CASTER: Direct species tree inference from whole-genome alignments

Area of Science:

• Evolutionary biology
• Genomics
• Phylogenetics

Background:

• Genomes exhibit mosaic evolutionary histories, complicating the accurate inference of the tree of life.
• Current phylogenomic methods often exclude large portions of genomes due to modeling and scalability constraints.
• There is a need for advanced methods to infer species trees directly from multiple genome alignments, accounting for discordance.

Purpose of the Study:

• To introduce a novel, theoretically grounded method for species tree inference that directly utilizes multiple genome alignments.
• To address the limitations of existing methods in handling genomic discordance and scalability.
• To provide a tool capable of analyzing whole-genome data for accurate phylogenetic reconstruction.

Main Methods:

• Developed Coalescence-Aware Alignment-Based Species Tree Estimator (CASTER), a site-based phylogenomic method.
• CASTER eliminates the requirement for pre-defining recombination-free loci.
• The method is designed for scalability to hundreds of mammalian whole genomes.

Main Results:

• Demonstrated the accuracy and scalability of CASTER through simulations incorporating recombination.
• Applied CASTER to diverse biological datasets, showcasing its performance on real-world genomic data.
• Showcased CASTER's ability to identify biological and artifactual patterns of genomic discordance using per-site scores.

Conclusions:

• CASTER offers a theoretically justified and scalable approach to phylogenomic inference from whole-genome alignments.
• The method effectively handles genomic discordance, improving the accuracy of tree of life reconstruction.
• CASTER's per-site analysis provides valuable insights into evolutionary processes and potential data artifacts across the genome.