AlphaFold predictions are valuable hypotheses and accelerate but do not replace experimental structure determination.

Artificial intelligence-based protein structure prediction methods such as AlphaFold have revolutionized structural biology. The accuracies of these predictions vary, however, and they do not take into account ligands, covalent modifications or other environmental factors. Here, we evaluate how well AlphaFold predictions can be expected to describe the structure of a protein by comparing predictions directly with experimental crystallographic maps. In many cases, AlphaFold predictions matched experimental maps remarkably closely. In other cases, even very high-confidence predictions differed from experimental maps on a global scale through distortion and domain orientation, and on a local scale in backbone and side-chain conformation. We suggest considering AlphaFold predictions as exceptionally useful hypotheses. We further suggest that it is important to consider the confidence in prediction when interpreting AlphaFold predictions and to carry out experimental structure determination to verify structural details, particularly those that involve interactions not included in the prediction.

Outcomes of the EMDataResource cryo-EM Ligand Modeling Challenge.

The EMDataResource Ligand Model Challenge aimed to assess the reliability and reproducibility of modeling ligands bound to protein and protein-nucleic acid complexes in cryogenic electron microscopy (cryo-EM) maps determined at near-atomic (1.9-2.5 Å) resolution. Three published maps were selected as targets: Escherichia coli beta-galactosidase with inhibitor, SARS-CoV-2 virus RNA-dependent RNA polymerase with covalently bound nucleotide analog and SARS-CoV-2 virus ion channel ORF3a with bound lipid. Sixty-one models were submitted from 17 independent research groups, each with supporting workflow details. The quality of submitted ligand models and surrounding atoms were analyzed by visual inspection and quantification of local map quality, model-to-map fit, geometry, energetics and contact scores. A composite rather than a single score was needed to assess macromolecule+ligand model quality. These observations lead us to recommend best practices for assessing cryo-EM structures of liganded macromolecules reported at near-atomic resolution.

Improved AlphaFold modeling with implicit experimental information.

Machine-learning prediction algorithms such as AlphaFold and RoseTTAFold can create remarkably accurate protein models, but these models usually have some regions that are predicted with low confidence or poor accuracy. We hypothesized that by implicitly including new experimental information such as a density map, a greater portion of a model could be predicted accurately, and that this might synergistically improve parts of the model that were not fully addressed by either machine learning or experiment alone. An iterative procedure was developed in which AlphaFold models are automatically rebuilt on the basis of experimental density maps and the rebuilt models are used as templates in new AlphaFold predictions. We show that including experimental information improves prediction beyond the improvement obtained with simple rebuilding guided by the experimental data. This procedure for AlphaFold modeling with density has been incorporated into an automated procedure for interpretation of crystallographic and electron cryo-microscopy maps.

Cryo-EM model validation recommendations based on outcomes of the 2019 EMDataResource challenge.

This paper describes outcomes of the 2019 Cryo-EM Model Challenge. The goals were to (1) assess the quality of models that can be produced from cryogenic electron microscopy (cryo-EM) maps using current modeling software, (2) evaluate reproducibility of modeling results from different software developers and users and (3) compare performance of current metrics used for model evaluation, particularly Fit-to-Map metrics, with focus on near-atomic resolution. Our findings demonstrate the relatively high accuracy and reproducibility of cryo-EM models derived by 13 participating teams from four benchmark maps, including three forming a resolution series (1.8 to 3.1 Å). The results permit specific recommendations to be made about validating near-atomic cryo-EM structures both in the context of individual experiments and structure data archives such as the Protein Data Bank. We recommend the adoption of multiple scoring parameters to provide full and objective annotation and assessment of the model, reflective of the observed cryo-EM map density.

Improvement of cryo-EM maps by density modification.

A density-modification procedure for improving maps from single-particle electron cryogenic microscopy (cryo-EM) is presented. The theoretical basis of the method is identical to that of maximum-likelihood density modification, previously used to improve maps from macromolecular X-ray crystallography. Key differences from applications in crystallography are that the errors in Fourier coefficients are largely in the phases in crystallography but in both phases and amplitudes in cryo-EM, and that half-maps with independent errors are available in cryo-EM. These differences lead to a distinct approach for combination of information from starting maps with information obtained in the density-modification process. The density-modification procedure was applied to a set of 104 datasets and improved map-model correlation and increased the visibility of details in many of the maps. The procedure requires two unmasked half-maps and a sequence file or other source of information on the volume of the macromolecule that has been imaged.

The cryo-electron microscopy structure of human transcription factor IIH.

Human transcription factor IIH (TFIIH) is part of the general transcriptional machinery required by RNA polymerase II for the initiation of eukaryotic gene transcription. Composed of ten subunits that add up to a molecular mass of about 500 kDa, TFIIH is also essential for nucleotide excision repair. The seven-subunit TFIIH core complex formed by XPB, XPD, p62, p52, p44, p34, and p8 is competent for DNA repair, while the CDK-activating kinase subcomplex, which includes the kinase activity of CDK7 as well as the cyclin H and MAT1 subunits, is additionally required for transcription initiation. Mutations in the TFIIH subunits XPB, XPD, and p8 lead to severe premature ageing and cancer propensity in the genetic diseases xeroderma pigmentosum, Cockayne syndrome, and trichothiodystrophy, highlighting the importance of TFIIH for cellular physiology. Here we present the cryo-electron microscopy structure of human TFIIH at 4.4 Å resolution. The structure reveals the molecular architecture of the TFIIH core complex, the detailed structures of its constituent XPB and XPD ATPases, and how the core and kinase subcomplexes of TFIIH are connected. Additionally, our structure provides insight into the conformational dynamics of TFIIH and the regulation of its activity.

A fully automatic method yielding initial models from high-resolution cryo-electron microscopy maps.

We report a fully automated procedure for the optimization and interpretation of reconstructions from cryo-electron microscopy (cryo-EM) data, available in Phenix as phenix.map_to_model. We applied our approach to 476 datasets with resolution of 4.5 Å or better, including reconstructions of 47 ribosomes and 32 other protein-RNA complexes. The median fraction of residues in the deposited structures reproduced automatically was 71% for reconstructions determined at resolutions of 3 Å or better and 47% for those at resolutions worse than 3 Å.

Accurate model annotation of a near-atomic resolution cryo-EM map.

Electron cryomicroscopy (cryo-EM) has been used to determine the atomic coordinates (models) from density maps of biological assemblies. These models can be assessed by their overall fit to the experimental data and stereochemical information. However, these models do not annotate the actual density values of the atoms nor their positional uncertainty. Here, we introduce a computational procedure to derive an atomic model from a cryo-EM map with annotated metadata. The accuracy of such a model is validated by a faithful replication of the experimental cryo-EM map computed using the coordinates and associated metadata. The functional interpretation of any structural features in the model and its utilization for future studies can be made in the context of its measure of uncertainty. We applied this protocol to the 3.3-Å map of the mature P22 bacteriophage capsid, a large and complex macromolecular assembly. With this protocol, we identify and annotate previously undescribed molecular interactions between capsid subunits that are crucial to maintain stability in the absence of cementing proteins or cross-linking, as occur in other bacteriophages.

Cryo_fit: Democratization of flexible fitting for cryo-EM.

Cryo-electron microscopy (cryo-EM) is becoming a method of choice for describing native conformations of biomolecular complexes at high resolution. The rapid growth of cryo-EM in recent years has created a high demand for automated solutions, both in hardware and software. Flexible fitting of atomic models to three-dimensional (3D) cryo-EM reconstructions by molecular dynamics (MD) simulation is a popular technique but often requires technical expertise in computer simulation. This work introduces cryo_fit, a package for the automatic flexible fitting of atomic models in cryo-EM maps using MD simulation. The package is integrated with the Phenix software suite. The module was designed to automate the multiple steps of MD simulation in a reproducible manner, as well as facilitate refinement and validation through Phenix. Through the use of cryo_fit, scientists with little experience in MD simulation can produce high quality atomic models automatically and better exploit the potential of cryo-EM.

Map segmentation, automated model-building and their application to the Cryo-EM Model Challenge.

A recently-developed method for identifying a compact, contiguous region representing the unique part of a density map was applied to 218 Cryo-EM maps with resolutions of 4.5 Šor better. The key elements of the segmentation procedure are (1) identification of all regions of density above a threshold and (2) choice of a unique set of these regions, taking symmetry into consideration, that maximize connectivity and compactness. This segmentation approach was then combined with tools for automated map sharpening and model-building to generate models for the 12 maps in the 2016 Cryo-EM Model Challenge in a fully automated manner. The resulting models have completeness from 24% to 82% and RMS distances from reference interpretations of 0.6 Å-2.1 Å.

Validated near-atomic resolution structure of bacteriophage epsilon15 derived from cryo-EM and modeling.

High-resolution structures of viruses have made important contributions to modern structural biology. Bacteriophages, the most diverse and abundant organisms on earth, replicate and infect all bacteria and archaea, making them excellent potential alternatives to antibiotics and therapies for multidrug-resistant bacteria. Here, we improved upon our previous electron cryomicroscopy structure of Salmonella bacteriophage epsilon15, achieving a resolution sufficient to determine the tertiary structures of both gp7 and gp10 protein subunits that form the T = 7 icosahedral lattice. This study utilizes recently established best practice for near-atomic to high-resolution (3-5 Å) electron cryomicroscopy data evaluation. The resolution and reliability of the density map were cross-validated by multiple reconstructions from truly independent data sets, whereas the models of the individual protein subunits were validated adopting the best practices from X-ray crystallography. Some sidechain densities are clearly resolved and show the subunit-subunit interactions within and across the capsomeres that are required to stabilize the virus. The presence of the canonical phage and jellyroll viral protein folds, gp7 and gp10, respectively, in the same virus suggests that epsilon15 may have emerged more recently relative to other bacteriophages.

The solvent component of macromolecular crystals.

The mother liquor from which a biomolecular crystal is grown will contain water, buffer molecules, native ligands and cofactors, crystallization precipitants and additives, various metal ions, and often small-molecule ligands or inhibitors. On average, about half the volume of a biomolecular crystal consists of this mother liquor, whose components form the disordered bulk solvent. Its scattering contributions can be exploited in initial phasing and must be included in crystal structure refinement as a bulk-solvent model. Concomitantly, distinct electron density originating from ordered solvent components must be correctly identified and represented as part of the atomic crystal structure model. Herein, are reviewed (i) probabilistic bulk-solvent content estimates, (ii) the use of bulk-solvent density modification in phase improvement, (iii) bulk-solvent models and refinement of bulk-solvent contributions and (iv) modelling and validation of ordered solvent constituents. A brief summary is provided of current tools for bulk-solvent analysis and refinement, as well as of modelling, refinement and analysis of ordered solvent components, including small-molecule ligands.

From deep TLS validation to ensembles of atomic models built from elemental motions.

The translation-libration-screw model first introduced by Cruickshank, Schomaker and Trueblood describes the concerted motions of atomic groups. Using TLS models can improve the agreement between calculated and experimental diffraction data. Because the T, L and S matrices describe a combination of atomic vibrations and librations, TLS models can also potentially shed light on molecular mechanisms involving correlated motions. However, this use of TLS models in mechanistic studies is hampered by the difficulties in translating the results of refinement into molecular movement or a structural ensemble. To convert the matrices into a constituent molecular movement, the matrix elements must satisfy several conditions. Refining the T, L and S matrix elements as independent parameters without taking these conditions into account may result in matrices that do not represent concerted molecular movements. Here, a mathematical framework and the computational tools to analyze TLS matrices, resulting in either explicit decomposition into descriptions of the underlying motions or a report of broken conditions, are described. The description of valid underlying motions can then be output as a structural ensemble. All methods are implemented as part of the PHENIX project.

FEM: feature-enhanced map.

A method is presented that modifies a 2mFobs - DFmodel σA-weighted map such that the resulting map can strengthen a weak signal, if present, and can reduce model bias and noise. The method consists of first randomizing the starting map and filling in missing reflections using multiple methods. This is followed by restricting the map to regions with convincing density and the application of sharpening. The final map is then created by combining a series of histogram-equalized intermediate maps. In the test cases shown, the maps produced in this way are found to have increased interpretability and decreased model bias compared with the starting 2mFobs - DFmodel σA-weighted map.

Predicting X-ray diffuse scattering from translation-libration-screw structural ensembles.

Identifying the intramolecular motions of proteins and nucleic acids is a major challenge in macromolecular X-ray crystallography. Because Bragg diffraction describes the average positional distribution of crystalline atoms with imperfect precision, the resulting electron density can be compatible with multiple models of motion. Diffuse X-ray scattering can reduce this degeneracy by reporting on correlated atomic displacements. Although recent technological advances are increasing the potential to accurately measure diffuse scattering, computational modeling and validation tools are still needed to quantify the agreement between experimental data and different parameterizations of crystalline disorder. A new tool, phenix.diffuse, addresses this need by employing Guinier's equation to calculate diffuse scattering from Protein Data Bank (PDB)-formatted structural ensembles. As an example case, phenix.diffuse is applied to translation-libration-screw (TLS) refinement, which models rigid-body displacement for segments of the macromolecule. To enable the calculation of diffuse scattering from TLS-refined structures, phenix.tls_as_xyz builds multi-model PDB files that sample the underlying T, L and S tensors. In the glycerophosphodiesterase GpdQ, alternative TLS-group partitioning and different motional correlations between groups yield markedly dissimilar diffuse scattering maps with distinct implications for molecular mechanism and allostery. These methods demonstrate how, in principle, X-ray diffuse scattering could extend macromolecular structural refinement, validation and analysis.

Metrics for comparison of crystallographic maps.

Numerical comparison of crystallographic contour maps is used extensively in structure solution and model refinement, analysis and validation. However, traditional metrics such as the map correlation coefficient (map CC, real-space CC or RSCC) sometimes contradict the results of visual assessment of the corresponding maps. This article explains such apparent contradictions and suggests new metrics and tools to compare crystallographic contour maps. The key to the new methods is rank scaling of the Fourier syntheses. The new metrics are complementary to the usual map CC and can be more helpful in map comparison, in particular when only some of their aspects, such as regions of high density, are of interest.

Flexible torsion-angle noncrystallographic symmetry restraints for improved macromolecular structure refinement.

One of the great challenges in refining macromolecular crystal structures is a low data-to-parameter ratio. Historically, knowledge from chemistry has been used to help to improve this ratio. When a macromolecule crystallizes with more than one copy in the asymmetric unit, the noncrystallographic symmetry relationships can be exploited to provide additional restraints when refining the working model. However, although globally similar, NCS-related chains often have local differences. To allow for local differences between NCS-related molecules, flexible torsion-based NCS restraints have been introduced, coupled with intelligent rotamer handling for protein chains, and are available in phenix.refine for refinement of models at all resolutions.

Automated identification of elemental ions in macromolecular crystal structures.

Many macromolecular model-building and refinement programs can automatically place solvent atoms in electron density at moderate-to-high resolution. This process frequently builds water molecules in place of elemental ions, the identification of which must be performed manually. The solvent-picking algorithms in phenix.refine have been extended to build common ions based on an analysis of the chemical environment as well as physical properties such as occupancy, B factor and anomalous scattering. The method is most effective for heavier elements such as calcium and zinc, for which a majority of sites can be placed with few false positives in a diverse test set of structures. At atomic resolution, it is observed that it can also be possible to identify tightly bound sodium and magnesium ions. A number of challenges that contribute to the difficulty of completely automating the process of structure completion are discussed.

Automating crystallographic structure solution and refinement of protein-ligand complexes.

High-throughput drug-discovery and mechanistic studies often require the determination of multiple related crystal structures that only differ in the bound ligands, point mutations in the protein sequence and minor conformational changes. If performed manually, solution and refinement requires extensive repetition of the same tasks for each structure. To accelerate this process and minimize manual effort, a pipeline encompassing all stages of ligand building and refinement, starting from integrated and scaled diffraction intensities, has been implemented in Phenix. The resulting system is able to successfully solve and refine large collections of structures in parallel without extensive user intervention prior to the final stages of model completion and validation.

Accounting for nonuniformity of bulk-solvent: A mosaic model.

A flat mask-based model is almost universally used in macromolecular crystallography to account for disordered (bulk) solvent. This model assumes any voxel of the crystal unit cell that is not occupied by the atomic model is occupied by the solvent. The properties of this solvent are assumed to be exactly the same across the whole volume of the unit cell. While this is a reasonable approximation in practice, there are a number of scenarios where this model becomes suboptimal. In this work, we enumerate several of these scenarios and describe a new generalized approach to modeling the bulk-solvent which we refer to as mosaic bulk-solvent model. The mosaic bulk-solvent model allows nonuniform features of the solvent in the crystal to be accounted for in a computationally efficient way. It is implemented in the computational crystallography toolbox and the Phenix software.