Network Diffusion on a Multiscale Interactome Reveals Novel Herbal Candidates and Bioactive Compounds Converging on Shared Effectors in Anemia
Won-Yung LeeIntroduction/Objective:
Anemia is a prevalent hematological disorder driven by heterogeneous mechanisms—iron deficiency, chronic inflammation, impaired erythropoiesis, and hemolysis—that single-target therapies often fail to address. We systematically prioritized candidate herbs and their active compounds for anemia using a multiscale interactome–based network pharmacology framework.
Methods:
Herb–compound–target data were aggregated from OASIS, DrugBank, TTD, and STITCH; anemia disease targets came from DisGeNET. We generated diffusion profiles using a biased random walk with restart on the tripartite multiscale interactome, and ranked herbs by cosine similarity to the anemia profile. Hypergeometric testing, GO/KEGG enrichment, compoundlevel subnetwork analysis, and HERB-based transcriptome validation followed.
Results:
The top 10 prioritized herbs—led by Swertiae Herba, Geranii Herba, and Salviae Miltiorrhizae Radix—were enriched in TNF/NF-κB, HIF-1, and JAK-STAT/PI3K-Akt pathways. Four high-ranking herbs and seven constituent compounds (e.g., swertiamarin, p-hydroxybenzoic acid, physcion, salvianolic acids A/B) had not previously been proposed as therapeutic candidates for anemia. Transcriptome validation of tanshinone IIA implicated the cell cycle, PI3KAkt, and FoxO pathways and recovered GDF15 and CXCL12.
Discussion:
Across all subnetworks, divergent compound–target architectures converged on shared downstream effectors (PLEC, ITGB4, ELANE, TNF), supporting a multi-target paradigm. Mechanistic profiles further suggested subtype relevance: swertiamarin to inflammationassociated anemia; p-hydroxybenzoic acid to telomere/bone-marrow-failure anemias; physcion/epicatechin to oxidative-stress and hemolytic anemias.
Conclusion:
The framework offers a reproducible route to nominating multi-target herbal therapeutics for anemia, identifying novel candidates that warrant experimental validation.