Atlas of predicted protein complex structures across kingdoms

Area of Science:

• Structural Biology
• Bioinformatics
• Evolutionary Biology

Background:

• Protein complexes are crucial for biological processes.
• Existing protein-protein interaction (PPI) data lacks comprehensive structural characterization across species.
• This gap hinders understanding of molecular mechanisms and evolutionary relationships.

Purpose of the Study:

• To create a large-scale, cross-kingdom atlas of predicted protein-protein interaction structures.
• To identify conserved structural motifs and evolutionary events.
• To provide a resource for enhancing downstream applications like protein binding prediction.

Main Methods:

• Utilized the AlphaFold2-based ColabFold framework for structure prediction.
• Generated a dataset of 1.1 million predicted protein-protein interaction structures.
• Performed structural clustering and comparative analyses with existing databases.

Main Results:

• Identified 181,671 high-confidence protein complex structures, including 37,855 in the human interactome.
• Discovered conserved protein complex architectures shared across bacteria, archaea, humans, mice, and plants.
• Uncovered widespread gene fusion and fission events during evolution and identified potential viral receptors.

Conclusions:

• The comprehensive atlas provides an extensive cross-kingdom resource for structural biology and bioinformatics.
• The findings offer insights into conserved biological functions and evolutionary processes.
• The dataset serves as a valuable tool for advancing protein binding prediction and biomedical research.