MixTreEM

MixTreEM reconstructs species trees from genomic sequence data by combining a mixture-model structural expectation-maximization phase with PrIME-DLRS DLRS-based selection to account for gene duplication and loss.

Key Features:

• Two-Phase Approach: MixTreEM uses a structural expectation-maximization algorithm based on a mixture model to generate candidate species trees from monocopy gene family sequence data, and PrIME-DLRS applies the DLRS (Duplication, Loss, and Reconciliation) model to select the optimal species tree.
• Handling of Gene Duplications and Losses: Incorporates modeling of gene duplication and loss to manage discordance between species trees and gene trees.
• Scalability and Efficiency: Addresses scalability limitations of probabilistic coestimation methods and provides accuracy comparable to PHYLDOG with substantially faster processing.
• Comparative Performance: Evaluated on synthetic and biological data, it outperforms Duptree in accuracy and the standalone MixTreEM phase offers a rapid alternative for species tree selection on large datasets.

Scientific Applications:

• Phylogenetics: Reconstruction of species phylogenies in datasets affected by gene duplication and loss.
• Comparative Genomics: Genome-scale species tree inference using monocopy gene family sequence data.
• Evolutionary Biology: Analysis of evolutionary relationships accounting for gene tree discordance driven by duplication and loss events.

Methodology:

MixTreEM applies a structural expectation-maximization algorithm based on a mixture model to generate candidate species trees from monocopy gene family sequences, followed by PrIME-DLRS using the DLRS model to select the optimal species tree.