From interaction networks to interfaces, scanning intrinsically disordered regions using AlphaFold2

From interaction networks to interfaces, scanning intrinsically disordered regions using AlphaFold2

18 January 2024 | Hélène Bret, Jinmei Gao, Diego Javier Zea, Jessica Andreani & Raphaël Guerois
The study explores the application of AlphaFold2 in predicting protein-protein interactions, particularly those involving intrinsically disordered regions (IDRs). Using a dataset of 42 protein-peptide complexes, the researchers found that AlphaFold2-Multimer, when provided with full-length protein sequences, achieved a 40% success rate in identifying the correct interface. By segmenting the proteins into smaller fragments and integrating evolutionary information, the success rate increased to 90%. This method was also effective on a larger dataset of 923 protein-peptide interactions from the ELM database. The study highlights the importance of fragment scanning and the combination of different multiple sequence alignment (MSA) strategies for improving prediction accuracy. Additionally, the research discusses the challenges and limitations of using AlphaFold2 confidence scores to discriminate between binding partners, especially in cases with small interaction motifs. The findings underscore the potential of AlphaFold2 in advancing the understanding of complex protein interaction networks.The study explores the application of AlphaFold2 in predicting protein-protein interactions, particularly those involving intrinsically disordered regions (IDRs). Using a dataset of 42 protein-peptide complexes, the researchers found that AlphaFold2-Multimer, when provided with full-length protein sequences, achieved a 40% success rate in identifying the correct interface. By segmenting the proteins into smaller fragments and integrating evolutionary information, the success rate increased to 90%. This method was also effective on a larger dataset of 923 protein-peptide interactions from the ELM database. The study highlights the importance of fragment scanning and the combination of different multiple sequence alignment (MSA) strategies for improving prediction accuracy. Additionally, the research discusses the challenges and limitations of using AlphaFold2 confidence scores to discriminate between binding partners, especially in cases with small interaction motifs. The findings underscore the potential of AlphaFold2 in advancing the understanding of complex protein interaction networks.
Reach us at info@study.space