Time varying causal network reconstruction of a mouse cell cycle

Area of Science:

• Systems Biology
• Molecular Biology
• Computational Biology

Background:

• Biochemical networks are typically studied using static or time-averaged data.
• Temporal variations in macromolecules are crucial for understanding network dynamics and causal mechanisms.
• Analyzing time-series data for complex systems to identify temporal regimes and causal networks is challenging.

Purpose of the Study:

• To develop and apply methods for analyzing temporal transcriptional data to identify dynamic networks and mechanisms in the cell cycle.
• To extract phase-specific causal interactions of cell cycle genes and temporal interdependencies of biological mechanisms.
• To provide a comprehensive picture of molecular crosstalk within a cell cycle.

Main Methods:

• Utilized Granger causality, Vector Autoregression, Estimation Stability with Cross Validation (ES-CV), and a nonparametric change point detection algorithm.
• Applied methods to RNA-sequencing (RNA-seq) time-course data from Mouse Embryonic Fibroblast (MEF) primary cells over nearly two cell cycles.
• Employed Least Absolute Shrinkage and Selection Operator (LASSO) for gene interaction analysis.

Main Results:

• The change-point detection algorithm precisely determined the timing and duration of cell cycle phases.
• Identified phase-specific causal interactions among cell cycle genes and temporal interdependencies of biological mechanisms.
• Successfully extracted dynamic network information without requiring prior biological knowledge.

Conclusions:

• The developed model captures temporal dependencies of cellular components beyond existing literature.
• Inferred dynamic interplay of intracellular mechanisms can predict time-varying cellular responses.
• Provides insights for designing precise experiments to modulate cell cycle regulation.