Assessment of three-dimensional RNA structure prediction in CASP15.

The prediction of RNA three-dimensional structures remains an unsolved problem. Here, we report assessments of RNA structure predictions in CASP15, the first CASP exercise that involved RNA structure modeling. Forty-two predictor groups submitted models for at least one of twelve RNA-containing targets. These models were evaluated by the RNA-Puzzles organizers and, separately, by a CASP-recruited team using metrics (GDT, lDDT) and approaches (Z-score rankings) initially developed for assessment of proteins and generalized here for RNA assessment. The two assessments independently ranked the same predictor groups as first (AIchemy_RNA2), second (Chen), and third (RNAPolis and GeneSilico, tied); predictions from deep learning approaches were significantly worse than these top ranked groups, which did not use deep learning. Further analyses based on direct comparison of predicted models to cryogenic electron microscopy (cryo-EM) maps and x-ray diffraction data support these rankings. With the exception of two RNA-protein complexes, models submitted by CASP15 groups correctly predicted the global fold of the RNA targets. Comparisons of CASP15 submissions to designed RNA nanostructures as well as molecular replacement trials highlight the potential utility of current RNA modeling approaches for RNA nanotechnology and structural biology, respectively. Nevertheless, challenges remain in modeling fine details such as noncanonical pairs, in ranking among submitted models, and in prediction of multiple structures resolved by cryo-EM or crystallography.

Tertiary structure assessment at CASP15.

The results of tertiary structure assessment at CASP15 are reported. For the first time, recognizing the outstanding performance of AlphaFold 2 (AF2) at CASP14, all single-chain predictions were assessed together, irrespective of whether a template was available. At CASP15, there was no single stand-out group, with most of the best-scoring groups-led by PEZYFoldings, UM-TBM, and Yang Server-employing AF2 in one way or another. Many top groups paid special attention to generating deep Multiple Sequence Alignments (MSAs) and testing variant MSAs, thereby allowing them to successfully address some of the hardest targets. Such difficult targets, as well as lacking templates, were typically proteins with few homologues. Local divergence between prediction and target correlated with localization at crystal lattice or chain interfaces, and with regions exhibiting high B-factor factors in crystal structure targets, and should not necessarily be considered as representing error in the prediction. However, analysis of exposed and buried side chain accuracy showed room for improvement even in the latter. Nevertheless, a majority of groups produced high-quality predictions for most targets, which are valuable for experimental structure determination, functional analysis, and many other tasks across biology. These include those applying methods similar to those used to generate major resources such as the AlphaFold Protein Structure Database and the ESM Metagenomic atlas: the confidence estimates of the former were also notably accurate.

Predicted models and CCP4.

In late 2020, the results of CASP14, the 14th event in a series of competitions to assess the latest developments in computational protein structure-prediction methodology, revealed the giant leap forward that had been made by Google's Deepmind in tackling the prediction problem. The level of accuracy in their predictions was the first instance of a competitor achieving a global distance test score of better than 90 across all categories of difficulty. This achievement represents both a challenge and an opportunity for the field of experimental structural biology. For structure determination by macromolecular X-ray crystallography, access to highly accurate structure predictions is of great benefit, particularly when it comes to solving the phase problem. Here, details of new utilities and enhanced applications in the CCP4 suite, designed to allow users to exploit predicted models in determining macromolecular structures from X-ray diffraction data, are presented. The focus is mainly on applications that can be used to solve the phase problem through molecular replacement.

Structural insights into pink-eyed dilution protein (Oca2).

Recent innovations in computational structural biology have opened an opportunity to revise our current understanding of the structure and function of clinically important proteins. This study centres on human Oca2 which is located on mature melanosomal membranes. Mutations of Oca2 can result in a form of oculocutanous albinism, which is the most prevalent and visually identifiable form of albinism. Sequence analysis predicts Oca2 to be a member of the SLC13 transporter family, but it has not been classified into any existing SLC families. The modelling of Oca2 with AlphaFold2 and other advanced methods show that, like SLC13 members, it consists of a scaffold and transport domain and displays a pseudo inverted repeat topology that includes re-entrant loops. This finding contradicts the prevailing consensus view of its topology. In addition to the scaffold and transport domains, the presence of a cryptic GOLD domain is revealed that is likely responsible for its trafficking from the endoplasmic reticulum to the Golgi prior to localisation at the melanosomes. The GOLD domain harbours some known glycosylation sites. Analysis of the putative ligand binding site of the model shows the presence of highly conserved key asparagine residues that suggest Oca2 may be a Na+/dicarboxylate symporter. Known critical pathogenic mutations map to structural features present in the repeat regions that form the transport domain. Exploiting the AlphaFold2 multimeric modelling protocol in combination with conventional homology modelling allowed the building of plausible homodimers in both inward- and outward-facing conformations, supporting an elevator-type transport mechanism.

Assessment of three-dimensional RNA structure prediction in CASP15.

The prediction of RNA three-dimensional structures remains an unsolved problem. Here, we report assessments of RNA structure predictions in CASP15, the first CASP exercise that involved RNA structure modeling. Forty two predictor groups submitted models for at least one of twelve RNA-containing targets. These models were evaluated by the RNA-Puzzles organizers and, separately, by a CASP-recruited team using metrics (GDT, lDDT) and approaches (Z-score rankings) initially developed for assessment of proteins and generalized here for RNA assessment. The two assessments independently ranked the same predictor groups as first (AIchemy_RNA2), second (Chen), and third (RNAPolis and GeneSilico, tied); predictions from deep learning approaches were significantly worse than these top ranked groups, which did not use deep learning. Further analyses based on direct comparison of predicted models to cryogenic electron microscopy (cryo-EM) maps and X-ray diffraction data support these rankings. With the exception of two RNA-protein complexes, models submitted by CASP15 groups correctly predicted the global fold of the RNA targets. Comparisons of CASP15 submissions to designed RNA nanostructures as well as molecular replacement trials highlight the potential utility of current RNA modeling approaches for RNA nanotechnology and structural biology, respectively. Nevertheless, challenges remain in modeling fine details such as non-canonical pairs, in ranking among submitted models, and in prediction of multiple structures resolved by cryo-EM or crystallography.

CCP4 Cloud for structure determination and project management in macromolecular crystallography.

Nowadays, progress in the determination of three-dimensional macromolecular structures from diffraction images is achieved partly at the cost of increasing data volumes. This is due to the deployment of modern high-speed, high-resolution detectors, the increased complexity and variety of crystallographic software, the use of extensive databases and high-performance computing. This limits what can be accomplished with personal, offline, computing equipment in terms of both productivity and maintainability. There is also an issue of long-term data maintenance and availability of structure-solution projects as the links between experimental observations and the final results deposited in the PDB. In this article, CCP4 Cloud, a new front-end of the CCP4 software suite, is presented which mitigates these effects by providing an online, cloud-based environment for crystallographic computation. CCP4 Cloud was developed for the efficient delivery of computing power, database services and seamless integration with web resources. It provides a rich graphical user interface that allows project sharing and long-term storage for structure-solution projects, and can be linked to data-producing facilities. The system is distributed with the CCP4 software suite version 7.1 and higher, and an online publicly available instance of CCP4 Cloud is provided by CCP4.

MrParse: finding homologues in the PDB and the EBI AlphaFold database for molecular replacement and more.

Crystallographers have an array of search-model options for structure solution by molecular replacement (MR). The well established options of homologous experimental structures and regular secondary-structure elements or motifs are increasingly supplemented by computational modelling. Such modelling may be carried out locally or may use pre-calculated predictions retrieved from databases such as the EBI AlphaFold database. MrParse is a new pipeline to help to streamline the decision process in MR by consolidating bioinformatic predictions in one place. When reflection data are provided, MrParse can rank any experimental homologues found using eLLG, which indicates the likelihood that a given search model will work in MR. Inbuilt displays of predicted secondary structure, coiled-coil and transmembrane regions further inform the choice of MR protocol. MrParse can also identify and rank homologues in the EBI AlphaFold database, a function that will also interest other structural biologists and bioinformaticians.

Pathogenomic analyses of Shigella isolates inform factors limiting shigellosis prevention and control across LMICs.

Shigella spp. are the leading bacterial cause of severe childhood diarrhoea in low- and middle-income countries (LMICs), are increasingly antimicrobial resistant and have no widely available licenced vaccine. We performed genomic analyses of 1,246 systematically collected shigellae sampled from seven countries in sub-Saharan Africa and South Asia as part of the Global Enteric Multicenter Study (GEMS) between 2007 and 2011, to inform control and identify factors that could limit the effectiveness of current approaches. Through contemporaneous comparison among major subgroups, we found that S. sonnei contributes ≥6-fold more disease than other Shigella species relative to its genomic diversity, and highlight existing diversity and adaptative capacity among S. flexneri that may generate vaccine escape variants in <6 months. Furthermore, we show convergent evolution of resistance against ciprofloxacin, the current WHO-recommended antimicrobial for the treatment of shigellosis, among Shigella isolates. This demonstrates the urgent need to integrate existing genomic diversity into vaccine and treatment plans for Shigella, providing a framework for the focused application of comparative genomics to guide vaccine development, and the optimization of control and prevention strategies for other pathogens relevant to public health policy considerations.

findMySequence: a neural-network-based approach for identification of unknown proteins in X-ray crystallography and cryo-EM.

Although experimental protein-structure determination usually targets known proteins, chains of unknown sequence are often encountered. They can be purified from natural sources, appear as an unexpected fragment of a well characterized protein or appear as a contaminant. Regardless of the source of the problem, the unknown protein always requires characterization. Here, an automated pipeline is presented for the identification of protein sequences from cryo-EM reconstructions and crystallographic data. The method's application to characterize the crystal structure of an unknown protein purified from a snake venom is presented. It is also shown that the approach can be successfully applied to the identification of protein sequences and validation of sequence assignments in cryo-EM protein structures.

Redeployment of automated MrBUMP search-model identification for map fitting in cryo-EM.

In crystallography, the phase problem can often be addressed by the careful preparation of molecular-replacement search models. This has led to the development of pipelines such as MrBUMP that can automatically identify homologous proteins from an input sequence and edit them to focus on the areas that are most conserved. Many of these approaches can be applied directly to cryo-EM to help discover, prepare and correctly place models (here called cryo-EM search models) into electrostatic potential maps. This can significantly reduce the amount of manual model building that is required for structure determination. Here, MrBUMP is repurposed to fit automatically obtained PDB-derived chains and domains into cryo-EM maps. MrBUMP was successfully able to identify and place cryo-EM search models across a range of resolutions. Methods such as map segmentation are also explored as potential routes to improved performance. Map segmentation was also found to improve the effectiveness of the pipeline for higher resolution (<8 Å) data sets.

ConPlot: web-based application for the visualization of protein contact maps integrated with other data.

SUMMARY: Covariance-based predictions of residue contacts and inter-residue distances are an increasingly popular data type in protein bioinformatics. Here we present ConPlot, a web-based application for convenient display and analysis of contact maps and distograms. Integration of predicted contact data with other predictions is often required to facilitate inference of structural features. ConPlot can therefore use the empty space near the contact map diagonal to display multiple coloured tracks representing other sequence-based predictions. Popular file formats are natively read and bespoke data can also be flexibly displayed. This novel visualization will enable easier interpretation of predicted contact maps. AVAILABILITY AND IMPLEMENTATION: available online at www.conplot.org, along with documentation and examples. Alternatively, ConPlot can be installed and used locally using the docker image from the project's Docker Hub repository. ConPlot is licensed under the BSD 3-Clause. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

Assessing the utility of CASP14 models for molecular replacement.

The assessment of CASP models for utility in molecular replacement is a measure of their use in a valuable real-world application. In CASP7, the metric for molecular replacement assessment involved full likelihood-based molecular replacement searches; however, this restricted the assessable targets to crystal structures with only one copy of the target in the asymmetric unit, and to those where the search found the correct pose. In CASP10, full molecular replacement searches were replaced by likelihood-based rigid-body refinement of models superimposed on the target using the LGA algorithm, with the metric being the refined log-likelihood-gain (LLG) score. This enabled multi-copy targets and very poor models to be evaluated, but a significant further issue remained: the requirement of diffraction data for assessment. We introduce here the relative-expected-LLG (reLLG), which is independent of diffraction data. This reLLG is also independent of any crystal form, and can be calculated regardless of the source of the target, be it X-ray, NMR or cryo-EM. We calibrate the reLLG against the LLG for targets in CASP14, showing that it is a robust measure of both model and group ranking. Like the LLG, the reLLG shows that accurate coordinate error estimates add substantial value to predicted models. We find that refinement by CASP groups can often convert an inadequate initial model into a successful MR search model. Consistent with findings from others, we show that the AlphaFold2 models are sufficiently good, and reliably so, to surpass other current model generation strategies for attempting molecular replacement phasing.

High-accuracy protein structure prediction in CASP14.

The application of state-of-the-art deep-learning approaches to the protein modeling problem has expanded the "high-accuracy" category in CASP14 to encompass all targets. Building on the metrics used for high-accuracy assessment in previous CASPs, we evaluated the performance of all groups that submitted models for at least 10 targets across all difficulty classes, and judged the usefulness of those produced by AlphaFold2 (AF2) as molecular replacement search models with AMPLE. Driven by the qualitative diversity of the targets submitted to CASP, we also introduce DipDiff as a new measure for the improvement in backbone geometry provided by a model versus available templates. Although a large leap in high-accuracy is seen due to AF2, the second-best method in CASP14 out-performed the best in CASP13, illustrating the role of community-based benchmarking in the development and evolution of the protein structure prediction field.

Evaluation of model refinement in CASP14.

We report here an assessment of the model refinement category of the 14th round of Critical Assessment of Structure Prediction (CASP14). As before, predictors submitted up to five ranked refinements, along with associated residue-level error estimates, for targets that had a wide range of starting quality. The ability of groups to accurately rank their submissions and to predict coordinate error varied widely. Overall, only four groups out-performed a "naïve predictor" corresponding to the resubmission of the starting model. Among the top groups, there are interesting differences of approach and in the spread of improvements seen: some methods are more conservative, others more adventurous. Some targets were "double-barreled" for which predictors were offered a high-quality AlphaFold 2 (AF2)-derived prediction alongside another of lower quality. The AF2-derived models were largely unimprovable, many of their apparent errors being found to reside at domain and, especially, crystal lattice contacts. Refinement is shown to have a mixed impact overall on structure-based function annotation methods to predict nucleic acid binding, spot catalytic sites, and dock protein structures.

Helical ensembles outperform ideal helices in molecular replacement.

The conventional approach in molecular replacement is the use of a related structure as a search model. However, this is not always possible as the availability of such structures can be scarce for poorly characterized families of proteins. In these cases, alternative approaches can be explored, such as the use of small ideal fragments that share high, albeit local, structural similarity with the unknown protein. Earlier versions of AMPLE enabled the trialling of a library of ideal helices, which worked well for largely helical proteins at suitable resolutions. Here, the performance of libraries of helical ensembles created by clustering helical segments is explored. The impacts of different B-factor treatments and different degrees of structural heterogeneity are explored. A 30% increase in the number of solutions obtained by AMPLE was observed when using this new set of ensembles compared with the performance with ideal helices. The boost in performance was notable across three different fold classes: transmembrane, globular and coiled-coil structures. Furthermore, the increased effectiveness of these ensembles was coupled to a reduction in the time required by AMPLE to reach a solution. AMPLE users can now take full advantage of this new library of search models by activating the `helical ensembles' mode.

Using Phaser and ensembles to improve the performance of SIMBAD.

The conventional approach to search-model identification in molecular replacement (MR) is to screen a database of known structures using the target sequence. However, this strategy is not always effective, for example when the relationship between sequence and structural similarity fails or when the crystal contents are not those expected. An alternative approach is to identify suitable search models directly from the experimental data. SIMBAD is a sequence-independent MR pipeline that uses either a crystal lattice search or MR functions to directly locate suitable search models from databases. The previous version of SIMBAD used the fast AMoRe rotation-function search. Here, a new version of SIMBAD which makes use of Phaser and its likelihood scoring to improve the sensitivity of the pipeline is presented. It is shown that the additional compute time potentially required by the more sophisticated scoring is counterbalanced by the greater sensitivity, allowing more cases to trigger early-termination criteria, rather than running to completion. Using Phaser solved 17 out of 25 test cases in comparison to the ten solved with AMoRe, and it is shown that use of ensemble search models produces additional performance benefits.

Molecular replacement using structure predictions from databases.

Molecular replacement (MR) is the predominant route to solution of the phase problem in macromolecular crystallography. Where the lack of a suitable homologue precludes conventional MR, one option is to predict the target structure using bioinformatics. Such modelling, in the absence of homologous templates, is called ab initio or de novo modelling. Recently, the accuracy of such models has improved significantly as a result of the availability, in many cases, of residue-contact predictions derived from evolutionary covariance analysis. Covariance-assisted ab initio models representing structurally uncharacterized Pfam families are now available on a large scale in databases, potentially representing a valuable and easily accessible supplement to the PDB as a source of search models. Here, the unconventional MR pipeline AMPLE is employed to explore the value of structure predictions in the GREMLIN and PconsFam databases. It was tested whether these deposited predictions, processed in various ways, could solve the structures of PDB entries that were subsequently deposited. The results were encouraging: nine of 27 GREMLIN cases were solved, covering target lengths of 109-355 residues and a resolution range of 1.4-2.9 Å, and with target-model shared sequence identity as low as 20%. The cluster-and-truncate approach in AMPLE proved to be essential for most successes. For the overall lower quality structure predictions in the PconsFam database, remodelling with Rosetta within the AMPLE pipeline proved to be the best approach, generating ensemble search models from single-structure deposits. Finally, it is shown that the AMPLE-obtained search models deriving from GREMLIN deposits are of sufficiently high quality to be selected by the sequence-independent MR pipeline SIMBAD. Overall, the results help to point the way towards the optimal use of the expanding databases of ab initio structure predictions.

SIMBAD: a sequence-independent molecular-replacement pipeline.

The conventional approach to finding structurally similar search models for use in molecular replacement (MR) is to use the sequence of the target to search against those of a set of known structures. Sequence similarity often correlates with structure similarity. Given sufficient similarity, a known structure correctly positioned in the target cell by the MR process can provide an approximation to the unknown phases of the target. An alternative approach to identifying homologous structures suitable for MR is to exploit the measured data directly, comparing the lattice parameters or the experimentally derived structure-factor amplitudes with those of known structures. Here, SIMBAD, a new sequence-independent MR pipeline which implements these approaches, is presented. SIMBAD can identify cases of contaminant crystallization and other mishaps such as mistaken identity (swapped crystallization trays), as well as solving unsequenced targets and providing a brute-force approach where sequence-dependent search-model identification may be nontrivial, for example because of conformational diversity among identifiable homologues. The program implements a three-step pipeline to efficiently identify a suitable search model in a database of known structures. The first step performs a lattice-parameter search against the entire Protein Data Bank (PDB), rapidly determining whether or not a homologue exists in the same crystal form. The second step is designed to screen the target data for the presence of a crystallized contaminant, a not uncommon occurrence in macromolecular crystallography. Solving structures with MR in such cases can remain problematic for many years, since the search models, which are assumed to be similar to the structure of interest, are not necessarily related to the structures that have actually crystallized. To cater for this eventuality, SIMBAD rapidly screens the data against a database of known contaminant structures. Where the first two steps fail to yield a solution, a final step in SIMBAD can be invoked to perform a brute-force search of a nonredundant PDB database provided by the MoRDa MR software. Through early-access usage of SIMBAD, this approach has solved novel cases that have otherwise proved difficult to solve.

Recent developments in MrBUMP: better search-model preparation, graphical interaction with search models, and solution improvement and assessment.

Increasing sophistication in molecular-replacement (MR) software and the rapid expansion of the PDB in recent years have allowed the technique to become the dominant method for determining the phases of a target structure in macromolecular X-ray crystallography. In addition, improvements in bioinformatic techniques for finding suitable homologous structures for use as MR search models, combined with developments in refinement and model-building techniques, have pushed the applicability of MR to lower sequence identities and made weak MR solutions more amenable to refinement and improvement. MrBUMP is a CCP4 pipeline which automates all stages of the MR procedure. Its scope covers everything from the sourcing and preparation of suitable search models right through to rebuilding of the positioned search model. Recent improvements to the pipeline include the adoption of more sensitive bioinformatic tools for sourcing search models, enhanced model-preparation techniques including better ensembling of homologues, and the use of phase improvement and model building on the resulting solution. The pipeline has also been deployed as an online service through CCP4 online, which allows its users to exploit large bioinformatic databases and coarse-grained parallelism to speed up the determination of a possible solution. Finally, the molecular-graphics application CCP4mg has been combined with MrBUMP to provide an interactive visual aid to the user during the process of selecting and manipulating search models for use in MR. Here, these developments in MrBUMP are described with a case study to explore how some of the enhancements to the pipeline and to CCP4mg can help to solve a difficult case.

GP0.4 from bacteriophage T7: in silico characterisation of its structure and interaction with E. coli FtsZ.

BACKGROUND: Proteins produced by bacteriophages can have potent antimicrobial activity. The study of phage-host interactions can therefore inform small molecule drug discovery by revealing and characterising new drug targets. Here we characterise in silico the predicted interaction of gene protein 0.4 (GP0.4) from the Escherichia coli (E. coli) phage T7 with E. coli filamenting temperature-sensitive mutant Z division protein (FtsZ). FtsZ is a tubulin homolog which plays a key role in bacterial cell division and that has been proposed as a drug target. RESULTS: Using ab initio, fragment assembly structure modelling, we predicted the structure of GP0.4 with two programs. A structure similarity-based network was used to identify a U-shaped helix-turn-helix candidate fold as being favoured. ClusPro was used to dock this structure prediction to a homology model of E. coli FtsZ resulting in a favourable predicted interaction mode. Alternative docking methods supported the proposed mode which offered an immediate explanation for the anti-filamenting activity of GP0.4. Importantly, further strong support derived from a previously characterised insertion mutation, known to abolish GP0.4 activity, that is positioned in close proximity to the proposed GP0.4/FtsZ interface. CONCLUSIONS: The mode of interaction predicted by bioinformatics techniques strongly suggests a mechanism through which GP0.4 inhibits FtsZ and further establishes the latter's druggable intrafilament interface as a potential drug target.