Deep learning for reconstructing protein structures from cryo-EM density maps: Recent advances and future directions

Cryo-Electron Microscopy (cryo-EM) has emerged as a key technology to determine the structure of proteins, particularly large protein complexes and assemblies in recent years. A key challenge in cryo-EM data analysis is to automatically reconstruct accurate protein structures from cryo-EM density maps. In this review, we briefly overview various deep learning methods for building protein structures from cryo-EM density maps, analyze their impact, and discuss the challenges of preparing high-quality data sets for training deep learning models. Looking into the future, more advanced deep learning models of effectively integrating cryo-EM data with other sources of complementary data such as protein sequences and AlphaFold-predicted structures need to be developed to further advance the field.     

Cryo-electron microscopy modeling by the molecular dynamics flexible fitting method

The increasing power and popularity of cryo-electron microscopy (cryo-EM) in structural biology brought about the development of so-called hybrid methods, which permit the interpretation of cryo-EM density maps beyond their nominal resolution in terms of atomic models. The Cryo-EM Modeling Challenge 2010 is the first community effort to bring together developers of hybrid methods as well as cryo-EM experimentalists. Participating in the challenge, the molecular dynamics flexible fitting (MDFF) method was applied to a number of cryo-EM density maps. The results are described here with special emphasis on the use of symmetry-based restraints to improve the quality of atomic models derived from density maps of symmetric complexes; on a comparison of the stereochemical quality of atomic models resulting from different hybrid methods; and on application of MDFF to electron crystallography data.     

Small angle X-ray scattering-assisted protein structure prediction in CASP13 and emergence of solution structure differences

Small angle X-ray scattering (SAXS) measures comprehensive distance information on a protein's structure, which can constrain and guide computational structure prediction algorithms. Here, we evaluate structure predictions of 11 monomeric and oligomeric proteins for which SAXS data were collected and provided to predictors in the 13th round of the Critical Assessment of protein Structure Prediction (CASP13). The category for SAXS-assisted predictions made gains in certain areas for CASP13 compared to CASP12. Improvements included higher quality data with size exclusion chromatography-SAXS (SEC-SAXS) and better selection of targets and communication of results by CASP organizers. In several cases, we can track improvements in model accuracy with use of SAXS data. For hard multimeric targets where regular folding algorithms were unsuccessful, SAXS data helped predictors to build models better resembling the global shape of the target. For most models, however, no significant improvement in model accuracy at the domain level was registered from use of SAXS data, when rigorously comparing SAXS-assisted models to the best regular server predictions. To promote future progress in this category, we identify successes, challenges, and opportunities for improved strategies in prediction, assessment, and communication of SAXS data to predictors. An important observation is that, for many targets, SAXS data were inconsistent with crystal structures, suggesting that these proteins adopt different conformation(s) in solution. This CASP13 result, if representative of PDB structures and future CASP targets, may have substantive implications for the structure training databases used for machine learning, CASP, and use of prediction models for biology.     

Comparison of Segger and other methods for segmentation and rigid-body docking of molecular components in cryo-EM density maps

Segmentation and docking are useful methods for the discovery of molecular components in electron cryo-microscopy (cryo-EM) density maps of macromolecular complexes. In this article, we describe the segmentation and docking methods implemented in Segger. For 11 targets posted in the 2010 cryo-EM challenge, we segmented the regions corresponding to individual molecular components using Segger. We then used the segmented regions to guide rigid-body docking of individual components. Docking results were evaluated by comparing the docked components with published structures, and by calculation of several scores, such as atom inclusion, density occupancy, and geometry clash. The accuracy of the component segmentation using Segger and other methods was assessed by comparing segmented regions with docked components.     

Estimation of model accuracy in CASP13

Methods to reliably estimate the accuracy of 3D models of proteins are both a fundamental part of most protein folding pipelines and important for reliable identification of the best models when multiple pipelines are used. Here, we describe the progress made from CASP12 to CASP13 in the field of estimation of model accuracy (EMA) as seen from the progress of the most successful methods in CASP13. We show small but clear progress, that is, several methods perform better than the best methods from CASP12 when tested on CASP13 EMA targets. Some progress is driven by applying deep learning and residue-residue contacts to model accuracy prediction. We show that the best EMA methods select better models than the best servers in CASP13, but that there exists a great potential to improve this further. Also, according to the evaluation criteria based on local similarities, such as lDDT and CAD, it is now clear that single model accuracy methods perform relatively better than consensus-based methods.     

Improving a consensus approach for protein structure selection by removing redundancy

In protein tertiary structure prediction, a crucial step is to select near-native structures from a large number of predicted structural models. Over the years, extensive research has been conducted for the protein structure selection problem with most approaches focusing on developing more accurate energy or scoring functions. Despite significant advances in this area, the discerning power of current approaches is still unsatisfactory. In this paper, we propose a novel consensus-based algorithm for the selection of predicted protein structures. Given a set of predicted models, our method first removes redundant structures to derive a subset of reference models. Then, a structure is ranked based on its average pairwise similarity to the reference models. Using the CASP8 data set containing a large collection of predicted models for 122 targets, we compared our method with the best CASP8 quality assessment (QA) servers, which are all consensus based, and showed that our QA scores correlate better with the GDT-TSs than those of the CASP8 QA servers. We also compared our method with the state-of-the-art scoring functions and showed its improved performance for near-native model selection. The GDT-TSs of the top models picked by our method are on average more than 8 percent better than the ones selected by the best performing scoring function.     

A further leap of improvement in tertiary structure prediction in CASP13 prompts new routes for future assessments

We present our assessment of tertiary structure predictions for hard targets in Critical Assessment of Structure Prediction round 13 (CASP13). The analysis includes (a) assignment and discussion of best models through scores-aided visual inspection of models for each evaluation unit (EU); (b) ranking of predictors resulting from this evaluation and from global scores; and (c) evaluation of progress, state of the art, and current limitations of protein structure prediction. We witness a sizable improvement in tertiary structure prediction building on the progress observed from CASP11 to CASP12, with (a) top models reaching backbone RMSD <3 å for several EUs of size <150 residues, contributed by many groups; (b) at least one model that roughly captures global topology for all EUs, probably unprecedented in this track of CASP; and (c) even quite good models for full, unsplit targets. Better structure predictions are brought about mainly by improved residue-residue contact predictions, and since this CASP also by distance predictions, achieved through state-of-the-art machine learning methods which also progressed to work with slightly shallower alignments compared to CASP12. As we reach a new realm of tertiary structure prediction quality, new directions are proposed and explored for future CASPs: (a) dropping splitting into EUs, (b) rethinking difficulty metrics probably in terms of contact and distance predictions, (c) assessing also side chains for models of high backbone accuracy, and (d) assessing residue-wise and possibly residue-residue quality estimates.     

Applications of the molecular dynamics flexible fitting method

In recent years, cryo-electron microscopy (cryo-EM) has established itself as a key method in structural biology, permitting the structural characterization of large biomolecular complexes in various functional states. The data obtained through single-particle cryo-EM has recently seen a leap in resolution thanks to landmark advances in experimental and computational techniques, resulting in sub-nanometer resolution structures being obtained routinely. The remaining gap between these data and revealing the mechanisms of molecular function can be closed through hybrid modeling tools that incorporate known atomic structures into the cryo-EM data. One such tool, molecular dynamics flexible fitting (MDFF), uses molecular dynamics simulations to combine structures from X-ray crystallography with cryo-EM density maps to derive atomic models of large biomolecular complexes. The structures furnished by MDFF can be used subsequently in computational investigations aimed at revealing the dynamics of the complexes under study. In the present work, recent applications of MDFF are presented, including the interpretation of cryo-EM data of the ribosome at different stages of translation and the structure of a membrane-curvature-inducing photosynthetic complex.     

Atomic-level protein structure refinement using fragment-guided molecular dynamics conformation sampling

One of critical difficulties of molecular dynamics (MD)?simulations in protein structure refinement is that the?physics-based energy landscape lacks?a middle-range funnel to guide nonnative conformations toward near-native states. We propose to use the target model as a probe to identify fragmental analogs from PDB. The distance maps are then used to reshape the MD energy funnel. The protocol was tested on 181 benchmarking and 26 CASP targets. It was found that structure models of correct folds with TM-score >0.5 can be often pulled closer to native with higher GDT-HA score, but improvement for the models of incorrect folds (TM-score <0.5) are much less pronounced. These data indicate that template-based fragmental distance maps essentially reshaped the MD energy landscape from golf-course-like to funnel-like ones in the successfully refined targets with a radius of TM-score ～0.5. These results demonstrate a new avenue to improve high-resolution structures by combining knowledge-based template information with physics-based MD simulations.     

Fitting low-resolution cryo-EM maps of proteins using constrained geometric simulations

Recent experimental advances in producing density maps from cryo-electron microscopy (cryo-EM) have challenged theorists to develop improved techniques to provide structural models that are consistent with the data and that preserve all the local stereochemistry associated with the biomolecule. We develop a new technique that maintains the local geometry and chemistry at each stage of the fitting procedure. A geometric simulation is used to drive the structure from some appropriate starting point (a nearby experimental structure or a modeled structure) toward the experimental density, via a set of small incremental motions. Structural motifs such as α-helices can be held rigid during the fitting procedure as the starting structure is brought into alignment with the experimental density. After validating this procedure on simulated data for adenylate kinase and lactoferrin, we show how cryo-EM data for two different GroEL structures can be fit using a starting x-ray crystal structure. We show that by incorporating the correct local stereochemistry in the modeling, structures can be obtained with effective resolution that is significantly higher than might be expected from the nominal cryo-EM resolution.     

Assessment of protein model structure accuracy estimation in CASP13: Challenges in the era of deep learning

Scoring model structure is an essential component of protein structure prediction that can affect the prediction accuracy tremendously. Users of protein structure prediction results also need to score models to select the best models for their application studies. In Critical Assessment of techniques for protein Structure Prediction (CASP), model accuracy estimation methods have been tested in a blind fashion by providing models submitted by the tertiary structure prediction servers for scoring. In CASP13, model accuracy estimation results were evaluated in terms of both global and local structure accuracy. Global structure accuracy estimation was evaluated by the quality of the models selected by the global structure scores and by the absolute estimates of the global scores. Residue-wise, local structure accuracy estimations were evaluated by three different measures. A new measure introduced in CASP13 evaluates the ability to predict inaccurately modeled regions that may be improved by refinement. An intensive comparative analysis on CASP13 and the previous CASPs revealed that the tertiary structure models generated by the CASP13 servers show very distinct features. Higher consensus toward models of higher global accuracy appeared even for free modeling targets, and many models of high global accuracy were not well optimized at the atomic level. This is related to the new technology in CASP13, deep learning for tertiary contact prediction. The tertiary model structures generated by deep learning pose a new challenge for EMA (estimation of model accuracy) method developers. Model accuracy estimation itself is also an area where deep learning can potentially have an impact, although current EMA methods have not fully explored that direction.     

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 ¹⁵N-¹H residual dipolar coupling data, typical of that obtained for ¹⁵N,¹³C-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.     

DeepTracer-ID: De novo protein identification from cryo-EM maps

The recent revolution in cryo-electron microscopy (cryo-EM) has made it possible to determine macromolecular structures directly from cell extracts. However, identifying the correct protein from the cryo-EM map is still challenging and often needs additional sequence information from other techniques, such as tandem mass spectrometry and/or bioinformatics. Here, we present DeepTracer-ID, a server-based approach to identify the candidate protein in a user-provided organism de novo from a cryo-EM map, without the need for additional information. Our method first uses DeepTracer to generate a protein backbone model that best represents the cryo-EM map, and this model is then searched against the library of AlphaFold2 predictions for all proteins in the given organism. This method is highly accurate and robust for high-resolution cryo-EM maps: in all 13 experimental maps tested blindly, DeepTracer-ID identified the correct proteins as the top candidates. Eight of the maps were of known structures, while the other five unpublished maps were validated by prior protein annotation and careful inspection of the model refined into the map. The program also showed promising results for both homomeric and heteromeric protein complexes. This platform is possible because of the recent breakthroughs in large-scale three-dimensional protein structure prediction.     

Cryo-EM density maps adjustment for subtraction,consensus and sharpening

Electron cryomicroscopy (cryo-EM) has emerged as a powerful structural biology instrument to solve near-atomic three-dimensional structures. Despite the fast growth in the number of density maps generated from cryo-EM data, comparison tools among these reconstructions are still lacking. Current proposals to compare cryo-EM data derived volumes perform map subtraction based on adjustment of each volume grey level to the same scale. We present here a more sophisticated way of adjusting the volumes before comparing, which implies adjustment of grey level scale and spectrum energy, but keeping phases intact inside a mask and imposing the results to be strictly positive. The adjustment that we propose leaves the volumes in the same numeric frame, allowing to perform operations among the adjusted volumes in a more reliable way. This adjustment can be a preliminary step for several applications such as comparison through subtraction, map sharpening, or combination of volumes through a consensus that selects the best resolved parts of each input map. Our development might also be used as a sharpening method using an atomic model as a reference. We illustrate the applicability of this algorithm with the reconstructions derived of several experimental examples. This algorithm is implemented in Xmipp software package and its applications are user-friendly accessible through the cryo-EM image processing framework Scipion.     

Prediction of protein oligomer structures using GALAXY in CASP13

Many proteins need to form oligomers to be functional, so oligomer structures provide important clues to biological roles of proteins. Prediction of oligomer structures therefore can be a useful tool in the absence of experimentally resolved structures. In this article, we describe the server and human methods that we used to predict oligomer structures in the CASP13 experiment. Performances of the methods on the 42 CASP13 oligomer targets consisting of 30 homo-oligomers and 12 hetero-oligomers are discussed. Our server method, Seok-assembly, generated models with interface contact similarity measure greater than 0.2 as model 1 for 11 homo-oligomer targets when proper templates existed in the database. Model refinement methods such as loop modeling and molecular dynamics (MD)-based overall refinement failed to improve model qualities when target proteins have domains not covered by templates or when chains have very small interfaces. In human predictions, additional experimental data such as low-resolution electron microscopy (EM) map were utilized. EM data could assist oligomer structure prediction by providing a global shape of the complex structure.     

Introducing “best single template” models as reference baseline for the Continuous Automated Model Evaluation (CAMEO)

Critical blind assessment of structure prediction techniques is crucial for the scientific community to establish the state of the art, identify bottlenecks, and guide future developments. In Critical Assessment of Techniques in Structure Prediction (CASP), human experts assess the performance of participating methods in relation to the difficulty of the prediction task in a biennial experiment on approximately 100 targets. Yet, the development of automated computational modeling methods requires more frequent evaluation cycles and larger sets of data. The “Continuous Automated Model EvaluatiOn (CAMEO)” platform complements CASP by conducting fully automated blind prediction evaluations based on the weekly pre-release of sequences of those structures, which are going to be published in the next release of the Protein Data Bank (PDB). Each week, CAMEO publishes benchmarking results for predictions corresponding to a set of about 20 targets collected during a 4-day prediction window. CAMEO benchmarking data are generated consistently for all methods at the same point in time, enabling developers to cross-validate their method's performance, and referring to their results in publications. Many successful participants of CASP have used CAMEO—either by directly benchmarking their methods within the system or by comparing their own performance to CAMEO reference data. CAMEO offers a variety of scores reflecting different aspects of structure modeling, for example, binding site accuracy, homo-oligomer interface quality, or accuracy of local model confidence estimates. By introducing the "bestSingleTemplate" method based on structure superpositions as a reference for the accuracy of 3D modeling predictions, CAMEO facilitates objective comparison of techniques and fosters the development of advanced methods.     

Constructing and validating initial Cα models from subnanometer resolution density maps with pathwalking

A significant number of macromolecular structures solved by electron cryo-microscopy and X-ray crystallography obtain resolutions of 3.5-6?, at which direct atomistic interpretation is difficult. To address this, we developed pathwalking, a semi-automated protocol to enumerate reasonable Cα models from near-atomic resolution density maps without a structural template or sequence-structure correspondence. Pathwalking uses an approach derived from the Traveling Salesman Problem to rapidly generate an ensemble of initial models for individual proteins, which can later be optimized to produce full atomic models. Pathwalking can also be used to validate and identify potential structural ambiguities in models generated from near-atomic resolution density maps. In this work, examples from the EMDB and PDB are used to assess the broad applicability and accuracy of our method. With the growing number of near-atomic resolution density maps from cryo-EM and X-ray crystallography, pathwalking can become an important tool in modeling protein structures.     

Nhs: network-based hierarchical segmentation for cryo-electron microscopy density maps

Cryo-electron microscopy (cryo-EM) experiments yield low-resolution (3-30 ?) 3D-density maps of macromolecules. These density maps are segmented to identify structurally distinct proteins, protein domains, and subunits. Such partitioning aids the inference of protein motions and guides fitting of high-resolution atomistic structures. Cryo-EM density map segmentation has traditionally required tedious and subjective manual partitioning or semisupervised computational methods, whereas validation of resulting segmentations has remained an open problem in this field. We introduce a network-based hierarchical segmentation (Nhs) method, that provides a multi-scale partitioning, reflecting local and global clustering, while requiring no user input. This approach models each map as a graph, where map voxels constitute nodes and weighted edges connect neighboring voxels. Nhs initiates Markov diffusion (or random walk) on the weighted graph. As Markov probabilities homogenize through diffusion, an intrinsic segmentation emerges. We validate the segmentations with ground-truth maps based on atomistic models. When implemented on density maps in the 2010 Cryo-EM Modeling Challenge, Nhs efficiently and objectively partitions macromolecules into structurally and functionally relevant subregions at multiple scales.     

Cryo-EM of macromolecular assemblies at near-atomic resolution

With single-particle electron cryomicroscopy (cryo-EM), it is possible to visualize large, macromolecular assemblies in near-native states. Although subnanometer resolutions have been routinely achieved for many specimens, state of the art cryo-EM has pushed to near-atomic (3.3-4.6 ?) resolutions. At these resolutions, it is now possible to construct reliable atomic models directly from the cryo-EM density map. In this study, we describe our recently developed protocols for performing the three-dimensional reconstruction and modeling of Mm-cpn, a group II chaperonin, determined to 4.3 ? resolution. This protocol, utilizing the software tools EMAN, Gorgon and Coot, can be adapted for use with nearly all specimens imaged with cryo-EM that target beyond 5 ? resolution. Additionally, the feature recognition and computational modeling tools can be applied to any near-atomic resolution density maps, including those from X-ray crystallography.     

Exploring global distortions of biological macromolecules and assemblies from low-resolution structural information and elastic network theory

A theory of elastic normal modes is described for the exploration of global distortions of biological structures and their assemblies based upon low-resolution image data. Structural information at low resolution, e.g. from density maps measured by cryogenic electron microscopy (cryo-EM), is used to construct discrete multi-resolution models for the electron density using the techniques of vector quantization. The elastic normal modes computed based on these discretized low-resolution models are found to compare well with the normal modes obtained at atomic resolution. The quality of the normal modes describing global displacements of the molecular system is found to depend on the resolution of the synthetic EM data and the extent of reductionism in the discretized representation. However, models that reproduce the functional rearrangements of our test set of molecules are achieved for realistic values of experimental resolution. Thus large conformational changes as occur during the functioning of biological macromolecules and assemblies can be elucidated directly from low-resolution structural data through the application of elastic normal mode theory and vector quantization.