Single-cell Sequence Analysis Combined with Multiple Machine Learning to Identify Markers in Sepsis Patients: LILRA5

Area of Science:

• Immunology
• Genomics
• Computational Biology

Background:

• Sepsis, a life-threatening condition from bacterial infection, involves immune dysregulation.
• Current research often lacks single-cell resolution, limiting understanding of specific cell subsets and disease markers.
• Investigating sepsis at the single-cell transcriptome level is crucial for identifying novel diagnostic and therapeutic targets.

Purpose of the Study:

• To identify sepsis-specific cell phenotypes and gene markers using single-cell RNA sequencing.
• To elucidate cellular communication networks and regulatory relationships in sepsis.
• To validate the diagnostic value of identified marker genes and explore potential therapeutic interventions.

Main Methods:

• Single-cell peripheral blood sequencing of sepsis patients and healthy controls.
• Machine learning algorithms (LASSO, SVM, RF) for marker gene identification.
• Cellular communication analysis ('cell chat') and Bayesian networks for regulatory relationships.
• External dataset validation and molecular docking for therapeutic target assessment.

Main Results:

• Identified distinct sepsis and healthy cell phenotypes, focusing on monocytes with inflammatory profiles.
• Discovered LILRA5 as a robust sepsis marker gene with high diagnostic accuracy (AUCs up to 0.980) across multiple datasets and clinical samples.
• Elucidated LILRA5 co-expression networks and demonstrated atenolol's potential to target LILRA5 for sepsis treatment via molecular docking and survival analyses.

Conclusions:

• LILRA5 is a validated diagnostic marker for sepsis, offering improved single-cell resolution insights.
• Atenolol demonstrates potential as a therapeutic agent for sepsis by targeting LILRA5.
• This study provides a foundation for developing targeted therapies for sepsis based on single-cell transcriptomic data.