Inferring Time-Lagged Causality Using the Derivative of Single-Cell Expression.

Area of Science:

• Computational Biology
• Genomics
• Systems Biology

Background:

• Existing computational methods for gene causality inference often use cross-sectional data, potentially missing time-lagged relationships.
• Single-cell RNA sequencing (scRNA-seq) data offers a powerful resource but has limitations in capturing dynamic processes.

Purpose of the Study:

• To introduce a novel method, causal inference with time-lagged information (CITL), for inferring time-lagged gene causality from scRNA-seq data.
• To address the limitations of current methods in identifying dynamic gene regulatory interactions.

Main Methods:

• CITL infers time-lagged causal relationships by assessing conditional independence between changing and current gene expression levels.
• RNA velocity is utilized to estimate the changing expression levels of genes.
• The method was validated using simulation data and compared against leading computational approaches.

Main Results:

• CITL demonstrated accuracy and stability in inferring time-lagged causality on simulation data.
• Application to real scRNA-seq data identified 878 pairs of time-lagged causal relationships.
• The number of regulatory relationships identified by CITL was significantly greater than expected by chance.

Conclusions:

• CITL effectively infers time-lagged gene causality from scRNA-seq data, overcoming limitations of existing methods.
• The method provides a valuable tool for uncovering complex gene regulatory networks.
• An R package and command-line tool are available for broader application.