Detection of circular permutations by Protein Language Models

Area of Science:

• Biochemistry
• Computational Biology
• Structural Biology

Background:

• Protein circular permutations (CP) are important for understanding protein evolution and function.
• Existing detection methods, including sequence-based and structure-based approaches, have limitations in accuracy and scope.
• Protein Language Models (PLMs) show promise for sequence-based analysis but struggle with CP detection due to linear sequence dependency.

Purpose of the Study:

• To develop a novel method for accurately detecting protein circular permutations.
• To overcome the limitations of existing sequence alignment algorithms in identifying CP.
• To leverage the power of PLMs combined with genetic principles for improved CP detection.

Main Methods:

• Developed plmCP, a novel approach integrating classical genetic principles with advanced PLM-based alignment techniques.
• The method is designed to circumvent the linear sequence order dependency inherent in traditional alignment algorithms.
• Employs PLMs to effectively utilize sequence information without requiring structural input.

Main Results:

• plmCP successfully addresses the challenges faced by current PLM-based alignment tools in detecting circular permutations.
• The integration of genetic knowledge enables the method to overcome sequence order dependency.
• Demonstrates effective detection of circular permutations, enhancing capabilities in protein research.

Conclusions:

• plmCP offers a significant advancement in the detection of protein circular permutations.
• The method's ability to handle structural flexibility and avoid sequence order dependency broadens its applicability.
• Contributes to enhanced protein research and engineering by providing a robust tool for CP identification.