AF2Ring: A Memory Efficient Method for Predicting Quaternary Structure of Cyclic Protein Oligomers

Area of Science:

• Biochemistry
• Structural Biology
• Computational Biology

Background:

• Cyclic protein oligomers are crucial for biological processes.
• Predicting protein quaternary structure is vital for protein design.
• Current deep learning methods for structure prediction are resource-intensive.

Purpose of the Study:

• To develop a memory-efficient method for predicting quaternary structures of cyclic protein assemblies.
• To facilitate large-scale validation in protein design.

Main Methods:

• AF2Ring utilizes AlphaFold2 with a reduced number of input chains.
• Partial prediction results are used to construct the complete quaternary structure.

Main Results:

• AF2Ring achieves accurate quaternary structure prediction for cyclic protein assemblies.
• Significantly reduced memory consumption and computation time compared to standard methods.
• Maintained prediction accuracy comparable to full AlphaFold2 runs.

Conclusions:

• AF2Ring offers an efficient solution for predicting cyclic protein structures.
• The method supports protein design by enabling large-scale validation and accommodating longer sequences.