Protein complex structure prediction powered by multiple sequence alignments of interologs from multiple taxonomic...

Area of Science:

• Computational biology
• Structural biology
• Bioinformatics

Background:

• AlphaFold2 accurately predicts protein structures but requires multiple sequence alignments (MSAs) of interologs for protein-protein interactions (PPIs).
• Generating effective MSAs for PPIs, especially across different species, remains a challenge.

Purpose of the Study:

• To develop and benchmark a simplified phylogeny-based approach for generating MSAs of interologs for AlphaFold2-based protein complex structure prediction.
• To improve the accuracy and success rate of predicting protein complex structures.

Main Methods:

• A simplified phylogeny-based approach was used to generate MSAs of interologs.
• These MSAs were used as input for AlphaFold2 to predict protein complex structures.
• The protocol was benchmarked on a nonredundant dataset of 107 bacterial and 442 eukaryotic PPIs.
• Interologs were further restricted to specific taxonomic ranks to assess impact on prediction success.

Main Results:

• The phylogeny-based MSA generation method significantly outperformed existing approaches.
• Successful prediction rates were 79.5% for bacterial PPIs and 49.8% for eukaryotic PPIs using the initial method.
• Restricting interologs to specific taxonomic ranks increased success rates to 87.9% for bacterial and 56.3% for eukaryotic PPIs.
• Predicted template modeling (TM) scores can guide the selection of optimal taxonomic ranks.

Conclusions:

• The developed phylogeny-based method provides a robust and effective strategy for generating MSAs for protein complex structure prediction with AlphaFold2.
• Taxonomic rank restriction is crucial for optimizing prediction accuracy, particularly for eukaryotic PPIs.
• This approach enhances the utility of AlphaFold2 for studying protein-protein interactions across diverse species.