Accurate prediction of protein structures and interactions using a 3-track neural network

By B. M, D. f, A. I, D. J, O. S, Lee Gr, W. J, C. Q., Kinch Ln, Schaeffer Rd and 22 more. Science, 2021

3,390 citations recorded by the source corpus.

Deep learning takes on protein folding In 1972, Anfinsen won a Nobel prize for demonstrating a connection between a protein's amino acid sequence and its three-dimensional structure. Since 1994, scientists have competed in the biannual Critical Assessment of Structure Prediction (CASP) protein-folding challenge. Deep learning methods took center stage at CASP14, with DeepMind's Alphafold2 achieving remarkable accuracy. Baek et al. explored network architectures based on the DeepMind framework. They used a three-track network to process sequence, distance, and coordinate information simultaneously and achieved accuracies approaching those of DeepMind. The method, RoseTTA fold, can solve challenging x-ray crystallography and cryo–electron microscopy modeling problems and generate accurate models of protein-protein complexes. Science, abj8754, this issue p. 871 Protein structure modeling enables the rapid solution of protein structures and provides insights into function. DeepMind presented notably accurate predictions at the recent 14th Critical Assessment of Structure Prediction (CASP14) conference. We explored network architectures that incorporate related ideas and obtained the best performance with a three-track network in which information at the one-dimensional (1D) sequence level, the 2D distance map level, and the 3D coordinate level is successively transformed and integrated. The three-track network produces structure predictions with accuracies approaching those of DeepMind in CASP14, enables the rapid solution of challenging x-ray crystallography and cryo–electron microscopy structure modeling problems, and provides insights into the functions of proteins of currently unknown structure. The network also enables rapid generation of accurate protein-protein complex models from sequence information alone, short-circuiting traditional approaches that require modeling of individual subunits followed by docking. We make the method available to the scientific community to speed biological research.

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

View Paper: https://doi.org/10.1126/science.abj8754

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

Robust deep learning based protein sequence design using ProteinMPNN (2022, positive): “…(MPNN) with 3 encoder and 3 decoder layers and 128 hidden dimensions which predicts protein sequences in an autoregressive manner from N to C terminus using protein backbone features – distances between Ca-Ca atoms, relative Ca-Ca-Ca frame orientations and rotations, and backbone dihedral…”
Multi-State Modeling of G-protein Coupled Receptors at Experimental Accuracy (2021, positive): “demonstrated that RoseTTAFold can model GPCRs in the active and inactive states by providing activation-state specific templates.(6) The”
De novo protein design by deep network hallucination (2020, methodology): “Our hallucination approach can be readily extended to design new proteins using the recently developed RoseTTAFold 26 and AlphaFold2 27 networks 28,29 .”
Hallucinating symmetric protein assemblies (2022, methodology): “In addition, we independently evaluated the sequences using an updated version of RoseTTAFold (RF2) (33), and found that RF2 did not confidently predict the structure of most of the original AF2 hallucinated sequences, but successfully predicted almost all ProteinMPNN sequences (Fig.”
Critical Assessment of Methods of Protein Structure Prediction (CASP) – Round XIV (2021, methodology): “the modeling community now, exploring the new techniques, and the Baker group has already reported modeling accuracy similar to that of AlphaFold2 using deep-learning methods.(28)”
Scaffolding protein functional sites using deep learning (2022, methodology): “Following the development of RosettaFold (RF) (16) we found that it performed better than trRosetta in guiding protein design by functional-site-constrained hallucination (Fig.”
Geometric deep learning on molecular representations (2021, background): “Synthesis planning [56], prediction of reaction yields [57], generative molecular design [58], prediction of properties in drug discovery [59], and protein structure prediction [6, 7].”
Fast, accurate antibody structure prediction from deep learning on massive set of natural antibodies (2022, methodology): “Deep learning methods have brought about a revolution in protein structure prediction 10,11 .”
The trRosetta server for fast and accurate protein structure prediction (2021, methodology): “On the basis of the success of AlphaFold2, we believe that more deep learning – based methods will be developed in the future (e.g., the recent RoseTTAFold 12 ).”