High-precision binary trait association on phylogenetic trees

Area of Science:

• Microbial genomics
• Computational biology
• Population genetics

Background:

• Traditional microbial genome-wide association study (mGWAS) methods struggle with bacterial genome complexity, leading to high false discovery rates, low statistical power, and computational limitations.
• Existing mGWAS approaches often fail to accurately detect negative interactions and are not scalable for pangenome-wide analyses.

Purpose of the Study:

• To introduce Simulation-based Phylogenetic iNteraction Inference (SimPhyNI), a computationally optimized framework designed for efficient and rigorous mGWAS.
• To develop a method that overcomes the limitations of existing mGWAS approaches, particularly in identifying both positive and negative genetic associations in bacterial populations.

Main Methods:

• SimPhyNI constructs null co-occurrence distributions by simulating traits using phylogenetically informed parameters, including time to first event.
• The framework utilizes constrained variation in simulations and log odds ratio scoring to robustly identify associations.
• The method was validated using synthetic datasets and applied to real-world datasets, including the entire *Escherichia coli* pangenome.

Main Results:

• SimPhyNI demonstrates high precision and recall in identifying both positive and negative gene-gene and gene-trait associations using synthetic data.
• The framework successfully detected interactions within phage defense systems in *Escherichia coli*.
• Application to the *E. coli* pangenome involved over 9 million tests, showcasing SimPhyNI's scalability and performance.

Conclusions:

• SimPhyNI provides a computationally efficient and statistically rigorous framework for mGWAS, significantly advancing the field.
• The method's ability to accurately detect diverse genetic interactions enables deeper understanding of microbial function, ecology, and disease.
• SimPhyNI's design supports future extensions to various trait types, broadening its applicability in microbial genomics research.