Distinguishing Felsenstein Zone from Farris Zone Using Neural Networks

Area of Science:

• Evolutionary biology
• Bioinformatics
• Computational phylogenetics

Background:

• Phylogenetic tree reconstruction is crucial for understanding evolutionary relationships.
• Maximum likelihood and maximum parsimony are widely used but can yield conflicting results.
• Long-branch attraction and repulsion can complicate phylogenetic analyses.

Purpose of the Study:

• To develop a neural network method for distinguishing phylogenetic reconstruction challenges.
• To assess the suitability of maximum likelihood versus maximum parsimony methods.
• To provide insights into contentious evolutionary cases like Strepsiptera.

Main Methods:

• Training a neural network on four-taxon alignments simulated under different evolutionary conditions.
• Evaluating the neural network's ability to identify conditions prone to long-branch attraction or repulsion.
• Applying the neural network to discordant phylogenies from maximum likelihood and parsimony analyses.

Main Results:

• The neural network successfully distinguishes between alignment conditions susceptible to long-branch attraction and repulsion.
• The method provides guidance on choosing between maximum likelihood and maximum parsimony when results conflict.
• Analysis of Strepsiptera evolution using the neural network supports its placement with beetles.

Conclusions:

• Neural networks offer a novel approach to improving phylogenetic tree reconstruction accuracy.
• This method can help resolve phylogenetic ambiguities and disputes.
• The findings reinforce the current understanding of Strepsiptera evolutionary placement.