Figure 1.

Visualization and functional characterization of the human protein–lipid–metabolite interaction network. (A) Network integration of multi-omics data embedded in the hyperbolic space (see Section 2 for details). Node colors represent different molecular classes (proteins in black, lipids in cyan, metabolites in red). The circular sectors indicate clusters (numbered 1 to 15) identified by the gaps between consecutive nodes in the angular dimension of the hyperbolic space (see Section 2 for details). Node coordinates are available in Supplementary Table S1. Edges (not represented) are available in Supplementary Table S2). (B) The over-represented biological function in each cluster was determined via GO enrichment analysis (BP: biological process) to reveal the biological relevance based on the term with the lowest adjusted p-value of enrichment (see Section 2 for details). The numbers in parentheses indicate how many proteins had the indicated GO term out of all the proteins in the respective cluster. (C) The distribution of θ coordinates for each omics layer: proteins, lipids, and metabolites. The different maxima positions of lipids and metabolites in the angular dimension indicate that they are grouped in different clusters. The angular coordinates of the 15 clusters are indicated with a colored bar for reference under the plot. (D) Brain-related proteins, lipids, and metabolites in the hyperbolic map (identifiers from UniProt, SLM, and PubChem, respectively).