Unicore Enables Scalable and Accurate Phylogenetic Reconstruction with Structural Core Genes

Area of Science:

• Genomics
• Bioinformatics
• Structural Biology

Background:

• Single-copy core genes are crucial for phylogenetics and genome quality assessment.
• Traditional methods rely on amino acid similarity, limiting deep clade analysis.
• AI-driven 3D protein structure prediction is computationally intensive for proteomic scales.

Purpose of the Study:

• Introduce Unicore, a scalable method for identifying structural core genes.
• Enable efficient phylogenetic analysis using structural information.
• Overcome limitations of existing ortholog identification methods.

Main Methods:

• Utilize ProstT5 to generate 3Di structural strings from proteomes, reducing runtime.
• Employ Foldseek for clustering universally present single-copy structures.
• Apply FoldMason for structural alignment and project back to amino acid data.

Main Results:

• Unicore achieves significant speedups (over 1000x) compared to full 3D prediction.
• Demonstrates linear runtime scaling with the number of species.
• Performs up to seven times faster than OrthoFinder while yielding congruent phylogenetic results.

Conclusions:

• Unicore provides a fast and scalable approach for structural core gene identification.
• The method is universally applicable across diverse taxa, including superkingdoms.
• Unicore enhances phylogenetic reconstruction and genome quality assessment capabilities.