Ghost interactions: revealing missing protein-ligand interactions using AlphaFold predictions
Escobedo, N.; Saldano, T. E.; Mac Donagh, J.; Rodriguez Sawicki, L.; Palopoli, N.; Fernandez Alberti, S.; Fornasari, M. S.; Parisi, G.
Show abstract
Protein-ligand interactions represent an essential step in understanding molecular recognition, an intense field of research for many scientific areas. Structural biology has played a central role in unveiling protein-ligand interactions, but current techniques are still not able to reliably describe the interactions of ligands with highly flexible regions. In this work we explored the capacity of AlphaFold2 (AF2) to estimate the presence of interactions between ligands and residues belonging to disordered regions, which we called "ghost interactions" as they are missing in the crystallographic derived structures. We found that AF2 models are good predictors of regions associated with order-disorder transitions. Additionally, we found that AF2 predicts residues making ghost interactions with ligands, which are mostly buried and show a differential evolutionary conservation. Our findings could fuel current areas of research that consider intrinsically disordered proteins as potentially valuable targets for drug development, given their biological relevance and associated diseases.
Matching journals
The top 6 journals account for 50% of the predicted probability mass.
Similar papers in this journal
Similar papers in this journal
Similar papers in this journal
- Intrinsically disordered protein ensembles shape evolutionary rates revealing conformational patterns 97%
- Gradations in protein dynamics captured by experimental NMR are not well represented by AlphaFold2 models and other computational metrics 96%
- Conformational variation in enzyme catalysis: A structural study on catalytic residues 95%
Similar papers in this journal
- PRosettaC: Rosetta based modeling of PROTAC mediated ternary complexes 96%
- Fast Local Alignment of Protein Pockets (FLAPP): A system-compiled program for large-scale binding site alignment 96%
- Are Deep Learning Structural Models Sufficiently Accurate for Free Energy Calculations? Application of FEP+ to AlphaFold2 Predicted Structures 96%
"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.