Machine-learning-based integrative -'omics analyses reveal immunologic and metabolic dysregulation in environmental...

Area of Science:

• Gastroenterology and Hepatology
• Computational Biology
• Biochemistry

Background:

• Environmental enteric dysfunction (EED) is a subclinical enteropathy with diagnostic challenges due to overlapping features with other inflammatory conditions.
• Accurate diagnosis of EED is crucial for understanding its impact on gut health and development.

Purpose of the Study:

• To develop a machine-learning-based image analysis model for classifying EED.
• To identify multi-omic signatures correlating with EED histology and clinical data.
• To characterize metabolic alterations in EED using computational and lipidomic approaches.

Main Methods:

• Development of a machine-learning classification model using histologic features from EED subjects (Pakistan cohort) and controls.
• Correlation analysis of histologic features with transcriptomics and clinical biomarkers.
• In-silico metabolic network modeling and untargeted lipidomics to assess metabolic flux alterations.

Main Results:

• A machine-learning model was developed to classify EED based on histologic features.
• Significant alterations in fatty acid and glycerophospholipid metabolism were identified in EED.
• Specific lipid changes (increased phosphatidylcholine, LPCs; decreased ester-linked LPCs) and elevated plasma bile acids were observed in EED patients.

Conclusions:

• The study elucidates a multi-omic signature for environmental enteric dysfunction.
• Machine learning and integrated omics approaches provide novel insights into EED pathophysiology.
• Findings may contribute to improved diagnostic strategies for EED.