Accurate Protein Structure Prediction Research Articles

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

DeepMind presented remarkably accurate predictions at the recent CASP14 protein structure prediction assessment conference. We explored network architectures incorporating related ideas and obtained the best performance with a 3-track network in which information at the 1D sequence level, the 2D distance map level, and the 3D coordinate level is successively transformed and integrated. The 3-track network produces structure predictions with accuracies approaching those of DeepMind in CASP14, enables the rapid solution of challenging X-ray crystallography and cryo-EM 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 which require modeling of individual subunits followed by docking. We make the method available to the scientific community to speed biological research.

Accurate protein structure prediction AI made openly available

Accurate protein structure prediction AI made openly available

Highly accurate protein structure prediction with AlphaFold

Proteins are essential to life, and understanding their structure can facilitate a mechanistic understanding of their function. Through an enormous experimental effort1–4, the structures of around 100,000 unique proteins have been determined5, but this represents a small fraction of the billions of known protein sequences6,7. Structural coverage is bottlenecked by the months to years of painstaking effort required to determine a single protein structure. Accurate computational approaches are needed to address this gap and to enable large-scale structural bioinformatics. Predicting the three-dimensional structure that a protein will adopt based solely on its amino acid sequence—the structure prediction component of the ‘protein folding problem’8—has been an important open research problem for more than 50 years9. Despite recent progress10–14, existing methods fall far short of atomic accuracy, especially when no homologous structure is available. Here we provide the first computational method that can regularly predict protein structures with atomic accuracy even in cases in which no similar structure is known. We validated an entirely redesigned version of our neural network-based model, AlphaFold, in the challenging 14th Critical Assessment of protein Structure Prediction (CASP14)15, demonstrating accuracy competitive with experimental structures in a majority of cases and greatly outperforming other methods. Underpinning the latest version of AlphaFold is a novel machine learning approach that incorporates physical and biological knowledge about protein structure, leveraging multi-sequence alignments, into the design of the deep learning algorithm.

PyconsFold: a fast and easy tool for modeling and docking using distance predictions.

MotivationContact predictions within a protein have recently become a viable method for accurate prediction of protein structure. Using predicted distance distributions has been shown in many cases to be superior to only using a binary contact annotation. Using predicted interprotein distances has also been shown to be able to dock some protein dimers.ResultsHere, we present pyconsFold. Using CNS as its underlying folding mechanism and predicted contact distance it outperforms regular contact prediction-based modeling on our dataset of 210 proteins. It performs marginally worse than the state-of-the-art pyRosetta folding pipeline but is on average about 20 times faster per model. More importantly pyconsFold can also be used as a fold-and-dock protocol by using predicted interprotein contacts/distances to simultaneously fold and dock two protein chains.Availability and implementationpyconsFold is implemented in Python 3 with a strong focus on using as few dependencies as possible for longevity. It is available both as a pip package in Python 3 and as source code on GitHub and is published under the GPLv3 license. The data underlying this article together with source code are available on github, at https://github.com/johnlamb/pyconsfold.Supplementary information Supplementary data are available at Bioinformatics online.

SPOT-1D-Single: improving the single-sequence-based prediction of protein secondary structure, backbone angles, solvent accessibility and half-sphere exposures using a large training set and ensembled deep learning.

Knowing protein secondary and other one-dimensional structural properties are essential for accurate protein structure and function prediction. As a result, many methods have been developed for predicting these one-dimensional structural properties. However, most methods relied on evolutionary information that may not exist for many proteins due to a lack of sequence homologs. Moreover, it is computationally intensive for obtaining evolutionary information as the library of protein sequences continues to expand exponentially. Here, we developed a new single-sequence method called SPOT-1D-Single based on a large training dataset of 39 120 proteins deposited prior to 2016 and an ensemble of hybrid long-short-term-memory bidirectional neural network and convolutional neural network. We showed that SPOT-1D-Single consistently improves over SPIDER3-Single and ProteinUnet for secondary structure, solvent accessibility, contact number and backbone angles prediction for all seven independent test sets (TEST2018, SPOT-2016, SPOT-2016-HQ, SPOT-2018, SPOT-2018-HQ, CASP12 and CASP13 free-modeling targets). For example, the predicted three-state secondary structure's accuracy ranges from 72.12% to 74.28% by SPOT-1D-Single, compared to 69.1-72.6% by SPIDER3-Single and 70.6-73% by ProteinUnet. SPOT-1D-Single also predicts SS3 and SS8 with 6.24% and 6.98% better accuracy than SPOT-1D on SPOT-2018 proteins with no homologs (Neff = 1), respectively. The new method's improvement over existing techniques is due to a larger training set combined with ensembled learning. Standalone-version of SPOT-1D-Single is available at https://github.com/jas-preet/SPOT-1D-Single. Direct prediction can also be made at https://sparks-lab.org/server/spot-1d-single. The datasets used in this research can also be downloaded from GitHub. Supplementary data are available at Bioinformatics online.

Progress in Determination of Protein Spatial Structure Based on Machine Learning

Introduction. The task of determining the spatial structure of proteins is one of the most important unsolved problems of mankind. Life on the planet Earth is called protein, because protein molecules are the drivers of life processes in living organisms. Proteins make up about 80% of the dry mass of the cell and coordinate the processes of metabolism. The functions of proteins are defined by its spatial structure. The results of recent competitions in methods for determining protein structures have shown significant progress in this important area. One of the research groups presented the AlphaFold 2 method, the accuracy of which reached the accuracy of experimental methods. Purpose of the article. The aim of the work is to consider and analyze the basic principles of the AlphaFold software package for determining the spatial structure of proteins. Results. We consider the main stages in the process of recognizing the structure of a protein using the AlphaFold program complex. The stages and corresponding methods include: search for homologous proteins based on multiple alignment methods, construction of protein-specific differentiated potential using artificial neural networks and protein structure energy optimization using gradient descent and limited sampling. We discuss how combination of various bioinformatics techniques powered by data from open data sources can lead to significant improvements in accuracy of protein structure prediction. Special attention is paid to the use of artificial neural networks for building the smooth protein-specific potential and following energy minimization based on constructed potential. Conclusions. The combination of a number of methods and the use of information from protein and genetic data banks allows us to make significant progress in solving the extremely important task of determining the structure of a protein. Keywords: protein spatial structure, Machine Learning, AlphaFold.

Accurate protein structure prediction with hydroxyl radical protein footprinting data

Hydroxyl radical protein footprinting (HRPF) in combination with mass spectrometry reveals the relative solvent exposure of labeled residues within a protein, thereby providing insight into protein tertiary structure. HRPF labels nineteen residues with varying degrees of reliability and reactivity. Here, we are presenting a dynamics-driven HRPF-guided algorithm for protein structure prediction. In a benchmark test of our algorithm, usage of the dynamics data in a score term resulted in notable improvement of the root-mean-square deviations of the lowest-scoring ab initio models and improved the funnel-like metric Pnear for all benchmark proteins. We identified models with accurate atomic detail for three of the four benchmark proteins. This work suggests that HRPF data along with side chain dynamics sampled by a Rosetta mover ensemble can be used to accurately predict protein structure.

DISTEVAL: a web server for evaluating predicted protein distances

BackgroundProtein inter-residue contact and distance prediction are two key intermediate steps essential to accurate protein structure prediction. Distance prediction comes in two forms: real-valued distances and ‘binned’ distograms, which are a more finely grained variant of the binary contact prediction problem. The latter has been introduced as a new challenge in the 14th Critical Assessment of Techniques for Protein Structure Prediction (CASP14) 2020 experiment. Despite the recent proliferation of methods for predicting distances, few methods exist for evaluating these predictions. Currently only numerical metrics, which evaluate the entire prediction at once, are used. These give no insight into the structural details of a prediction. For this reason, new methods and tools are needed.ResultsWe have developed a web server for evaluating predicted inter-residue distances. Our server, DISTEVAL, accepts predicted contacts, distances, and a true structure as optional inputs to generate informative heatmaps, chord diagrams, and 3D models. All of these outputs facilitate visual and qualitative assessment. The server also evaluates predictions using other metrics such as mean absolute error, root mean squared error, and contact precision.ConclusionsThe visualizations generated by DISTEVAL complement each other and collectively serve as a powerful tool for both quantitative and qualitative assessments of predicted contacts and distances, even in the absence of a true 3D structure.

Sequence Alignment Using Machine Learning for Accurate Template-based Protein Structure Prediction.

Template-based modeling, the process of predicting the tertiary structure of a protein by using homologous protein structures, is useful when good templates can be available. Indeed, modern homology detection methods can find remote homologs with high sensitivity. However, the accuracy of template-based models generated from the homology-detection-based alignments is often lower than that from ideal alignments. In this study, we propose a new method that generates pairwise sequence alignments for more accurate template-based modeling. Our method trains a machine learning model using the structural alignment of known homologs. When calculating sequence alignments, instead of a fixed substitution matrix, this method dynamically predicts a substitution score from the trained model.

CGLFold: a contact-assisted de novo protein structure prediction using global exploration and loop perturbation sampling algorithm.

Regions that connect secondary structure elements in a protein are known as loops, whose slight change will produce dramatic effect on the entire topology. This study investigates whether the accuracy of protein structure prediction can be improved using a loop-specific sampling strategy. A novel de novo protein structure prediction method that combines global exploration and loop perturbation is proposed in this study. In the global exploration phase, the fragment recombination and assembly are used to explore the massive conformational space and generate native-like topology. In the loop perturbation phase, a loop-specific local perturbation model is designed to improve the accuracy of the conformation and is solved by differential evolution algorithm. These two phases enable a cooperation between global exploration and local exploitation. The filtered contact information is used to construct the conformation selection model for guiding the sampling. The proposed CGLFold is tested on 145 benchmark proteins, 14 free modeling (FM) targets of CASP13 and 29 FM targets of CASP12. The experimental results show that the loop-specific local perturbation can increase the structure diversity and success rate of conformational update and gradually improve conformation accuracy. CGLFold obtains template modeling score ≥ 0.5 models on 95 standard test proteins, 7 FM targets of CASP13 and 9 FM targets of CASP12. The source code and executable versions are freely available at https://github.com/iobio-zjut/CGLFold. Supplementary data are available at Bioinformatics online.

Protein structure prediction assisted with sparse NMR data in CASP13.

CASP13 has investigated the impact of sparse NMR data on the accuracy of protein structure prediction. NOESY and 15 N-1 H residual dipolar coupling data, typical of that obtained for 15 N,13 C-enriched, perdeuterated proteins up to about 40 kDa, were simulated for 11 CASP13 targets ranging in size from 80 to 326 residues. For several targets, two prediction groups generated models that are more accurate than those produced using baseline methods. Real NMR data collected for a de novo designed protein were also provided to predictors, including one data set in which only backbone resonance assignments were available. Some NMR-assisted prediction groups also did very well with these data. CASP13 also assessed whether incorporation of sparse NMR data improves the accuracy of protein structure prediction relative to nonassisted regular methods. In most cases, incorporation of sparse, noisy NMR data results in models with higher accuracy. The best NMR-assisted models were also compared with the best regular predictions of any CASP13 group for the same target. For six of 13 targets, the most accurate model provided by any NMR-assisted prediction group was more accurate than the most accurate model provided by any regular prediction group; however, for the remaining seven targets, one or more regular prediction method provided a more accurate model than even the best NMR-assisted model. These results suggest a novel approach for protein structure determination, in which advanced prediction methods are first used to generate structural models, and sparse NMR data is then used to validate and/or refine these models.

Detecting distant-homology protein structures by aligning deep neural-network based contact maps.

Accurate prediction of atomic-level protein structure is important for annotating the biological functions of protein molecules and for designing new compounds to regulate the functions. Template-based modeling (TBM), which aims to construct structural models by copying and refining the structural frameworks of other known proteins, remains the most accurate method for protein structure prediction. Due to the difficulty in recognizing distant-homology templates, however, the accuracy of TBM decreases rapidly when the evolutionary relationship between the query and template vanishes. In this study, we propose a new method, CEthreader, which first predicts residue-residue contacts by coupling evolutionary precision matrices with deep residual convolutional neural-networks. The predicted contact maps are then integrated with sequence profile alignments to recognize structural templates from the PDB. The method was tested on two independent benchmark sets consisting collectively of 1,153 non-homologous protein targets, where CEthreader detected 176% or 36% more correct templates with a TM-score >0.5 than the best state-of-the-art profile- or contact-based threading methods, respectively, for the Hard targets that lacked homologous templates. Moreover, CEthreader was able to identify 114% or 20% more correct templates with the same Fold as the query, after excluding structures from the same SCOPe Superfamily, than the best profile- or contact-based threading methods. Detailed analyses show that the major advantage of CEthreader lies in the efficient coupling of contact maps with profile alignments, which helps recognize global fold of protein structures when the homologous relationship between the query and template is weak. These results demonstrate an efficient new strategy to combine ab initio contact map prediction with profile alignments to significantly improve the accuracy of template-based structure prediction, especially for distant-homology proteins.

Sequence alignment using machine learning for accurate template-based protein structure prediction.

Template-based modeling, the process of predicting the tertiary structure of a protein by using homologous protein structures, is useful if good templates can be found. Although modern homology detection methods can find remote homologs with high sensitivity, the accuracy of template-based models generated from homology-detection-based alignments is often lower than that from ideal alignments. In this study, we propose a new method that generates pairwise sequence alignments for more accurate template-based modeling. The proposed method trains a machine learning model using the structural alignment of known homologs. It is difficult to directly predict sequence alignments using machine learning. Thus, when calculating sequence alignments, instead of a fixed substitution matrix, this method dynamically predicts a substitution score from the trained model. We evaluate our method by carefully splitting the training and test datasets and comparing the predicted structure's accuracy with that of state-of-the-art methods. Our method generates more accurate tertiary structure models than those produced from alignments obtained by other methods. https://github.com/shuichiro-makigaki/exmachina. Supplementary data are available at Bioinformatics online.

Four Challenges for Better Biocatalysts

Biocatalysis (the use of biological molecules or materials to catalyse chemical reactions) has considerable potential. The use of biological molecules as catalysts enables new and more specific syntheses. It also meets many of the core principles of “green chemistry”. While there have been some considerable successes in biocatalysis, the full potential has yet to be realised. This results, partly, from some key challenges in understanding the fundamental biochemistry of enzymes. This review summarises four of these challenges: the need to understand protein folding, the need for a qualitative understanding of the hydrophobic effect, the need to understand and quantify the effects of organic solvents on biomolecules and the need for a deep understanding of enzymatic catalysis. If these challenges were addressed, then the number of successful biocatalysis projects is likely to increase. It would enable accurate prediction of protein structures, and the effects of changes in sequence or solution conditions on these structures. We would be better able to predict how substrates bind and are transformed into products, again leading to better enzyme engineering. Most significantly, it may enable the de novo design of enzymes to catalyse specific reactions.

Does inclusion of residue-residue contact information boost protein threading?

Template-based modeling is considered as one of the most successful approaches for protein structure prediction. However, reliably and accurately selecting optimal template proteins from a library of known protein structures having similar folds as the target protein and making correct alignments between the target sequence and the template structures, a template-based modeling technique known as threading, remains challenging, particularly for non- or distantly-homologous protein targets. With the recent advancement in protein residue-residue contact map prediction powered by sequence co-evolution and machine learning, here we systematically analyze the effect of inclusion of residue-residue contact information in improving the accuracy and reliability of protein threading. We develop a new threading algorithm by incorporating various sequential and structural features, and subsequently integrate residue-residue contact information as an additional scoring term for threading template selection. We show that the inclusion of contact information attains statistically significantly better threading performance compared to a baseline threading algorithm that does not utilize contact information when everything else remains the same. Experimental results demonstrate that our contact based threading approach outperforms popular threading method MUSTER, contact-assisted ab initio folding method CONFOLD2, and recent state-of-the-art contact-assisted protein threading methods EigenTHREADER and map_align on several benchmarks. Our study illustrates that the inclusion of contact maps is a promising avenue in protein threading to ultimately help to improve the accuracy of protein structure prediction.

DeepCDpred: Inter-residue distance and contact prediction for improved prediction of protein structure.

Rapid, accurate prediction of protein structure from amino acid sequence would accelerate fields as diverse as drug discovery, synthetic biology and disease diagnosis. Massively improved prediction of protein structures has been driven by improving the prediction of the amino acid residues that contact in their 3D structure. For an average globular protein, around 92% of all residue pairs are non-contacting, therefore accurate prediction of only a small percentage of inter-amino acid distances could increase the number of constraints to guide structure determination. We have trained deep neural networks to predict inter-residue contacts and distances. Distances are predicted with an accuracy better than most contact prediction techniques. Addition of distance constraints improved de novo structure predictions for test sets of 158 protein structures, as compared to using the best contact prediction methods alone. Importantly, usage of distance predictions allows the selection of better models from the structure pool without a need for an external model assessment tool. The results also indicate how the accuracy of distance prediction methods might be improved further.

Template-Guided Protein Structure Prediction and Refinement Using Optimized Folding Landscape Force Fields.

When good structural templates can be identified, template-based modeling is the most reliable way to predict the tertiary structure of proteins. In this study, we combine template-based modeling with a realistic coarse-grained force field, AWSEM, that has been optimized using the principles of energy landscape theory. The Associative memory, Water mediated, Structure and Energy Model (AWSEM) is a coarse-grained force field having both transferable tertiary interactions and knowledge-based local-in-sequence interaction terms. We incorporate template information into AWSEM by introducing soft collective biases to the template structures, resulting in a model that we call AWSEM-Template. Structure prediction tests on eight targets, four of which are in the low sequence identity "twilight zone" of homology modeling, show that AWSEM-Template can achieve high-resolution structure prediction. Our results also confirm that using a combination of AWSEM and a template-guided potential leads to more accurate prediction of protein structures than simply using a template-guided potential alone. Free energy profile analyses demonstrate that the soft collective biases to the template effectively increase funneling toward native-like structures while still allowing significant flexibility so as to allow for correction of discrepancies between the target structure and the template. A further stage of refinement using all-atom molecular dynamics augmented with soft collective biases to the structures predicted by AWSEM-Template leads to a further improvement of both backbone and side-chain accuracy by maintaining sufficient flexibility but at the same time discouraging unproductive unfolding events often seen in unrestrained all-atom refinement simulations. The all-atom refinement simulations also reduce patches of frustration of the initial predictions. Some of the backbones found among the structures produced during the initial coarse-grained prediction step already have CE-RMSD values of less than 3 Å with 90% or more of the residues aligned to the experimentally solved structure for all targets. All-atom structures generated during the following all-atom refinement simulations, which started from coarse-grained structures that were chosen without reference to any knowledge about the native structure, have CE-RMSD values of less than 2.5 Å with 90% or more of the residues aligned for 6 out of 8 targets. Clustering low energy structures generated during the initial coarse-grained annealing picks out reliably structures that are within 1 Å of the best sampled structures in 5 out of 8 cases. After the all-atom refinement, structures that are within 1 Å of the best sampled structures can be selected using a simple algorithm based on energetic features alone in 7 out of 8 cases.

Predicting improved protein conformations with a temporal deep recurrent neural network.

Accurate protein structure prediction from amino acid sequence is still an unsolved problem. The most reliable methods centre on template based modelling. However, the accuracy of these models entirely depends on the availability of experimentally resolved homologous template structures. In order to generate more accurate models, extensive physics based molecular dynamics (MD) refinement simulations are performed to sample many different conformations to find improved conformational states. In this study, we propose a deep recurrent network model, called DeepTrajectory, that is able to identify these improved conformational states, with high precision, from a variety of different MD based sampling protocols. The proposed model learns the temporal patterns of features computed from MD trajectory data in order to classify whether each recorded simulation snapshot is an improved quality conformational state, decreased quality conformational state or whether there is no perceivable change in state with respect to the starting conformation. The model was trained and tested on 904 trajectories from 42 different protein systems with a cumulative number of more than 1.7 million snapshots. We show that our model outperforms other state of the art machine-learning algorithms that do not consider temporal dependencies. To our knowledge, DeepTrajectory is the first implementation of a time-dependent deep-learning protocol that is re-trainable and able to adapt to any new MD based sampling procedure, thereby demonstrating how a neural network can be used to learn the latter part of the protein folding funnel.

Fast Implicit Potentials for Accurate Prediction and Design of Membrane Protein Structures

Fast Implicit Potentials for Accurate Prediction and Design of Membrane Protein Structures

Exploring the effects of sparse restraints on protein structure prediction.

One of the main barriers to accurate computational protein structure prediction is searching the vast space of protein conformations. Distance restraints or inter-residue contacts have been used to reduce this search space, easing the discovery of the correct folded state. It has been suggested that about 1 contact for every 12 residues may be sufficient to predict structure at fold level accuracy. Here, we use coarse-grained structure-based models in conjunction with molecular dynamics simulations to examine this empirical prediction. We generate sparse contact maps for 15 proteins of varying sequence lengths and topologies and find that given perfect secondary-structural information, a small fraction of the native contact map (5%-10%) suffices to fold proteins to their correct native states. We also find that different sparse maps are not equivalent and we make several observations about the type of maps that are successful at such structure prediction. Long range contacts are found to encode more information than shorter range ones, especially for α and αβ-proteins. However, this distinction reduces for β-proteins. Choosing contacts that are a consensus from successful maps gives predictive sparse maps as does choosing contacts that are well spread out over the protein structure. Additionally, the folding of proteins can also be used to choose predictive sparse maps. Overall, we conclude that structure-based models can be used to understand the efficacy of structure-prediction restraints and could, in future, be tuned to include specific force-field interactions, secondary structure errors and noise in the sparse maps.