Discovering Fragile Clades and Causal Sequences in Phylogenomics by Evolutionary Sparse Learning

Area of Science:

• Evolutionary biology
• Bioinformatics
• Genomics

Background:

• Phylogenomic analyses use extensive genetic data to reconstruct organismal relationships.
• Inferred evolutionary relationships can be unreliable due to the influence of a few specific sequences.
• Current methods lack direct ways to identify fragile clades or the responsible gene sequences.

Purpose of the Study:

• To develop novel metrics for assessing gene-species sequence concordance and clade probability.
• To provide a streamlined method for identifying sequences that may distort phylogenetic reconstructions.
• To enhance the statistical confidence and accuracy of phylogenomic analyses.

Main Methods:

• Introduction of evolutionary sparse learning models.
• Development of new metrics for gene-species sequence concordance.
• Development of new metrics for clade probability.
• Validation using diverse phylogenomic datasets (fungi, plants, animals).

Main Results:

• The new metrics effectively pinpoint fragile clades within phylogenies.
• Specific gene sequences responsible for distorting phylogenetic inferences were identified.
• The approach demonstrated high accuracy across different taxonomic groups.
• The methodology avoids the need for extensive re-analysis with alternative models or data subsets.

Conclusions:

• The developed metrics offer a robust and efficient way to evaluate inferred clades in phylogenomics.
• This method aids in identifying and addressing sequences that compromise phylogenetic accuracy.
• The approach streamlines the process of validating large-scale phylogenomic datasets.