Abstract
BACKGROUND: Autoimmune and autoinflammatory diseases (AIIDs) are characterized by significant heterogeneity and comorbidities, complicating their mechanisms and classification. Disease associations studies, or diseasome, facilitate the exploration of disease mechanisms and development of novel therapeutic strategies. However, the diseasome for AIIDs is still in its infancy. To address this gap, we developed a novel framework that utilizes multi-modal data and biomedical ontologies to explore AIID associations. METHODS: We curated disease terms from Mondo/DO/MeSH/ICD, and three specialized AIID knowledge bases, creating an integrated repository of 484 autoimmune diseases (ADs), 110 autoinflammatory diseases (AIDs), and 284 associated diseases. By leveraging genetic, transcriptomic (bulk and single-cell), and phenotypic data, we built multi-layered AIID association networks and an integrated network supported by cross-scale evidence. Our ontology-aware disease similarity (OADS) strategy incorporates not only multi-modal data, but also continuous biomedical ontologies. RESULTS: Network modularity analysis identified 10 robust disease communities and their representative phenotypes and dysfunctional pathways. Focusing on 10 highly concerning AIIDs, such as Behçet's disease and Systemic lupus erythematosus, we provide insights into the information flow from genetic susceptibilities to transcriptional dysregulation, alteration in immune microenvironment, and clinical phenotypes, and thus the mechanisms underlying comorbidity. For instance, in systemic sclerosis and psoriasis, dysregulated genes like CCL2 and CCR7 contribute to fibroblast activation and the infiltration of CD4+ T and NK cells through IL-17 signaling pathway, PPAR signaling pathway, leading to skin involvement and arthritis. CONCLUSION: These findings enhance our understanding of AIID pathogenesis, improving disease classification and supporting drug repurposing and targeted therapy development.