Improved prediction of protein-protein interactions using AlphaFold2

By Patrick Bryant, Gabriele Pozzati, Arne Elofsson. 2021

933 citations recorded by the source corpus.

Predicting the structure of interacting protein chains is a fundamental step towards understanding protein function. Unfortunately, no computational method can produce accurate structures of protein complexes. AlphaFold2, has shown unprecedented levels of accuracy in modelling single chain protein structures. Here, we apply AlphaFold2 for the prediction of heterodimeric protein complexes. We find that the AlphaFold2 protocol together with optimised multiple sequence alignments, generate models with acceptable quality (DockQ ≥ 0.23) for 63% of the dimers. From the predicted interfaces we create a simple function to predict the DockQ score which distinguishes acceptable from incorrect models as well as interacting from non-interacting proteins with state-of-art accuracy. We find that, using the predicted DockQ scores, we can identify 51% of all interacting pairs at 1% FPR. Predicting the structure of protein complexes is extremely difficult. Here, authors apply AlphaFold2 with optimized multiple sequence alignments to model complexes of interacting proteins, enabling prediction of both if and how proteins interact with state-of-art accuracy.

Subfields: Materials Chemistry, Molecular Biology, Computational Biology & Bioinformatics

How researchers see this paper

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

In Materials Chemistry

In Molecular Biology

In Computational Biology & Bioinformatics

Citations in the corpus

AFsample: improving multimer prediction with AlphaFold using massive sampling (2023, background): “However, for multimers, this is not necessarily true (Bryant et al. 2022), the evolutionary signal constraining multimers is much weaker, and for these cases, more sampling might improve the prediction.”
In silico protein interaction screening uncovers DONSON’s role in replication initiation (2023, methodology): “of human CMG (PDB: 7PLO) (64) by aligning”
AlphaFold2 and its applications in the fields of biology and medicine (2023, positive): “Recently, Bryant et al. 142 applied AF2 to predict heterodimeric protein complexes.”
ColabFold: making protein folding accessible to all (2022, methodology): “22 ).”
Benchmarking AlphaFold for protein complex modeling reveals accuracy determinants (2021, methodology): “While this manuscript was under review, a separate study 57 was published that also reported benchmarking of AlphaFold with a set of protein – protein complexes.”
A structural biology community assessment of AlphaFold2 applications (2021, methodology): “To evaluate performance, one model for each pair was generated with AF2 (using default parameters, except that model_2 was used, providing a complementary set of results to those derived in (Bryant et al., 2021)).”
AI-based structure prediction empowers integrative structural analysis of human nuclear pores (2022, background): “However, AlphaFold and RoseTTAfold have recently demonstrated unprecedented accuracy in predicting structures ofmonomeric proteins (26, 27, 60–65) and complexes (34, 36, 61, 66).”
OpenFold: retraining AlphaFold2 yields new insights into its learning mechanisms and capacity for generalization (2023, neutral): “Predicting protein structure from sequence has been a defining challenge of biology for decades 1,2 .”
Complementing machine learning‐based structure predictions with native mass spectrometry (2022, methodology): “For example, using protein sequence data as the only user input, AF2(3) can generate models of ordered, monomeric proteins that rival in quality experimentally derived structures,(4) which can be assembled into complexes using AF2 Multimer.(5) Received: 17 March 2022 Revised: 7 April 2022 Accepted: ”
Improved multimer prediction using massive sampling with AlphaFold in CASP15 (2023, background): “However, this is not necessarily the case for multimers, where the evolutionary signal42 constraining the prediction is much weaker (Bryant et al., 2022), and thus, more sampling may be43 necessary to improve the prediction.”
Towards a structurally resolved human protein interaction network (2021, methodology): “Recently, neural network-based approaches have demonstrated the ability to accurately predict the structures of individual proteins 16,17 and protein complexes 16,18–21 .”