AlphaFold-Multimer Predictions: Which Scores Best Identify True Protein-Protein Interactions in the TGF-β Activation Network?
Year: 2026
Type: conference-paper
Abstract: Protein-protein interactions (PPIs) play a central role in cellular processes. Changes in PPIs, such as mutations or the absence of a partner, can lead to cellular dysfunction and diseases. Identifying these interactions and obtaining their structures is therefore crucial for the development of therapeutic strategies. Recent advances in computational methods, notably AlphaFold and, by extension, AlphaFold-Multimer, have enabled large-scale prediction of protein complexes from sequence information. The predicted structure corresponds to the most plausible structure based on the sequences, however it does not ensure that the predicted structure is a true interaction. The state-of-the-art approach consists of applying thresholds to the scores associated with structure predictions to distinguish the likely interactions from those that are not. In this study, we evaluated whether if some of these scores can be used to distinguish experimentally validated PPIs from other predicted interactions, using our expertise on the PPI network involved in the activation of the profibrotic factor, TGF-β.
Source: HAL
Language: English
Subfield: Molecular Biology
Domain: Life Sciences
Open Access status: green
OpenAlex ID: W7208267173