Identification of useful genes from multiple microarrays for ulcerative colitis diagnosis based on machine learning...

Area of Science:

• Genomics and Bioinformatics
• Immunology
• Gastroenterology

Background:

• Ulcerative colitis (UC) is a chronic inflammatory bowel disease with increasing global incidence.
• Current UC diagnosis relies on clinical symptoms and laboratory tests.
• Gene expression signatures are linked to UC severity, suggesting potential diagnostic biomarkers.

Purpose of the Study:

• To assess the diagnostic potential of gene expression in ulcerative colitis (UC).
• To investigate the correlation between specific genes and immune regulation in UC.
• To identify reliable gene biomarkers for UC diagnosis.

Main Methods:

• Meta-analysis of ten microarray datasets (387 UC patients, 139 healthy controls).
• Application of six machine learning algorithms (SVM, LASSO, RF, GBM, PCA, NN) for gene selection.
• Validation using receiver operating characteristic (ROC) analysis and correlation with immune cell types.

Main Results:

• Identified 87 differentially expressed genes, narrowing down to six for model establishment.
• Two genes, OLFM4 and C4BPB, demonstrated high diagnostic accuracy (AUC > 0.8) for UC.
• OLFM4 and C4BPB showed significant correlations with six immune cell types (Macrophages M1/M2, Mast cells, Monocytes, NK cells).

Conclusions:

• OLFM4 and C4BPB are promising biomarkers for diagnosing ulcerative colitis (UC).
• These genes may play a role in UC-associated immune dysregulation.
• Further validation studies are recommended to confirm the clinical utility of OLFM4 and C4BPB.