Single-cell Ca

Area of Science:

• Computational Biology
• Systems Biology
• Genomics

Background:

• Single-cell genomic technologies provide unprecedented insights into cellular heterogeneity.
• Integrating gene expression and dynamic signaling data in single cells remains a challenge for parameter inference.

Purpose of the Study:

• To develop Bayesian methods for parameter inference using joint single-cell gene expression and Ca2+ dynamics data.
• To leverage transfer learning for efficient parameter inference across populations of single cells.
• To investigate the relationship between transcriptional states and cellular signaling dynamics.

Main Methods:

• Bayesian parameter inference framework.
• Transfer learning approach to share information between sequentially analyzed single cells.
• Dynamical modeling of intracellular Ca2+ signaling.
• Analysis of thousands of single cells with heterogeneous responses.

Main Results:

• Transfer learning accelerates parameter inference for cell sequences, irrespective of cell ordering.
• Ordering cells by transcriptional similarity is crucial for distinguishing Ca2+ dynamic profiles and associated marker genes.
• Inference reveals complex cellular heterogeneity, with divergent parameter covariation in intracellular versus intercellular contexts.

Conclusions:

• Single-cell parameter inference, informed by transcriptional similarity, can quantify relationships between gene expression and signaling dynamics.
• The developed methods enhance the utility of single-cell data for understanding cell dynamics and heterogeneity.
• Transcriptional similarity ordering is key for robustly identifying cell-specific signaling behaviors and molecular markers.