Large-scale clustering of AlphaFold2 3D models shines light on the structure and function of proteins.

Two recent studies exploited ultra-fast structural aligners and deep-learning approaches to cluster the protein structure space in the AlphaFold Database. Barrio-Hernandez et al.1 and Durairaj et al.2 uncovered fascinating new protein functions and structural features previously unknown.

Novel machine learning approaches revolutionize protein knowledge.

Breakthrough methods in machine learning (ML), protein structure prediction, and novel ultrafast structural aligners are revolutionizing structural biology. Obtaining accurate models of proteins and annotating their functions on a large scale is no longer limited by time and resources. The most recent method to be top ranked by the Critical Assessment of Structure Prediction (CASP) assessment, AlphaFold 2 (AF2), is capable of building structural models with an accuracy comparable to that of experimental structures. Annotations of 3D models are keeping pace with the deposition of the structures due to advancements in protein language models (pLMs) and structural aligners that help validate these transferred annotations. In this review we describe how recent developments in ML for protein science are making large-scale structural bioinformatics available to the general scientific community.

Chainsaw: protein domain segmentation with fully convolutional neural networks.

MOTIVATION: Protein domains are fundamental units of protein structure and play a pivotal role in understanding folding, function, evolution, and design. The advent of accurate structure prediction techniques has resulted in an influx of new structural data, making the partitioning of these structures into domains essential for inferring evolutionary relationships and functional classification. RESULTS: This article presents Chainsaw, a supervised learning approach to domain parsing that achieves accuracy that surpasses current state-of-the-art methods. Chainsaw uses a fully convolutional neural network which is trained to predict the probability that each pair of residues is in the same domain. Domain predictions are then derived from these pairwise predictions using an algorithm that searches for the most likely assignment of residues to domains given the set of pairwise co-membership probabilities. Chainsaw matches CATH domain annotations in 78% of protein domains versus 72% for the next closest method. When predicting on AlphaFold models, expert human evaluators were twice as likely to prefer Chainsaw's predictions versus the next best method. AVAILABILITY AND IMPLEMENTATION: github.com/JudeWells/Chainsaw.

InterPro in 2022.

The InterPro database (https://www.ebi.ac.uk/interpro/) provides an integrative classification of protein sequences into families, and identifies functionally important domains and conserved sites. Here, we report recent developments with InterPro (version 90.0) and its associated software, including updates to data content and to the website. These developments extend and enrich the information provided by InterPro, and provide a more user friendly access to the data. Additionally, we have worked on adding Pfam website features to the InterPro website, as the Pfam website will be retired in late 2022. We also show that InterPro's sequence coverage has kept pace with the growth of UniProtKB. Moreover, we report the development of a card game as a method of engaging the non-scientific community. Finally, we discuss the benefits and challenges brought by the use of artificial intelligence for protein structure prediction.

Characterizing and explaining the impact of disease-associated mutations in proteins without known structures or structural homologs.

Mutations in human proteins lead to diseases. The structure of these proteins can help understand the mechanism of such diseases and develop therapeutics against them. With improved deep learning techniques, such as RoseTTAFold and AlphaFold, we can predict the structure of proteins even in the absence of structural homologs. We modeled and extracted the domains from 553 disease-associated human proteins without known protein structures or close homologs in the Protein Databank. We noticed that the model quality was higher and the Root mean square deviation (RMSD) lower between AlphaFold and RoseTTAFold models for domains that could be assigned to CATH families as compared to those which could only be assigned to Pfam families of unknown structure or could not be assigned to either. We predicted ligand-binding sites, protein-protein interfaces and conserved residues in these predicted structures. We then explored whether the disease-associated missense mutations were in the proximity of these predicted functional sites, whether they destabilized the protein structure based on ddG calculations or whether they were predicted to be pathogenic. We could explain 80% of these disease-associated mutations based on proximity to functional sites, structural destabilization or pathogenicity. When compared to polymorphisms, a larger percentage of disease-associated missense mutations were buried, closer to predicted functional sites, predicted as destabilizing and pathogenic. Usage of models from the two state-of-the-art techniques provide better confidence in our predictions, and we explain 93 additional mutations based on RoseTTAFold models which could not be explained based solely on AlphaFold models.

CATHe: detection of remote homologues for CATH superfamilies using embeddings from protein language models.

MOTIVATION: CATH is a protein domain classification resource that exploits an automated workflow of structure and sequence comparison alongside expert manual curation to construct a hierarchical classification of evolutionary and structural relationships. The aim of this study was to develop algorithms for detecting remote homologues missed by state-of-the-art hidden Markov model (HMM)-based approaches. The method developed (CATHe) combines a neural network with sequence representations obtained from protein language models. It was assessed using a dataset of remote homologues having less than 20% sequence identity to any domain in the training set. RESULTS: The CATHe models trained on 1773 largest and 50 largest CATH superfamilies had an accuracy of 85.6 ± 0.4% and 98.2 ± 0.3%, respectively. As a further test of the power of CATHe to detect more remote homologues missed by HMMs derived from CATH domains, we used a dataset consisting of protein domains that had annotations in Pfam, but not in CATH. By using highly reliable CATHe predictions (expected error rate <0.5%), we were able to provide CATH annotations for 4.62 million Pfam domains. For a subset of these domains from Homo sapiens, we structurally validated 90.86% of the predictions by comparing their corresponding AlphaFold2 structures with structures from the CATH superfamilies to which they were assigned. AVAILABILITY AND IMPLEMENTATION: The code for the developed models is available on https://github.com/vam-sin/CATHe, and the datasets developed in this study can be accessed on https://zenodo.org/record/6327572. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

The impact of structural bioinformatics tools and resources on SARS-CoV-2 research and therapeutic strategies.

SARS-CoV-2 is the causative agent of COVID-19, the ongoing global pandemic. It has posed a worldwide challenge to human health as no effective treatment is currently available to combat the disease. Its severity has led to unprecedented collaborative initiatives for therapeutic solutions against COVID-19. Studies resorting to structure-based drug design for COVID-19 are plethoric and show good promise. Structural biology provides key insights into 3D structures, critical residues/mutations in SARS-CoV-2 proteins, implicated in infectivity, molecular recognition and susceptibility to a broad range of host species. The detailed understanding of viral proteins and their complexes with host receptors and candidate epitope/lead compounds is the key to developing a structure-guided therapeutic design. Since the discovery of SARS-CoV-2, several structures of its proteins have been determined experimentally at an unprecedented speed and deposited in the Protein Data Bank. Further, specialized structural bioinformatics tools and resources have been developed for theoretical models, data on protein dynamics from computer simulations, impact of variants/mutations and molecular therapeutics. Here, we provide an overview of ongoing efforts on developing structural bioinformatics tools and resources for COVID-19 research. We also discuss the impact of these resources and structure-based studies, to understand various aspects of SARS-CoV-2 infection and therapeutic development. These include (i) understanding differences between SARS-CoV-2 and SARS-CoV, leading to increased infectivity of SARS-CoV-2, (ii) deciphering key residues in the SARS-CoV-2 involved in receptor-antibody recognition, (iii) analysis of variants in host proteins that affect host susceptibility to infection and (iv) analyses facilitating structure-based drug and vaccine design against SARS-CoV-2.

Characterizing domain-specific open educational resources by linking ISCB Communities of Special Interest to Wikipedia.

MOTIVATION: Wikipedia is one of the most important channels for the public communication of science and is frequently accessed as an educational resource in computational biology. Joint efforts between the International Society for Computational Biology (ISCB) and the Computational Biology taskforce of WikiProject Molecular Biology (a group of expert Wikipedia editors) have considerably improved computational biology representation on Wikipedia in recent years. However, there is still an urgent need for further improvement in quality, especially when compared to related scientific fields such as genetics and medicine. Facilitating involvement of members from ISCB Communities of Special Interest (COSIs) would improve a vital open education resource in computational biology, additionally allowing COSIs to provide a quality educational resource highly specific to their subfield. RESULTS: We generate a list of around 1500 English Wikipedia articles relating to computational biology and describe the development of a binary COSI-Article matrix, linking COSIs to relevant articles and thereby defining domain-specific open educational resources. Our analysis of the COSI-Article matrix data provides a quantitative assessment of computational biology representation on Wikipedia against other fields and at a COSI-specific level. Furthermore, we conducted similarity analysis and subsequent clustering of COSI-Article data to provide insight into potential relationships between COSIs. Finally, based on our analysis, we suggest courses of action to improve the quality of computational biology representation on Wikipedia.

Dissecting peripheral protein-membrane interfaces.

Peripheral membrane proteins (PMPs) include a wide variety of proteins that have in common to bind transiently to the chemically complex interfacial region of membranes through their interfacial binding site (IBS). In contrast to protein-protein or protein-DNA/RNA interfaces, peripheral protein-membrane interfaces are poorly characterized. We collected a dataset of PMP domains representative of the variety of PMP functions: membrane-targeting domains (Annexin, C1, C2, discoidin C2, PH, PX), enzymes (PLA, PLC/D) and lipid-transfer proteins (START). The dataset contains 1328 experimental structures and 1194 AphaFold models. We mapped the amino acid composition and structural patterns of the IBS of each protein in this dataset, and evaluated which were more likely to be found at the IBS compared to the rest of the domains' accessible surface. In agreement with earlier work we find that about two thirds of the PMPs in the dataset have protruding hydrophobes (Leu, Ile, Phe, Tyr, Trp and Met) at their IBS. The three aromatic amino acids Trp, Tyr and Phe are a hallmark of PMPs IBS regardless of whether they protrude on loops or not. This is also the case for lysines but not arginines suggesting that, unlike for Arg-rich membrane-active peptides, the less membrane-disruptive lysine is preferred in PMPs. Another striking observation was the over-representation of glycines at the IBS of PMPs compared to the rest of their surface, possibly procuring IBS loops a much-needed flexibility to insert in-between membrane lipids. The analysis of the 9 superfamilies revealed amino acid distribution patterns in agreement with their known functions and membrane-binding mechanisms. Besides revealing novel amino acids patterns at protein-membrane interfaces, our work contributes a new PMP dataset and an analysis pipeline that can be further built upon for future studies of PMPs properties, or for developing PMPs prediction tools using for example, machine learning approaches.

CATH: increased structural coverage of functional space.

CATH (https://www.cathdb.info) identifies domains in protein structures from wwPDB and classifies these into evolutionary superfamilies, thereby providing structural and functional annotations. There are two levels: CATH-B, a daily snapshot of the latest domain structures and superfamily assignments, and CATH+, with additional derived data, such as predicted sequence domains, and functionally coherent sequence subsets (Functional Families or FunFams). The latest CATH+ release, version 4.3, significantly increases coverage of structural and sequence data, with an addition of 65,351 fully-classified domains structures (+15%), providing 500 238 structural domains, and 151 million predicted sequence domains (+59%) assigned to 5481 superfamilies. The FunFam generation pipeline has been re-engineered to cope with the increased influx of data. Three times more sequences are captured in FunFams, with a concomitant increase in functional purity, information content and structural coverage. FunFam expansion increases the structural annotations provided for experimental GO terms (+59%). We also present CATH-FunVar web-pages displaying variations in protein sequences and their proximity to known or predicted functional sites. We present two case studies (1) putative cancer drivers and (2) SARS-CoV-2 proteins. Finally, we have improved links to and from CATH including SCOP, InterPro, Aquaria and 2DProt.

The InterPro protein families and domains database: 20 years on.

The InterPro database (https://www.ebi.ac.uk/interpro/) provides an integrative classification of protein sequences into families, and identifies functionally important domains and conserved sites. InterProScan is the underlying software that allows protein and nucleic acid sequences to be searched against InterPro's signatures. Signatures are predictive models which describe protein families, domains or sites, and are provided by multiple databases. InterPro combines signatures representing equivalent families, domains or sites, and provides additional information such as descriptions, literature references and Gene Ontology (GO) terms, to produce a comprehensive resource for protein classification. Founded in 1999, InterPro has become one of the most widely used resources for protein family annotation. Here, we report the status of InterPro (version 81.0) in its 20th year of operation, and its associated software, including updates to database content, the release of a new website and REST API, and performance improvements in InterProScan.

Assigning protein function from domain-function associations using DomFun.

BACKGROUND: Protein function prediction remains a key challenge. Domain composition affects protein function. Here we present DomFun, a Ruby gem that uses associations between protein domains and functions, calculated using multiple indices based on tripartite network analysis. These domain-function associations are combined at the protein level, to generate protein-function predictions. RESULTS: We analysed 16 tripartite networks connecting homologous superfamily and FunFam domains from CATH-Gene3D with functional annotations from the three Gene Ontology (GO) sub-ontologies, KEGG, and Reactome. We validated the results using the CAFA 3 benchmark platform for GO annotation, finding that out of the multiple association metrics and domain datasets tested, Simpson index for FunFam domain-function associations combined with Stouffer's method leads to the best performance in almost all scenarios. We also found that using FunFams led to better performance than superfamilies, and better results were found for GO molecular function compared to GO biological process terms. DomFun performed as well as the highest-performing method in certain CAFA 3 evaluation procedures in terms of [Formula: see text] and [Formula: see text] We also implemented our own benchmark procedure, Pathway Prediction Performance (PPP), which can be used to validate function prediction for additional annotations sources, such as KEGG and Reactome. Using PPP, we found similar results to those found with CAFA 3 for GO, moreover we found good performance for the other annotation sources. As with CAFA 3, Simpson index with Stouffer's method led to the top performance in almost all scenarios. CONCLUSIONS: DomFun shows competitive performance with other methods evaluated in CAFA 3 when predicting proteins function with GO, although results vary depending on the evaluation procedure. Through our own benchmark procedure, PPP, we have shown it can also make accurate predictions for KEGG and Reactome. It performs best when using FunFams, combining Simpson index derived domain-function associations using Stouffer's method. The tool has been implemented so that it can be easily adapted to incorporate other protein features, such as domain data from other sources, amino acid k-mers and motifs. The DomFun Ruby gem is available from https://rubygems.org/gems/DomFun . Code maintained at https://github.com/ElenaRojano/DomFun . Validation procedure scripts can be found at https://github.com/ElenaRojano/DomFun_project .

CATH functional families predict functional sites in proteins.

MOTIVATION: Identification of functional sites in proteins is essential for functional characterization, variant interpretation and drug design. Several methods are available for predicting either a generic functional site, or specific types of functional site. Here, we present FunSite, a machine learning predictor that identifies catalytic, ligand-binding and protein-protein interaction functional sites using features derived from protein sequence and structure, and evolutionary data from CATH functional families (FunFams). RESULTS: FunSite's prediction performance was rigorously benchmarked using cross-validation and a holdout dataset. FunSite outperformed other publicly available functional site prediction methods. We show that conserved residues in FunFams are enriched in functional sites. We found FunSite's performance depends greatly on the quality of functional site annotations and the information content of FunFams in the training data. Finally, we analyze which structural and evolutionary features are most predictive for functional sites. AVAILABILITYAND IMPLEMENTATION: https://github.com/UCL/cath-funsite-predictor. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

Clustering FunFams using sequence embeddings improves EC purity.

MOTIVATION: Classifying proteins into functional families can improve our understanding of protein function and can allow transferring annotations within one family. For this, functional families need to be 'pure', i.e., contain only proteins with identical function. Functional Families (FunFams) cluster proteins within CATH superfamilies into such groups of proteins sharing function. 11% of all FunFams (22 830 of 203 639) contain EC annotations and of those, 7% (1526 of 22 830) have inconsistent functional annotations. RESULTS: We propose an approach to further cluster FunFams into functionally more consistent sub-families by encoding their sequences through embeddings. These embeddings originate from language models transferring knowledge gained from predicting missing amino acids in a sequence (ProtBERT) and have been further optimized to distinguish between proteins belonging to the same or a different CATH superfamily (PB-Tucker). Using distances between embeddings and DBSCAN to cluster FunFams and identify outliers, doubled the number of pure clusters per FunFam compared to random clustering. Our approach was not limited to FunFams but also succeeded on families created using sequence similarity alone. Complementing EC annotations, we observed similar results for binding annotations. Thus, we expect an increased purity also for other aspects of function. Our results can help generating FunFams; the resulting clusters with improved functional consistency allow more reliable inference of annotations. We expect this approach to succeed equally for any other grouping of proteins by their phenotypes. AVAILABILITY AND IMPLEMENTATION: Code and embeddings are available via GitHub: https://github.com/Rostlab/FunFamsClustering. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

Biological impact of mutually exclusive exon switching.

Alternative splicing can expand the diversity of proteomes. Homologous mutually exclusive exons (MXEs) originate from the same ancestral exon and result in polypeptides with similar structural properties but altered sequence. Why would some genes switch homologous exons and what are their biological impact? Here, we analyse the extent of sequence, structural and functional variability in MXEs and report the first large scale, structure-based analysis of the biological impact of MXE events from different genomes. MXE-specific residues tend to map to single domains, are highly enriched in surface exposed residues and cluster at or near protein functional sites. Thus, MXE events are likely to maintain the protein fold, but alter specificity and selectivity of protein function. This comprehensive resource of MXE events and their annotations is available at: http://gene3d.biochem.ucl.ac.uk/mxemod/. These findings highlight how small, but significant changes at critical positions on a protein surface are exploited in evolution to alter function.

Genome3D: integrating a collaborative data pipeline to expand the depth and breadth of consensus protein structure annotation.

Genome3D (https://www.genome3d.eu) is a freely available resource that provides consensus structural annotations for representative protein sequences taken from a selection of model organisms. Since the last NAR update in 2015, the method of data submission has been overhauled, with annotations now being 'pushed' to the database via an API. As a result, contributing groups are now able to manage their own structural annotations, making the resource more flexible and maintainable. The new submission protocol brings a number of additional benefits including: providing instant validation of data and avoiding the requirement to synchronise releases between resources. It also makes it possible to implement the submission of these structural annotations as an automated part of existing internal workflows. In turn, these improvements facilitate Genome3D being opened up to new prediction algorithms and groups. For the latest release of Genome3D (v2.1), the underlying dataset of sequences used as prediction targets has been updated using the latest reference proteomes available in UniProtKB. A number of new reference proteomes have also been added of particular interest to the wider scientific community: cow, pig, wheat and mycobacterium tuberculosis. These additions, along with improvements to the underlying predictions from contributing resources, has ensured that the number of annotations in Genome3D has nearly doubled since the last NAR update article. The new API has also been used to facilitate the dissemination of Genome3D data into InterPro, thereby widening the visibility of both the annotation data and annotation algorithms.

CATH: expanding the horizons of structure-based functional annotations for genome sequences.

This article provides an update of the latest data and developments within the CATH protein structure classification database (http://www.cathdb.info). The resource provides two levels of release: CATH-B, a daily snapshot of the latest structural domain boundaries and superfamily assignments, and CATH+, which adds layers of derived data, such as predicted sequence domains, functional annotations and functional clustering (known as Functional Families or FunFams). The most recent CATH+ release (version 4.2) provides a huge update in the coverage of structural data. This release increases the number of fully- classified domains by over 40% (from 308 999 to 434 857 structural domains), corresponding to an almost two- fold increase in sequence data (from 53 million to over 95 million predicted domains) organised into 6119 superfamilies. The coverage of high-resolution, protein PDB chains that contain at least one assigned CATH domain is now 90.2% (increased from 82.3% in the previous release). A number of highly requested features have also been implemented in our web pages: allowing the user to view an alignment between their query sequence and a representative FunFam structure and providing tools that make it easier to view the full structural context (multi-domain architecture) of domains and chains.

InterPro in 2019: improving coverage, classification and access to protein sequence annotations.

The InterPro database (http://www.ebi.ac.uk/interpro/) classifies protein sequences into families and predicts the presence of functionally important domains and sites. Here, we report recent developments with InterPro (version 70.0) and its associated software, including an 18% growth in the size of the database in terms on new InterPro entries, updates to content, the inclusion of an additional entry type, refined modelling of discontinuous domains, and the development of a new programmatic interface and website. These developments extend and enrich the information provided by InterPro, and provide greater flexibility in terms of data access. We also show that InterPro's sequence coverage has kept pace with the growth of UniProtKB, and discuss how our evaluation of residue coverage may help guide future curation activities.

Gene3D: Extensive prediction of globular domains in proteins.

Gene3D (http://gene3d.biochem.ucl.ac.uk) is a database of globular domain annotations for millions of available protein sequences. Gene3D has previously featured in the Database issue of NAR and here we report a significant update to the Gene3D database. The current release, Gene3D v16, has significantly expanded its domain coverage over the previous version and now contains over 95 million domain assignments. We also report a new method for dealing with complex domain architectures that exist in Gene3D, arising from discontinuous domains. Amongst other updates, we have added visualization tools for exploring domain annotations in the context of other sequence features and in gene families. We also provide web-pages to visualize other domain families that co-occur with a given query domain family.

FunFam protein families improve residue level molecular function prediction.

BACKGROUND: The CATH database provides a hierarchical classification of protein domain structures including a sub-classification of superfamilies into functional families (FunFams). We analyzed the similarity of binding site annotations in these FunFams and incorporated FunFams into the prediction of protein binding residues. RESULTS: FunFam members agreed, on average, in 36.9 ± 0.6% of their binding residue annotations. This constituted a 6.7-fold increase over randomly grouped proteins and a 1.2-fold increase (1.1-fold on the same dataset) over proteins with the same enzymatic function (identical Enzyme Commission, EC, number). Mapping de novo binding residue prediction methods (BindPredict-CCS, BindPredict-CC) onto FunFam resulted in consensus predictions for those residues that were aligned and predicted alike (binding/non-binding) within a FunFam. This simple consensus increased the F1-score (for binding) 1.5-fold over the original prediction method. Variation of the threshold for how many proteins in the consensus prediction had to agree provided a convenient control of accuracy/precision and coverage/recall, e.g. reaching a precision as high as 60.8 ± 0.4% for a stringent threshold. CONCLUSIONS: The FunFams outperformed even the carefully curated EC numbers in terms of agreement of binding site residues. Additionally, we assume that our proof-of-principle through the prediction of protein binding residues will be relevant for many other solutions profiting from FunFams to infer functional information at the residue level.