A Machine Learning Method for Detecting Autocorrelation of Evolutionary Rates in Large Phylogenies

Area of Science:

• Evolutionary Biology
• Genomics
• Bioinformatics

Background:

• Speciation predicts similar evolutionary rates between ancestors and descendants, leading to rate autocorrelation.
• Previous molecular data analyses have not confirmed this expectation, potentially due to limitations in detection methods.

Purpose of the Study:

• To introduce CorrTest, a novel machine learning method for detecting rate autocorrelation in large phylogenies.
• To assess the prevalence of rate autocorrelation across diverse taxa.

Main Methods:

• Developed CorrTest, a computationally efficient machine learning approach.
• Applied CorrTest to DNA and amino acid sequence data from various organisms.

Main Results:

• CorrTest demonstrates superior performance compared to existing methods.
• Extensive rate autocorrelation was detected in mammals, birds, insects, metazoans, plants, fungi, parasitic protozoans, and prokaryotes.
• Rate autocorrelation is a widespread phenomenon across the tree of life.

Conclusions:

• The findings suggest a concordance between molecular and non-molecular evolutionary patterns.
• Rate autocorrelation is a common evolutionary pattern, impacting phylogenetic analyses.
• This study will facilitate more accurate dating of the tree of life.