Excluding Loci With Substitution Saturation Improves Inferences From Phylogenomic Data

Area of Science:

• Evolutionary Biology
• Bioinformatics
• Genomics

Background:

• Nucleotide sequence evolution is affected by repeated substitutions, leading to substitution saturation.
• Substitution saturation obscures historical phylogenetic signals, posing challenges for deep-time evolutionary studies.
• Existing methods struggle to accurately assess substitution saturation in large genomic datasets.

Purpose of the Study:

• To introduce a novel statistical test for detecting substitution saturation in nucleotide sequences.
• To evaluate the performance of this new test using both simulated and empirical phylogenomic data.
• To assess the impact of substitution saturation on phylogenetic inference and propose mitigation strategies.

Main Methods:

• Development and application of an entropy-based statistical test for substitution saturation.
• Analysis of 36 empirical phylogenomic datasets to identify saturated loci.
• Simulation studies to evaluate test sensitivity to various evolutionary scenarios.

Main Results:

• The proposed entropy saturation test successfully detected substitution saturation in approximately 50% of loci across empirical datasets.
• Saturation was more prevalent in loci exhibiting discordant phylogenetic signals and fewer informative sites.
• The test proved sensitive to factors known to mislead phylogenetic inference, such as high evolutionary rates and tree imbalance.

Conclusions:

• The entropy saturation test is a valuable tool for identifying problematic substitution saturation in phylogenomic data.
• Excluding saturated loci can significantly improve the accuracy of deep-time phylogenetic reconstructions.
• This approach helps mitigate the negative effects of multiple substitutions on evolutionary inferences.