Harnessing protein folding neural networks for peptide–protein docking

By Tomer Tsaban, Julia K. Varga, O. Avraham, Ziv Ben-Aharon, A. Khramushin, O. Schueler‐Furman. 2021

299 citations recorded by the source corpus.

Highly accurate protein structure predictions by deep neural networks such as AlphaFold2 and RoseTTAFold have tremendous impact on structural biology and beyond. Here, we show that, although these deep learning approaches have originally been developed for the in silico folding of protein monomers, AlphaFold2 also enables quick and accurate modeling of peptide–protein interactions. Our simple implementation of AlphaFold2 generates peptide–protein complex models without requiring multiple sequence alignment information for the peptide partner, and can handle binding-induced conformational changes of the receptor. We explore what AlphaFold2 has memorized and learned, and describe specific examples that highlight differences compared to state-of-the-art peptide docking protocol PIPER-FlexPepDock. These results show that AlphaFold2 holds great promise for providing structural insight into a wide range of peptide–protein complexes, serving as a starting point for the detailed characterization and manipulation of these interactions. AlphaFold2 has originally been developed to provide highly accurate predictions of protein monomer structures. Here, the authors present a simple adaptation of AlphaFold2 that enables structural modeling of peptide–protein complexes, and explore the underlying mechanisms and limitations of this approach.

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

Citations in the corpus

AlphaFold2 and its applications in the fields of biology and medicine (2023, background): “For example, Tsaban et al. 146 suggested an AF2-based strategy to model peptide-protein complex which did not require MSA information for the peptide partner.”
Ins and outs of AlphaFold2 transmembrane protein structure predictions (2022, background): “partially caused by the footprint of these proteins themselves in the AF2 neural network, successful protein-peptide docking [50], when peptides were not involved in alignments, is an argument against this reasoning.”
Mass Spectrometry and Machine Learning Reveal Determinants of Client Recognition by Antiamyloid Chaperones (2022, background): “Thus, AF2 has also been found to reliably dock short peptides into binding pockets of folded protein domains (25).”
Using AlphaFold to predict the impact of single mutations on protein stability and function (2021, background): “Researchers in the field are already actively testing AlphaFold performance in various bioinformatics tasks, for instance, in peptide-protein docking [13, 14].”
Protein complex prediction with AlphaFold-Multimer (2021, background): “residue gap insertion or chains joined with a flexible linker) to the singlechain AlphaFold model is often successful at predicting multimer interactions [4, 5, 6, 7, 8, 9].”
Independent SE(3)-Equivariant Models for End-to-End Rigid Protein Docking (2021, unknown): “Peptide-protein docking is a subproblem of protein-protein docking, its performance has been refined recently using protein folding networks (Tsaban et al., 2021; Ko and Lee, 2021).”
The HADDOCK2.4 web server for integrative modeling of biomolecular complexes (2024, unknown): “Although AlphFold2 was originally trained only on single structures, its potential for predicting complexes was instantly recognized, with applications to protein–peptide complex prediction 24 and protein–protein complexes, sometimes in combination with classical docking methods 25 .”
Improving peptide-protein docking with AlphaFold-Multimer using forced sampling (2021, background): “, 2022) or using a flexible linker (Ko and Lee, 2021; Tsaban et al., 2022).”