AI and Rare Diseases ── Platforms ── 2026-11-20

Rare-disease diagnosis and drug discovery platforms and AI
- Abstract
BACKGROUND: Kawasaki Disease (KD) is an acute, self-limited vasculitis of childhood, with its etiopathogenesis intricately tied to systemic inflammation and endothelial dysfunction. Disruption of circadian rhythms and mitochondrial transport, 2 fundamental homeostatic systems, are increasingly recognized as independent contributors to vascular pathology. However, their integrated molecular crosstalk in KD remains completely uncharted. OBJECTIVE: The primary objective was to identify and validate a mitochondrial transport andcircadian rhythm (MRC)-related molecular signature in KD. The secondary objectives were to resolve MRC-defined molecular subtypes, to nominate and characterize a candidate hub gene, to define its cellular distribution, and to prioritize candidate therapeutic compounds. METHODS: First, Peripheral blood bulk transcriptomic datasets of KD patients were analyzed. GSE68004 was interrogated via Limma, GSVA, and WGCNA to isolate MRC-associated shared DEGs. Second, non-negative matrix factorization (NMF) and 3 machine learning algorithms (RF, Lasso and SVM-RFE) were deployed in GSE63881 for molecular subtype identification and diagnostic model construction, with independent diagnostic validation in integrated GSE73463 and GSE73462. Next, leveraging single-cell transcriptomic data from peripheral blood samples of KD patients (GSE24757), we resolved the spatiotemporal dissemination and functional evolution of the hub gene. In addition, deep learning-driven drug screening (DrugReflector) and molecular docking nominated a candidate compound in GSE18606. Finally, differential expression of the hub gene was validated using q-RT-PCR in clinical samples from KD patients. RESULTS: The integrated MRC-related signature robustly stratified KD patients into 2 distinct molecular subgroups. FNDC5 was identified as an NK-distributed and MRC-related pathogenic factor in KD. BRD-K10482608 was nominated as a computationally predicted candidate targeting FNDC5 for KD. CONCLUSION: This study provides the first pathway-anchored MRC-related signature in KD together with a validated framework for molecular stratification. FNDC5 is put forward as a prioritized, hypothesis-generating candidate of KD vasculopathy, and BRD-K10482608 as a computationally nominated starting point for future functional and nursing pharmacological testing.
Journal IF-equivalent: 4.1 (OpenAlex 2-year mean citedness, value as of 2026-11-19, retrieved 2026-11-20; not the official Clarivate IF)Reference: Yang Y. Deciphering mitochondrial dysfunction and circadian rhythm-related pathogenic patterns for the pathogenesis of Kawasaki Disease and corresponding nursing therapeutic enrichment: Evidence from machine learning-driven multi-omics and clinical validation. Computational Biology and Chemistry. 2027 Feb;126:109575. doi:10.1016/j.compbiolchem.2026.109575.Checked: Abstract only