Learning Causal Networks from Time Course Data
Author Information
Author(s): Rainer Opgen-Rhein, Korbinian Strimmer
Primary Institution: Ludwig-Maximilians-Universität München
Hypothesis
Can a novel model selection procedure improve the estimation of vector autoregressive (VAR) networks from small sample genomic data?
Conclusion
The proposed procedure allows for efficient statistical learning of large-scale VAR causal models, even in challenging genomic data situations.
Supporting Evidence
- The proposed shrinkage method achieved around 80% true discovery rate in simulations.
- The method was applied to real-world data from Arabidopsis thaliana, identifying 7,381 significant edges.
- The approach is computationally efficient, taking about one minute for large datasets.
Takeaway
This study shows a new way to understand how genes interact over time, even when we have very little data.
Methodology
The study developed a two-step procedure for estimating VAR networks, involving improved estimation of regression coefficients and model selection through partial correlation testing.
Limitations
The method assumes equidistant time points and may not be applicable to all types of biological data.
Digital Object Identifier (DOI)
Want to read the original?
Access the complete publication on the publisher's website