AFragmenter: schema-free, tuneable protein domain segmentation for AlphaFold protein structures

Journal article

Summary: Protein domain segmentation is a crucial aspect of understanding protein functions and interactions, and it is vital for protein modelling exercises and evolutionary studies. Current segmentation methods often rely on predefined classification schemes, leading to inconsistencies and biases. AFragmenter provides a schema-free and tuneable approach to protein domain segmentation based on network analysis of AlphaFold-predicted structures. Utilizing Predicted Aligned Error values, AFragmenter constructs a fully connected network of protein residues and identifies distinct structural domains by using Leiden clustering. This method empowers users to adjust parameters including contrast threshold and resolution, providing control over the segmentation process. Availability and implementation: AFragmenter is implemented in Python3 and freely available under an MIT license. It can be found as a Python library and command line tool at https://github.com/sverwimp/AFragmenter, pip, and Conda.

Authors: Verwimp, S; Lavigne, R; Lood, C; van Noort, V

• Publication status: Published
• Peer review status: Peer reviewed
• Publisher: Oxford University Press
• Journal: Bioinformatics
• Volume: 41
• Issue: 11
• Article number: btaf588
• Publication date: 2025-10-27
• Acceptance date: 2025-10-21
• DOI: 10.1093/bioinformatics/btaf588
• EISSN: 1367-4811
• ISSN: 1367-4803
• Language: English