Benchmarking of AlphaFold2 accuracy self-estimates as indicators of empirical model quality and ranking: a comparison with independent model quality assessment programmes.

MOTIVATION: Despite an increase in protein modelling accuracy following the development of AlphaFold2, there remains an accuracy gap between predicted and observed model quality assessment (MQA) scores. In CASP15, variations in AlphaFold2 model accuracy prediction were noticed for quaternary models of very similar observed quality. In this study, we compare plDDT and pTM to their observed counterparts the local distance difference test (lDDT) and TM-score for both tertiary and quaternary models to examine whether reliability is retained across the scoring range under normal modelling conditions and in situations where AlphaFold2 functionality is customized. We also explore plDDT and pTM ranking accuracy in comparison with the published independent MQA programmes ModFOLD9 and ModFOLDdock. RESULTS: plDDT was found to be an accurate descriptor of tertiary model quality compared to observed lDDT-Cα scores (Pearson r = 0.97), and achieved a ranking agreement true positive rate (TPR) of 0.34 with observed scores, which ModFOLD9 could not improve. However, quaternary structure accuracy was reduced (plDDT r = 0.67, pTM r = 0.70) and significant overprediction was seen with both scores for some lower quality models. Additionally, ModFOLDdock was able to improve upon AF2-Multimer model ranking compared to TM-score (TPR 0.34) and oligo-lDDT score (TPR 0.43). Finally, evidence is presented for increased variability in plDDT and pTM when using custom template recycling, which is more pronounced for quaternary structures. AVAILABILITY AND IMPLEMENTATION: The ModFOLD9 and ModFOLDdock quality assessment servers are available at https://www.reading.ac.uk/bioinf/ModFOLD/ and https://www.reading.ac.uk/bioinf/ModFOLDdock/, respectively. A docker image is available at https://hub.docker.com/r/mcguffin/multifold.

Prediction of protein structures, functions and interactions using the IntFOLD7, MultiFOLD and ModFOLDdock servers.

The IntFOLD server based at the University of Reading has been a leading method over the past decade in providing free access to accurate prediction of protein structures and functions. In a post-AlphaFold2 world, accurate models of tertiary structures are widely available for even more protein targets, so there has been a refocus in the prediction community towards the accurate modelling of protein-ligand interactions as well as modelling quaternary structure assemblies. In this paper, we describe the latest improvements to IntFOLD, which maintains its competitive structure prediction performance by including the latest deep learning methods while also integrating accurate model quality estimates and 3D models of protein-ligand interactions. Furthermore, we also introduce our two new server methods: MultiFOLD for accurately modelling both tertiary and quaternary structures, with performance which has been independently verified to outperform the standard AlphaFold2 methods, and ModFOLDdock, which provides world-leading quality estimates for quaternary structure models. The IntFOLD7, MultiFOLD and ModFOLDdock servers are available at: https://www.reading.ac.uk/bioinf/.

Estimation of model accuracy in CASP15 using the ModFOLDdock server.

In CASP15, there was a greater emphasis on multimeric modeling than in previous experiments, with assembly structures nearly doubling in number (41 up from 22) since the previous round. CASP15 also included a new estimation of model accuracy (EMA) category in recognition of the importance of objective quality assessment (QA) for quaternary structure models. ModFOLDdock is a multimeric model QA server developed by the McGuffin group at the University of Reading, which brings together a range of single-model, clustering, and deep learning methods to form a consensus of approaches. For CASP15, three variants of ModFOLDdock were developed to optimize for the different facets of the quality estimation problem. The standard ModFOLDdock variant produced predicted scores optimized for positive linear correlations with the observed scores. The ModFOLDdockR variant produced predicted scores optimized for ranking, that is, the top-ranked models have the highest accuracy. In addition, the ModFOLDdockS variant used a quasi-single model approach to score each model on an individual basis. The scores from all three variants achieved strongly positive Pearson correlation coefficients with the CASP observed scores (oligo-lDDT) in excess of 0.70, which were maintained across both homomeric and heteromeric model populations. In addition, at least one of the ModFOLDdock variants was consistently ranked in the top two methods across all three EMA categories. Specifically, for overall global fold prediction accuracy, ModFOLDdock placed second and ModFOLDdockR placed third; for overall interface quality prediction accuracy, ModFOLDdockR, ModFOLDdock, and ModFOLDdockS were placed above all other predictor methods, and ModFOLDdockR and ModFOLDdockS were placed second and third respectively for individual residue confidence scores. The ModFOLDdock server is available at: https://www.reading.ac.uk/bioinf/ModFOLDdock/. ModFOLDdock is also available as part of the MultiFOLD docker package: https://hub.docker.com/r/mcguffin/multifold.

Structural, functional, and mechanistic insights uncover the fundamental role of orphan connexin-62 in platelets.

Connexins oligomerise to form hexameric hemichannels in the plasma membrane that can further dock together on adjacent cells to form gap junctions and facilitate intercellular trafficking of molecules. In this study, we report the expression and function of an orphan connexin, connexin-62 (Cx62), in human and mouse (Cx57, mouse homolog) platelets. A novel mimetic peptide (62Gap27) was developed to target the second extracellular loop of Cx62, and 3-dimensional structural models predicted its interference with gap junction and hemichannel function. The ability of 62Gap27 to regulate both gap junction and hemichannel-mediated intercellular communication was observed using fluorescence recovery after photobleaching analysis and flow cytometry. Cx62 inhibition by 62Gap27 suppressed a range of agonist-stimulated platelet functions and impaired thrombosis and hemostasis. This was associated with elevated protein kinase A-dependent signaling in a cyclic adenosine monophosphate-independent manner and was not observed in Cx57-deficient mouse platelets (in which the selectivity of 62Gap27 for this connexin was also confirmed). Notably, Cx62 hemichannels were observed to function independently of Cx37 and Cx40 hemichannels. Together, our data reveal a fundamental role for a hitherto uncharacterized connexin in regulating the function of circulating cells.

ReFOLD3: refinement of 3D protein models with gradual restraints based on predicted local quality and residue contacts.

ReFOLD3 is unique in its application of gradual restraints, calculated from local model quality estimates and contact predictions, which are used to guide the refinement of theoretical 3D protein models towards the native structures. ReFOLD3 achieves improved performance by using an iterative refinement protocol to fix incorrect residue contacts and local errors, including unusual bonds and angles, which are identified in the submitted models by our leading ModFOLD8 model quality assessment method. Following refinement, the likely resulting improvements to the submitted models are recognized by ModFOLD8, which produces both global and local quality estimates. During the CASP14 prediction season (May-Aug 2020), we used the ReFOLD3 protocol to refine hundreds of 3D models, for both the refinement and the main tertiary structure prediction categories. Our group improved the global and local quality scores for numerous starting models in the refinement category, where we ranked in the top 10 according to the official assessment. The ReFOLD3 protocol was also used for the refinement of the SARS-CoV-2 targets as a part of the CASP Commons COVID-19 initiative, and we provided a significant number of the top 10 models. The ReFOLD3 web server is freely available at https://www.reading.ac.uk/bioinf/ReFOLD/.

ModFOLD8: accurate global and local quality estimates for 3D protein models.

Methods for estimating the quality of 3D models of proteins are vital tools for driving the acceptance and utility of predicted tertiary structures by the wider bioscience community. Here we describe the significant major updates to ModFOLD, which has maintained its position as a leading server for the prediction of global and local quality of 3D protein models, over the past decade (>20 000 unique external users). ModFOLD8 is the latest version of the server, which combines the strengths of multiple pure-single and quasi-single model methods. Improvements have been made to the web server interface and there has been successive increases in prediction accuracy, which were achieved through integration of newly developed scoring methods and advanced deep learning-based residue contact predictions. Each version of the ModFOLD server has been independently blind tested in the biennial CASP experiments, as well as being continuously evaluated via the CAMEO project. In CASP13 and CASP14, the ModFOLD7 and ModFOLD8 variants ranked among the top 10 quality estimation methods according to almost every official analysis. Prior to CASP14, ModFOLD8 was also applied for the evaluation of SARS-CoV-2 protein models as part of CASP Commons 2020 initiative. The ModFOLD8 server is freely available at: https://www.reading.ac.uk/bioinf/ModFOLD/.

MAP4K4 expression in cardiomyocytes: multiple isoforms, multiple phosphorylations and interactions with striatins.

The Ser/Thr kinase MAP4K4, like other GCKIV kinases, has N-terminal kinase and C-terminal citron homology (CNH) domains. MAP4K4 can activate c-Jun N-terminal kinases (JNKs), and studies in the heart suggest it links oxidative stress to JNKs and heart failure. In other systems, MAP4K4 is regulated in striatin-interacting phosphatase and kinase (STRIPAK) complexes, in which one of three striatins tethers PP2A adjacent to a kinase to keep it dephosphorylated and inactive. Our aim was to understand how MAP4K4 is regulated in cardiomyocytes. The rat MAP4K4 gene was not properly defined. We identified the first coding exon of the rat gene using 5'-RACE, we cloned the full-length sequence and confirmed alternative-splicing of MAP4K4 in rat cardiomyocytes. We identified an additional α-helix C-terminal to the kinase domain important for kinase activity. In further studies, FLAG-MAP4K4 was expressed in HEK293 cells or cardiomyocytes. The Ser/Thr protein phosphatase inhibitor calyculin A (CalA) induced MAP4K4 hyperphosphorylation, with phosphorylation of the activation loop and extensive phosphorylation of the linker between the kinase and CNH domains. This required kinase activity. MAP4K4 associated with myosin in untreated cardiomyocytes, and this was lost with CalA-treatment. FLAG-MAP4K4 associated with all three striatins in cardiomyocytes, indicative of regulation within STRIPAK complexes and consistent with activation by CalA. Computational analysis suggested the interaction was direct and mediated via coiled-coil domains. Surprisingly, FLAG-MAP4K4 inhibited JNK activation by H2O2 in cardiomyocytes and increased myofibrillar organisation. Our data identify MAP4K4 as a STRIPAK-regulated kinase in cardiomyocytes, and suggest it regulates the cytoskeleton rather than activates JNKs.

IntFOLD: an integrated web resource for high performance protein structure and function prediction.

The IntFOLD server provides a unified resource for the automated prediction of: protein tertiary structures with built-in estimates of model accuracy (EMA), protein structural domain boundaries, natively unstructured or disordered regions in proteins, and protein-ligand interactions. The component methods have been independently evaluated via the successive blind CASP experiments and the continual CAMEO benchmarking project. The IntFOLD server has established its ranking as one of the best performing publicly available servers, based on independent official evaluation metrics. Here, we describe significant updates to the server back end, where we have focused on performance improvements in tertiary structure predictions, in terms of global 3D model quality and accuracy self-estimates (ASE), which we achieve using our newly improved ModFOLD7_rank algorithm. We also report on various upgrades to the front end including: a streamlined submission process, enhanced visualization of models, new confidence scores for ranking, and links for accessing all annotated model data. Furthermore, we now include an option for users to submit selected models for further refinement via convenient push buttons. The IntFOLD server is freely available at: http://www.reading.ac.uk/bioinf/IntFOLD/.

PINOT: an intuitive resource for integrating protein-protein interactions.

BACKGROUND: The past decade has seen the rise of omics data for the understanding of biological systems in health and disease. This wealth of information includes protein-protein interaction (PPI) data derived from both low- and high-throughput assays, which are curated into multiple databases that capture the extent of available information from the peer-reviewed literature. Although these curation efforts are extremely useful, reliably downloading and integrating PPI data from the variety of available repositories is challenging and time consuming. METHODS: We here present a novel user-friendly web-resource called PINOT (Protein Interaction Network Online Tool; available at http://www.reading.ac.uk/bioinf/PINOT/PINOT_form.html) to optimise the collection and processing of PPI data from IMEx consortium associated repositories (members and observers) and WormBase, for constructing, respectively, human and Caenorhabditis elegans PPI networks. RESULTS: Users submit a query containing a list of proteins of interest for which PINOT extracts data describing PPIs. At every query submission PPI data are downloaded, merged and quality assessed. Then each PPI is confidence scored based on the number of distinct methods used for interaction detection and the number of publications that report the specific interaction. Examples of how PINOT can be applied are provided to highlight the performance, ease of use and potential utility of this tool. CONCLUSIONS: PINOT is a tool that allows users to survey the curated literature, extracting PPI data in relation to a list of proteins of interest. PINOT extracts a similar numbers of PPIs as other, analogous, tools and incorporates a set of innovative features. PINOT is able to process large queries, it downloads human PPIs live through PSICQUIC and it applies quality control filters on the downloaded PPI data (i.e. removing the need for manual inspection by the user). PINOT provides the user with information on detection methods and publication history for each downloaded interaction data entry and outputs the results in a table format that can be straightforwardly further customised and/or directly uploaded into network visualization software. Video abstract.

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.

ModFOLD6: an accurate web server for the global and local quality estimation of 3D protein models.

Methods that reliably estimate the likely similarity between the predicted and native structures of proteins have become essential for driving the acceptance and adoption of three-dimensional protein models by life scientists. ModFOLD6 is the latest version of our leading resource for Estimates of Model Accuracy (EMA), which uses a pioneering hybrid quasi-single model approach. The ModFOLD6 server integrates scores from three pure-single model methods and three quasi-single model methods using a neural network to estimate local quality scores. Additionally, the server provides three options for producing global score estimates, depending on the requirements of the user: (i) ModFOLD6_rank, which is optimized for ranking/selection, (ii) ModFOLD6_cor, which is optimized for correlations of predicted and observed scores and (iii) ModFOLD6 global for balanced performance. The ModFOLD6 methods rank among the top few for EMA, according to independent blind testing by the CASP12 assessors. The ModFOLD6 server is also continuously automatically evaluated as part of the CAMEO project, where significant performance gains have been observed compared to our previous server and other publicly available servers. The ModFOLD6 server is freely available at: http://www.reading.ac.uk/bioinf/ModFOLD/.

ReFOLD: a server for the refinement of 3D protein models guided by accurate quality estimates.

ReFOLD is a novel hybrid refinement server with integrated high performance global and local Accuracy Self Estimates (ASEs). The server attempts to identify and to fix likely errors in user supplied 3D models of proteins via successive rounds of refinement. The server is unique in providing output for multiple alternative refined models in a way that allows users to quickly visualize the key residue locations, which are likely to have been improved. This is important, as global refinement of a full chain model may not always be possible, whereas local regions, or individual domains, can often be much improved. Thus, users may easily compare the specific regions of the alternative refined models in which they are most interested e.g. key interaction sites or domains. ReFOLD was used to generate hundreds of alternative refined models for the CASP12 experiment, boosting our group's performance in the main tertiary structure prediction category. Our successful refinement of initial server models combined with our built-in ASEs were instrumental to our second place ranking on Template Based Modeling (TBM) and Free Modeling (FM)/TBM targets. The ReFOLD server is freely available at: http://www.reading.ac.uk/bioinf/ReFOLD/.

Methods for estimation of model accuracy in CASP12.

Methods to reliably estimate the quality of 3D models of proteins are essential drivers for the wide adoption and serious acceptance of protein structure predictions by life scientists. In this article, the most successful groups in CASP12 describe their latest methods for estimates of model accuracy (EMA). We show that pure single model accuracy estimation methods have shown clear progress since CASP11; the 3 top methods (MESHI, ProQ3, SVMQA) all perform better than the top method of CASP11 (ProQ2). Although the pure single model accuracy estimation methods outperform quasi-single (ModFOLD6 variations) and consensus methods (Pcons, ModFOLDclust2, Pcomb-domain, and Wallner) in model selection, they are still not as good as those methods in absolute model quality estimation and predictions of local quality. Finally, we show that when using contact-based model quality measures (CAD, lDDT) the single model quality methods perform relatively better.

Accurate template-based modeling in CASP12 using the IntFOLD4-TS, ModFOLD6, and ReFOLD methods.

Our aim in CASP12 was to improve our Template-Based Modeling (TBM) methods through better model selection, accuracy self-estimate (ASE) scores and refinement. To meet this aim, we developed two new automated methods, which we used to score, rank, and improve upon the provided server models. Firstly, the ModFOLD6_rank method, for improved global Quality Assessment (QA), model ranking and the detection of local errors. Secondly, the ReFOLD method for fixing errors through iterative QA guided refinement. For our automated predictions we developed the IntFOLD4-TS protocol, which integrates the ModFOLD6_rank method for scoring the multiple-template models that were generated using a number of alternative sequence-structure alignments. Overall, our selection of top models and ASE scores using ModFOLD6_rank was an improvement on our previous approaches. In addition, it was worthwhile attempting to repair the detected errors in the top selected models using ReFOLD, which gave us an overall gain in performance. According to the assessors' formula, the IntFOLD4 server ranked 3rd/5th (average Z-score > 0.0/-2.0) on the server only targets, and our manual predictions (McGuffin group) ranked 1st/2nd (average Z-score > -2.0/0.0) compared to all other groups.

IntFOLD: an integrated server for modelling protein structures and functions from amino acid sequences.

IntFOLD is an independent web server that integrates our leading methods for structure and function prediction. The server provides a simple unified interface that aims to make complex protein modelling data more accessible to life scientists. The server web interface is designed to be intuitive and integrates a complex set of quantitative data, so that 3D modelling results can be viewed on a single page and interpreted by non-expert modellers at a glance. The only required input to the server is an amino acid sequence for the target protein. Here we describe major performance and user interface updates to the server, which comprises an integrated pipeline of methods for: tertiary structure prediction, global and local 3D model quality assessment, disorder prediction, structural domain prediction, function prediction and modelling of protein-ligand interactions. The server has been independently validated during numerous CASP (Critical Assessment of Techniques for Protein Structure Prediction) experiments, as well as being continuously evaluated by the CAMEO (Continuous Automated Model Evaluation) project. The IntFOLD server is available at: http://www.reading.ac.uk/bioinf/IntFOLD/.

The ModFOLD4 server for the quality assessment of 3D protein models.

Once you have generated a 3D model of a protein, how do you know whether it bears any resemblance to the actual structure? To determine the usefulness of 3D models of proteins, they must be assessed in terms of their quality by methods that predict their similarity to the native structure. The ModFOLD4 server is the latest version of our leading independent server for the estimation of both the global and local (per-residue) quality of 3D protein models. The server produces both machine readable and graphical output, providing users with intuitive visual reports on the quality of predicted protein tertiary structures. The ModFOLD4 server is freely available to all at: http://www.reading.ac.uk/bioinf/ModFOLD/.

The FunFOLD2 server for the prediction of protein-ligand interactions.

The FunFOLD2 server is a new independent server that integrates our novel protein-ligand binding site and quality assessment protocols for the prediction of protein function (FN) from sequence via structure. Our guiding principles were, first, to provide a simple unified resource to make our function prediction software easily accessible to all via a simple web interface and, second, to produce integrated output for predictions that can be easily interpreted. The server provides a clean web interface so that results can be viewed on a single page and interpreted by non-experts at a glance. The output for the prediction is an image of the top predicted tertiary structure annotated to indicate putative ligand-binding site residues. The results page also includes a list of the most likely binding site residues and the types of predicted ligands and their frequencies in similar structures. The protein-ligand interactions can also be interactively visualized in 3D using the Jmol plug-in. The raw machine readable data are provided for developers, which comply with the Critical Assessment of Techniques for Protein Structure Prediction data standards for FN predictions. The FunFOLD2 webserver is freely available to all at the following web site: http://www.reading.ac.uk/bioinf/FunFOLD/FunFOLD_form_2_0.html.

In silico identification and three-dimensional modelling of the missense mutation in ADAMTS2 in a sheep flock with dermatosparaxis.

BACKGROUND: Dermatosparaxis (Ehlers-Danlos syndrome in humans) is characterized by extreme fragility of the skin. It is due to the lack of mature collagen caused by a failure in the enzymatic processing of procollagen I. We investigated the condition in a commercial sheep flock. HYPOTHESIS/OBJECTIVES: Mutations in the ADAM metallopeptidase with thrombospondin type 1 motif, 2 (ADAMTS2) locus, are involved in the development of dermatosparaxis in humans, cattle and the dorper sheep breed; consequently, this locus was investigated in the flock. ANIMALS: A single affected lamb, its dam, the dam of a second affected lamb and the rams in the flock were studied. METHODS: DNA was purified from blood, PCR primers were used to detect parts of the ADAMS2 gene and nucleotide sequencing was performed using Sanger's procedure. Skin samples were examined using standard histology procedures. RESULTS: A missense mutation was identified in the catalytic domain of ADAMTS2. The mutation is predicted to cause the substitution in the mature ADAMTS2 of a valine molecule by a methionine molecule (V15M) affecting the catalytic domain of the enzyme. Both the 'sorting intolerant from tolerant' (SIFT) and the PolyPhen-2 methodologies predicted a damaging effect for the mutation. Three-dimensional modelling suggested that this mutation may alter the stability of the protein folding or distort the structure, causing the protein to malfunction. CONCLUSIONS AND CLINICAL IMPORTANCE: Detection of the mutation responsible for the pathology allowed us to remove the heterozygote ram, thus preventing additional cases in the flock.

Disorder Prediction Methods, Their Applicability to Different Protein Targets and Their Usefulness for Guiding Experimental Studies.

The role and function of a given protein is dependent on its structure. In recent years, however, numerous studies have highlighted the importance of unstructured, or disordered regions in governing a protein's function. Disordered proteins have been found to play important roles in pivotal cellular functions, such as DNA binding and signalling cascades. Studying proteins with extended disordered regions is often problematic as they can be challenging to express, purify and crystallise. This means that interpretable experimental data on protein disorder is hard to generate. As a result, predictive computational tools have been developed with the aim of predicting the level and location of disorder within a protein. Currently, over 60 prediction servers exist, utilizing different methods for classifying disorder and different training sets. Here we review several good performing, publicly available prediction methods, comparing their application and discussing how disorder prediction servers can be used to aid the experimental solution of protein structure. The use of disorder prediction methods allows us to adopt a more targeted approach to experimental studies by accurately identifying the boundaries of ordered protein domains so that they may be investigated separately, thereby increasing the likelihood of their successful experimental solution.

SOcK, MiSTs, MASK and STicKs: the GCKIII (germinal centre kinase III) kinases and their heterologous protein-protein interactions.

The GCKIII (germinal centre kinase III) subfamily of the mammalian Ste20 (sterile 20)-like group of serine/threonine protein kinases comprises SOK1 (Ste20-like/oxidant-stress-response kinase 1), MST3 (mammalian Ste20-like kinase 3) and MST4. Initially, GCKIIIs were considered in the contexts of the regulation of mitogen-activated protein kinase cascades and apoptosis. More recently, their participation in multiprotein heterocomplexes has become apparent. In the present review, we discuss the structure and phosphorylation of GCKIIIs and then focus on their interactions with other proteins. GCKIIIs possess a highly-conserved, structured catalytic domain at the N-terminus and a less-well conserved C-terminal regulatory domain. GCKIIIs are activated by tonic autophosphorylation of a T-loop threonine residue and their phosphorylation is regulated primarily through protein serine/threonine phosphatases [especially PP2A (protein phosphatase 2A)]. The GCKIII regulatory domains are highly disorganized, but can interact with more structured proteins, particularly the CCM3 (cerebral cavernous malformation 3)/PDCD10 (programmed cell death 10) protein. We explore the role(s) of GCKIIIs (and CCM3/PDCD10) in STRIPAK (striatin-interacting phosphatase and kinase) complexes and their association with the cis-Golgi protein GOLGA2 (golgin A2; GM130). Recently, an interaction of GCKIIIs with MO25 has been identified. This exhibits similarities to the STRADα (STE20-related kinase adaptor α)-MO25 interaction (as in the LKB1-STRADα-MO25 heterotrimer) and, at least for MST3, the interaction may be enhanced by cis-autophosphorylation of its regulatory domain. In these various heterocomplexes, GCKIIIs associate with the Golgi apparatus, the centrosome and the nucleus, as well as with focal adhesions and cell junctions, and are probably involved in cell migration, polarity and proliferation. Finally, we consider the association of GCKIIIs with a number of human diseases, particularly cerebral cavernous malformations.