Bayesian hierarchical models for protein networks in single-cell mass cytometry

Area of Science:

• Proteomics
• Systems Biology
• Cellular Biology

Background:

• Understanding protein functional networks is crucial for deciphering cellular mechanisms.
• Single-cell proteomics offers high-resolution insights into protein expression within individual cells.
• Mass cytometry generates large-scale, multi-marker protein expression data from single cells.

Purpose of the Study:

• To develop and apply hierarchical Bayesian models for analyzing single-cell proteomics data.
• To investigate protein functional networks and identify differential networks across experimental conditions.
• To visualize and interpret the links between experimental agents and targeted proteins.

Main Methods:

• Utilized a novel dataset from single-cell mass cytometry experiments.
• Applied hierarchical Bayesian models to model protein expression in tens of thousands of single cells.
• Developed a method for novel visualization of differential networks based on posterior inference.

Main Results:

• Successfully constructed protein functional networks under various experimental conditions.
• Identified and characterized differential networks, highlighting condition-specific molecular interactions.
• Presented a novel visualization method for direct observation of agent-target protein relationships.

Conclusions:

• The proposed Bayesian models provide a powerful tool for studying molecular interactions at the single-cell level.
• The method enables the identification of differential protein functional networks.
• The novel visualization facilitates the understanding of experimental agent effects on protein targets.