size of the bubble plot on the right indicates the number of metabolite hits, the color of the dots corresponds to the p-value. C, all metabolites from M4 are displayed using Cytoscape. The top 10 hub metabolites in M4 are depicted as red circles (inner rim). The degree of connectivity of all metabolites, as calculated by CEMiTool, is indicated through a color scale ranging from red (high) to yellow (low).

Fig. 5

Gene set enrichment and transcription factor analyses of the transcriptomics dataset in MMA. Gene Set Enrichment Analysis (GSEA) revealed enrichment in gene sets related to the (A) TCA cycle and (B) glutathione metabolism pathway based on transcriptomics data. The top panel displays the running enrichment score (ES), which reflects the cumulative enrichment of the pathway by ranking the genes in the order of their relevance or impact. The peak ES represents the highest degree of enrichment. The bottom panel indicates the gene positions within the ranked gene list. The nominal p-value was calculated using the permutation test in GSEA. In panel (C), the blue bars represent the transcription factors (TFs) that showed significant changes in relation to the clinical severity score (CSS). Blue stars mark TFs associated with glutathione metabolism, as supported by literature, while red stars denote TFs related to the tricarboxylic acid (TCA) cycle, also supported by literature. D, the heatmap displays the transcription factor activity scores computed using VIPER, with each row representing a transcription factor and each column representing a sample. Red indicates upregulation of TF’s target genes, while blue indicates overall downregulation. The CSS for each patient sample is depicted as a gradient from yellow to red, with deeper colors signifying more severe disease.

Fig. 6

Glutathione pathway analysis. An integrated view of the changes in the KEGG glutathione metabolism pathway based on the pQTLs, transcriptomic, proteomic, and metabolomic changes. Ellipses represent metabolites, while squares represent SNPs/transcripts/proteins. Colors indicate the changes between MMA (mut⁰) and the control group (see also supplemental Figs. S3 and S6–S8), with green representing downregulation, orange representing upregulation, and blue representing no significant change.