Integrated-omics analysis with explainable deep networks on pathobiology of infant bronchiolitis

Area of Science:

• Pediatric Respiratory Medicine
• Systems Biology
• Molecular Medicine

Background:

• Bronchiolitis is a primary cause of infant hospitalization, yet its underlying molecular mechanisms are not fully understood.
• Identifying molecular networks is crucial for understanding disease severity and developing targeted therapies.

Purpose of the Study:

• To integrate nasopharyngeal transcriptome and metabolome data to uncover molecular networks driving bronchiolitis pathobiology.
• To identify biomarkers and potential therapeutic targets for infant bronchiolitis.

Main Methods:

• Integrated nasopharyngeal transcriptome and metabolome data from 397 infants hospitalized with bronchiolitis.
• Employed an explainable deep network model to identify an omics-cluster associated with disease severity.
• Performed network analyses to elucidate molecular pathways and identify potential drug candidates.

Main Results:

• An omics-cluster of 401 transcripts and 38 metabolites significantly distinguished bronchiolitis severity (AUC, 0.828).
• Innate immunity metabolites, such as ceramides, were central to the identified network, linked to toll-like receptor (TLR) and NF-κB signaling.
• Network analysis identified eight modules and 50 existing drug candidates, including prostaglandin I2 analogs, for repurposing.

Conclusions:

• This study successfully identified key molecular networks underlying infant bronchiolitis.
• The findings highlight the role of innate immunity and TLR signaling in disease severity.
• The identified drug candidates, like iloprost, offer potential for novel anti-inflammatory treatment strategies.