Development and validation of asthma risk prediction models using co-expression gene modules and machine learning...

Area of Science:

• Genomics and Bioinformatics
• Respiratory Medicine
• Molecular Biology

Background:

• Asthma is a complex respiratory disease with poorly understood genetic underpinnings.
• Gene expression profiling is a powerful tool for dissecting the molecular basis of complex diseases like asthma.

Purpose of the Study:

• To identify candidate genes and pathways involved in asthma pathogenesis using gene expression data.
• To develop and validate machine learning models for asthma classification and prediction based on gene signatures.

Main Methods:

• Differentially expressed genes (DEGs) analysis and weighted gene co-expression network analysis (WGCNA) were performed on airway epithelial cells (AECs) and nasal epithelial cells (NECs).
• Machine learning algorithms, including least absolute shrinkage and selection operator (LASSO) and support vector machine (SVM), were employed to identify gene signatures and build predictive models.
• Models were validated using independent datasets from bronchial epithelial cells (BECs), airway smooth muscle (ASM), and whole blood (WB).

Main Results:

• Gene signatures derived from AECs and NECs demonstrated high accuracy (AUC=1) in discriminating asthmatic subjects from controls.
• Validation in BECs, ASM, and WB showed significant diagnostic performance (AUCs ranging from 0.66 to 0.82).
• Functional annotation revealed enrichment in IL-13, PI3K/AKT, and apoptosis signaling pathways, highlighting key molecular mechanisms.

Conclusions:

• Epithelium-derived gene signatures provide a robust and potentially non-invasive approach for asthma classification and prediction.
• The identified gene signatures, including SERPINB2 and CTSC, offer insights into asthma pathogenesis and can serve as biomarkers.
• This epithelium-based model can act as a surrogate for less accessible tissues, advancing the molecular understanding of asthma.