Multiomics integration of single-cell transcriptomics and bulk sequencing data identifies key biomarkers and predictive models for IBD subtype classification.

Introduction: Inflammatory bowel disease (IBD), comprising ulcerative colitis (UC) and Crohn's disease (CD), represents a diagnostic challenge owing to its overlapping clinical features. Precise molecular characterization is essential for improved subtype classification and the development of reliable diagnostic tools. Methods: We integrated peripheral blood single-cell RNA sequencing (scRNA-seq) data (GSE125527) and bulk sequencing data (GSE75214) from UC and CD patients to identify molecular patterns distinguishing the two subtypes through cross-dataset integration analysis. Machine learning algorithms and network analysis were applied to identify hub genes and construct diagnostic models. Biomarker robustness was evaluated using an independent validation dataset (GSE179285), followed by immunohistochemical (IHC) confirmation in clinical samples. Results: Single-cell analysis revealed distinct immune signatures: UC patients exhibited IL-17A + effector memory T cells and enriched leukocyte migration pathways, whereas CD patients were characterized by IL-1β-producing immune cells and B cell activation processes. Cross-dataset integration identified 43 consistently dysregulated genes and 10 hub genes (THBS1, PLAUR, KLF4, CD36, CD44, CXCR4, FOS, S100A9, ANXA1, and TIMP1), spanning immune response, transcriptional regulation, metabolism, and tissue remodeling functions. An 18-gene machine learning model achieved an area under the curve (AUC) of 0.73 in independent validation. Notably, LDHB, MGAT4A, and PSME2 emerged as uniquely powerful biomarkers with limited prior reporting in UC/CD classification. IHC validation confirmed subtype-specific expression, with GZMA and PSME2 preferentially elevated in CD, and LDHB and FOSB preferentially elevated in UC. Discussion: This multiomics approach revealed robust molecular signatures distinguishing UC from CD, demonstrating moderate-to-good diagnostic capability (AUC = 0.73). The findings illuminate fundamental differences in immune regulation, metabolism, and tissue remodeling between IBD subtypes, providing a promising foundation for blood-based diagnostic tool development. Prospective clinical validation is warranted before implementation in routine practice.