Genome-wide epistasis and co-selection study using mutual information.

Area of Science:

• Population genomics
• Computational biology
• Evolutionary genetics

Background:

• Covariance-based methods are increasingly used to detect selection and epistasis in population genomics.
• Previous methods successfully identified selection patterns in bacterial populations but faced scalability challenges.
• Understanding co-selective pressures is crucial for deciphering complex evolutionary dynamics.

Purpose of the Study:

• Introduce SpydrPick, a computationally efficient, model-free method for discovering polymorphisms under co-selective pressure.
• Develop a method that can analyze large-scale pan-genome data.
• Provide a novel visualization tool for exploring co-evolutionary signals.

Main Methods:

• Developed SpydrPick, a model-free approach for detecting co-selected polymorphisms.
• Incorporated an efficient correction for population structure, independent of explicit phylogenetic trees.
• Implemented a Manhattan plot-like visualization for rapid signal exploration.

Main Results:

• SpydrPick demonstrates computational efficiency suitable for pan-genome analyses.
• The method effectively corrects for population structure without requiring phylogenetic trees.
• Application to Streptococcus pneumoniae and Neisseria meningitidis identified known and novel targets of co-selection, including those linked to virulence and antibiotic resistance.

Conclusions:

• SpydrPick offers a scalable and efficient solution for identifying co-selective pressures in large bacterial datasets.
• The method's ability to detect co-evolutionary signals, even without phenotypic data, holds significant potential for molecular discovery.
• The new visualization enhances the exploration and interpretation of complex genomic selection patterns.