Simple k-RF Metrics for Comparison of Labeled DAGs.

Area of Science:

• Computational Biology
• Bioinformatics
• Systems Biology

Background:

• Labeled directed acyclic graphs (DAGs) are vital for modeling causal relationships, particularly in biological and healthcare contexts like tumor progression.
• Comparing these DAGs is essential for advancing inference and evaluation methods for DAG models.

Purpose of the Study:

• To introduce novel dissimilarity measures for comparing labeled DAGs.
• To enhance existing comparison methods by refining the k-Robinson-Foulds distance.

Main Methods:

• The study refines the k-Robinson-Foulds distance, originally for labeled trees, to apply to labeled DAGs.
• New measures are based on comparing local node-induced multisets of labels within DAGs.

Main Results:

• The proposed dissimilarity measures effectively compare labeled DAGs.
• These measures can handle DAGs with differing label sets without requiring auxiliary nodes or label removal.

Conclusions:

• The developed dissimilarity metrics offer a robust approach for comparing labeled DAGs.
• This advancement is critical for the development and evaluation of DAG-based models in biology and healthcare.