A fast and robust statistical test based on likelihood ratio with Bartlett correction to identify Granger causality...

Area of Science:

• Bioinformatics
• Computational Biology
• Systems Biology

Background:

• Identifying causal relationships in complex biological systems, such as gene expression time series, is crucial for understanding cellular mechanisms.
• Existing methods for Granger causality detection in gene expression data may lack speed or statistical power, especially under non-ideal data distributions.

Purpose of the Study:

• To propose and evaluate a novel likelihood ratio test (LRT) with Bartlett correction for identifying Granger causality in gene expression time series.
• To compare the performance of the proposed LRT against a previously established bootstrap-based approach.

Main Methods:

• Development of a likelihood ratio test (LRT) incorporating Bartlett correction for Granger causality analysis.
• Comparative performance analysis of the LRT against a bootstrap-based Granger causality test.
• Implementation of both methods in an R package named gGranger.

Main Results:

• The proposed LRT demonstrates significantly faster computation times compared to the bootstrap-based method.
• LRT exhibits superior statistical power, maintaining effectiveness even with non-normally distributed gene expression data.
• The gGranger R package provides a readily available tool for implementing these Granger causality tests.

Conclusions:

• The LRT with Bartlett correction offers an efficient and powerful solution for Granger causality detection in gene expression time series.
• This method provides a valuable advancement for systems biology research, enabling more robust inference of gene regulatory networks.
• The availability of the gGranger package facilitates the application of these advanced statistical methods by researchers.