Episode 12: Phylogenetics with the arborists part 2

The microbinfie podcast delves into the complex world of phylogenetic modeling, exploring the nuanced approaches to constructing evolutionary trees and understanding genetic relationships across different biological systems.

In this episode we talk about models of evolution help estimate transitions between states (e.g., nucleotide mutations) and fit the evolutionary patterns to build phylogenetic trees.

Guests

• Dr. Conor Meehan
• Dr. Leo Martins

Types of Models

• Broad Categories: Models can vary based on:
  • Transition probabilities (e.g., from one nucleotide to another).
  • Covariance among sites (heterogeneity across sites).
  • Specific models discussed include:
    • CAT Models: Different implementations for nucleotide or amino acid data.
    • Gamma-Distributed Models: Addressing rate variation across sites.
    • Best Model Selection: Determined by model testing rather than a one-size-fits-all approach. The best model depends on the dataset.

Historical Context

• Model Evolution: Discussion on the evolution of phylogenetic models, such as DNAML and Fast DNAML, leading to the development of RAxML and other software.

Specific Models

• HKY Model: Favored for its complexity while providing an analytic solution.
• GTR Model: Commonly used for complex datasets but can lead to overparameterization.
• Indel Models: Need for models that account for insertion/deletion events, like TKF91/92 and Poisson-Indel models, highlighting that many current practices do not include indels in phylogenetic analysis.

Recombination Challenges

• Impact of Recombination: Acknowledgment that recombination complicates tree-building, particularly in recombination-rich species.
• Recommendations:
  • Avoid tree building with high recombination data.
  • Consider phylogenetic networks to understand relationships better.
  • Utilize tools like Bacter in a Bayesian framework for more accurate reconstructions.

The best phylogenetic approach should involve careful model selection based on the data, consideration of recombination, and a focus on using models that accurately represent the evolutionary processes at play.