Improved protein structure prediction using predicted interresidue orientations

By Jianyi Yang, Ivan V. Anishchenko, Hahnbeom Park, Zhenling Peng, Sergey Ovchinnikov, David Baker. 2019

1,274 citations recorded by the source corpus.

Significance Protein structure prediction is a longstanding challenge in computational biology. Through extension of deep learning-based prediction to interresidue orientations in addition to distances, and the development of a constrained optimization by Rosetta, we show that more accurate models can be generated. Results on a set of 18 de novo-designed proteins suggests the proposed method should be directly applicable to current challenges in de novo protein design. The prediction of interresidue contacts and distances from coevolutionary data using deep learning has considerably advanced protein structure prediction. Here, we build on these advances by developing a deep residual network for predicting interresidue orientations, in addition to distances, and a Rosetta-constrained energy-minimization protocol for rapidly and accurately generating structure models guided by these restraints. In benchmark tests on 13th Community-Wide Experiment on the Critical Assessment of Techniques for Protein Structure Prediction (CASP13)- and Continuous Automated Model Evaluation (CAMEO)-derived sets, the method outperforms all previously described structure-prediction methods. Although trained entirely on native proteins, the network consistently assigns higher probability to de novo-designed proteins, identifying the key fold-determining residues and providing an independent quantitative measure of the “ideality” of a protein structure. The method promises to be useful for a broad range of protein structure prediction and design problems.

Subfields: Computational Theory and Mathematics, 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 Computational Theory and Mathematics

In Materials Chemistry

In Molecular Biology

In Computational Biology & Bioinformatics

Citations in the corpus

Accurate prediction of protein structures and interactions using a 3-track neural network (2021, positive): “described at the CASP14 conference, the AlphaFold2 methodological advances included 1) starting from multiple sequence alignments (MSAs) rather than from more processed features such as inverse covariance matrices derived from MSAs, 2) replacement of 2D convolution with an attention mechanism that b”
End-to-end protein–ligand complex structure generation with diffusion-based generative models (2022, methodology): “The diffusion model used in this study is a likelihood-based generative model that is capable of estimating the likelihood of each sample [15, 29].”
Evolutionary-scale prediction of atomic level protein structure with a language model (2022, background): “[11] Jianyi Yang, Ivan Anishchenko, Hahnbeom Park, Zhenling Peng, Sergey Ovchinnikov, and David Baker.”
MSA Transformer (2021, methodology): “We also compare to trRosetta (Yang et al., 2019), a state-of-the-art supervised (Jumper et al., 2020).”
De novo protein design by deep network hallucination (2020, background): “Sequence optimization Given Eqs.”
Improved protein structure refinement guided by deep learning based accuracy estimation (2020, positive): “Distance prediction through deep learning on amino acid co-evolution data has considerably advanced protein structure prediction (1–3) .”
Learning inverse folding from millions of predicted structures (2022, methodology): “(2021) carried out Monte Carlo sampling in the sequence space to invert the trRosetta (Yang et al., 2020) structure prediction network for sequence design.”
Protein sequence design with a learned potential (2022, methodology): “We also use the trRosetta online server to get structure predictions for designed sequences 22 . trRosetta uses a deep neural network to predict inter-residue distance and orientation distributions from sequence and multiple sequence alignment (MSA) data.”
ProtTrans: Towards Cracking the Language of Life’s Code Through Self-Supervised Deep Learning and High Performance Computing (2020, methodology): “Top prediction methods in computational biology [15], [16], [17], [18], [19], [20] combine machine learning (ML) and evolutionary information (EI), ﬁrst established as the winning strategy to predict protein secondary structure [21], [22] in two steps.”
High‐accuracy protein structure prediction in CASP14 (2021, methodology): “Nevertheless, the second-best method, an updated version of the deep learning-based trRosetta 12 from the Baker group, reached a level of accuracy superior to that of AF in CASP13, illustrating the role of community-based benchmarking experiments in driving the further development of the field.”
Protein sequence design by conformational landscape optimization (2021, methodology): “Combining these predictions with Rosetta energy minimization yielded excellent predictions of structures in benchmark cases, as well as recent blind Continous Automated Model EvalutiOn (CAMEO) structure modeling evaluations (13).”
Scaffolding protein functional sites using deep learning (2022, methodology): “We previously demonstrated that the trRosetta structure-prediction neural network (11) can be used to generate new proteins by maximizing the trRosetta output probability that a sequence folds to some (unspecified) three dimensional structure during Monte Carlo sampling in sequence space (12).”
Transformer protein language models are unsupervised structure learners (2020, methodology): “Inputs are typically covariance statistics (Jones & Kandathil, 2018; Adhikari & Elofsson, 2020), or inferred coevolutionary parameters (Wang et al., 2017; Liu et al., 2018; Senior et al., 2020; Yang et al., 2019).”