Multicondition and multimodal temporal profile inference during mouse embryonic development

Area of Science:

• Computational Biology
• Genomics
• Developmental Biology

Background:

• Single-cell measurements offer insights into cellular dynamics but face challenges due to disruptive sampling and mismatched time points.
• Integrating multimodal (e.g., scRNA-seq, scATAC-seq) and multicondition (e.g., sex, batch) single-cell data across time is complex.

Purpose of the Study:

• To introduce Sunbear, a joint modeling framework for integrating multicondition and multimodal single-cell profiles over time.
• To enable imputation of temporal profile changes, alignment of profiles across datasets and modalities, and extrapolation of missing data.

Main Methods:

• Development of a novel joint modeling framework named Sunbear.
• Application of Sunbear to analyze mouse embryonic development datasets.
• Utilizing Sunbear for imputation, alignment, and extrapolation tasks.

Main Results:

• Sunbear successfully integrated multicondition and multimodal single-cell data across time.
• The framework revealed sex-biased transcription patterns during mouse embryonic development.
• Dynamic relationships between epigenetic priming and transcription were predicted for cells lacking multimodal profiles.

Conclusions:

• Sunbear provides a powerful approach for analyzing complex single-cell time-series data.
• The framework facilitates the projection of single-cell snapshots into comprehensive multimodal and multicondition views of cellular trajectories.
• Sunbear enhances our understanding of developmental processes and gene regulation by overcoming data integration challenges.