Dissecting Incongruence between Concatenation- and Quartet-Based Approaches in Phylogenomic Data

Area of Science:

• Phylogenetics and phylogenomics
• Evolutionary biology
• Bioinformatics

Background:

• Topological conflict is common in phylogenomic datasets, arising from both concatenation and coalescent-based methods.
• Characterizing the causes of incongruence between different phylogenetic signals (likelihood-based vs. quartet-based) is challenging.

Purpose of the Study:

• To investigate incongruence in phylogenomic data by comparing likelihood-based and quartet-based topological signals for individual genes.
• To identify genes that exhibit conflicting phylogenetic signals and assess their impact on overall tree reconstruction.

Main Methods:

• Analyzed three phylogenomic datasets from animals, fungi, and plants, comparing IQ-TREE (T1) and ASTRAL (T2) phylogenies.
• Quantified gene-wise conflict using the difference in log-likelihood scores (ΔGLS) and quartet scores (ΔGQS).
• Compared characteristics of inconsistent and consistent genes, including evolutionary rate, gene tree topology, branch lengths, hidden paralogy, and gene tree discordance.

Main Results:

• Found that 30-36% of genes in each dataset were inconsistent, showing conflicting signals between likelihood and quartet scores.
• Inconsistent genes were more likely to be discordant with both major topologies (T1 and T2) and exhibited higher gene tree discordance.
• Simulations showed that removing inconsistent genes reduced incongruence and increased accuracy, especially under low incomplete lineage sorting (ILS) and gene tree estimation error (GTEE).

Conclusions:

• A significant proportion of genes in phylogenomic datasets can harbor conflicting phylogenetic signals.
• Identifying and removing these inconsistent genes is a viable strategy to improve the accuracy and congruence of species phylogenies.
• The effectiveness of removing inconsistent genes depends on the levels of ILS and GTEE within the dataset.