The best of both worlds: Phylogenetic eigenvector regression and mapping

Area of Science:

• Phylogenetics
• Evolutionary Biology
• Comparative Genomics

Background:

• Eigenfunction analyses are crucial for modeling autocorrelation in time, space, and phylogeny.
• Phylogenetic Eigenvector Regression (PVR) uses principal coordinate analysis of phylogenetic distances.
• Phylogenetic Eigenvector Mapping (PEM) incorporates model-based warping using Ornstein-Uhlenbeck processes.

Purpose of the Study:

• Compare Phylogenetic Eigenvector Regression (PVR) and Phylogenetic Eigenvector Mapping (PEM).
• Evaluate their performance in estimating phylogenetic signal, correlated evolution, and phylogenetic imputation.
• Assess their utility with simulated data under alternative evolutionary models.

Main Methods:

• Utilized simulated data to compare PVR and PEM.
• Applied Principal Coordinate Analysis for PVR.
• Incorporated Ornstein-Uhlenbeck (O-U) process fitting for PEM before eigenvector extraction.
• Assessed phylogenetic signal, correlated evolution, and imputation accuracy.

Main Results:

• PEM demonstrated slightly higher prediction ability compared to PVR.
• PEM was found to be more general than the original PVR approach.
• Both methods showed similarities, but PEM offered enhanced performance and broader applicability.

Conclusions:

• PEM represents a significant advancement over PVR, offering improved predictive power and generality.
• PEM integrates the strengths of empirical eigenfunction analysis with the insights from evolutionary models.
• This approach provides a powerful, flexible tool for phylogenetic comparative analyses.