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

Evaluating sequence and structural similarity metrics for predicting shared paralog functions

Gene duplication is the primary source of new genes, resulting in most genes having identifiable paralogs. Over time, paralog pairs may diverge in some respects but many retain the ability to perform the same functional role. Protein sequence identity is often used as a proxy for functional similarity and can predict shared functions between paralogs as revealed by synthetic lethal experiments. However, the advent of alternative protein representations, including embeddings from protein language models (PLMs) and predicted structures from AlphaFold, raises the possibility that alternative similarity metrics could better capture functional similarity between paralogs. Here, using two species (budding yeast and human) and two different definitions of shared functionality (shared protein-protein interactions and synthetic lethality), we evaluated a variety of alternative similarity metrics. For some tasks, predicted structural similarity or PLM similarity outperform sequence identity, but more importantly these similarity metrics are not redundant with sequence identity, i.e. combining them with sequence identity leads to improved predictions of shared functionality. By adding contextual features, representing similarity to homologous proteins within and across species, we can significantly enhance our predictions of shared paralog functionality. Overall, our results suggest that alternative similarity metrics capture complementary aspects of functional similarity beyond sequence identity alone.

Assessing Structural Classification Using AlphaFold2 Models Through ECOD-Based Comparative Analysis

Identifying homologous proteins is a fundamental task in structural bioinformatics. While AlphaFold2 has revolutionized protein structure prediction, the extent to which structure comparison of its models can reliably detect homologs remains unclear. In this study, we evaluate the feasibility of homology detection using AlphaFold2-predicted structures through structural comparisons. We considered the classification of the ECOD database for experimental structures as the correct standard and obtained their corresponding predicted models from AlphaFoldDB. To ensure blind assessment, we divided the structures into test and train sets according to their release date. Predicted and experimental 3D structures in the test and train sets were compared using 3D structure comparisons (MATRAS, Dali, and Foldseek) and sequence comparisons (BLAST and HHsearch). The results were evaluated based on the homology annotations in the ECOD database. For top-1 accuracy, the performance of structural comparisons was comparable to that of HHsearch. However, when considering metrics that included all structural pairs, including more remote homology, structural comparisons outperformed HHsearch. No significant differences were observed between comparisons of experimental versus experimental, predicted versus experimental, and predicted versus predicted structures with pLDDT (prediction confidence) values greater than 60. We also demonstrate that predicted protein structures, determined by NMR, had lower pLDDT values and contained fewer coils than their experimental counterparts. These findings highlight the potential of AlphaFold2 models in structural classification and suggest that 3D structural searches should be conducted not only against the PDB but also against AlphaFoldDB to identify more potential homologs.

Unraveling the diversity of protein-carbohydrate interfaces: Insights from a multi-scale study

Protein-carbohydrate interactions play a crucial role in numerous fundamental biological processes. Thus, description and comparison of the carbohydrate binding site (CBS) architecture is of great importance for understanding of the underlying biological mechanisms. However, traditional approaches for carbohydrate-binding protein analysis and annotation rely primarily on the sequence-based methods applied to specific protein classes. The recently released DIONYSUS database aims to fill this gap by providing tools for CBS comparison at different levels: both in terms of protein properties and classification, as well as in terms of atomistic CBS organization. In the current study, we explore DIONYSUS content using a combination of the suggested approaches in order to evaluate the diversity of the currently resolved non-covalent protein-carbohydrate interfaces at different scales. Notably, our analysis reveals evolutionary convergence of CBS in proteins with distinct folds and coming from organisms across different kingdoms of life. Furthermore, we demonstrate that a CBS structure based approach has the potential to facilitate functional annotation for the proteins with missing information in the existing databases. In particular, it provides reliable information for numerous carbohydrate-binding proteins from rapidly evolving organisms, whose analysis is particularly challenging for classical sequence-based methods.

Are protein language models the new universal key?

Protein language models (pLMs) capture some aspects of the grammar of the language of life as written in protein sequences. The so-called pLM embeddings implicitly contain this information. Therefore, embeddings can serve as the exclusive input into downstream supervised methods for protein prediction. Over the last 33 years, evolutionary information extracted through simple averaging for specific protein families from multiple sequence alignments (MSAs) has been the most successful universal key to the success of protein prediction. For many applications, MSA-free pLM-based predictions now have become significantly more accurate. The reason for this is often a combination of two aspects. Firstly, embeddings condense the grammar so efficiently that downstream prediction methods succeed with small models, i.e., they need few free parameters in particular in the era of exploding deep neural networks. Secondly, pLM-based methods provide protein-specific solutions. As additional benefit, once the pLM pre-training is complete, pLM-based solutions tend to consume much fewer resources than MSA-based solutions. In fact, we appeal to the community to rather optimize foundation models than to retrain new ones and to evolve incentives for solutions that require fewer resources even at some loss in accuracy. Although pLMs have not, yet, succeeded to entirely replace the body of solutions developed over three decades, they clearly are rapidly advancing as the universal key for protein prediction.

Molecular dynamics simulation of wild and mutant proteasome subunit beta type 8 (PSMB8) protein: Implications for restoration of inflammation in experimental autoimmune encephalomyelitis pathogenesis

Multiple Sclerosis (MS) is an autoimmune and chronic disease in the brain and spinal cord. MS has inflammatory progression characterized by its hallmark inflammatory plaques. The histological and clinical characteristics of MS are shared by Experimental Autoimmune Encephalomyelitis (EAE). Genetic and environmental factors contribute to the development of MS. In EAE-MS disease, the level of proteasome subunit beta type-8 (PSMB8), encoded by the PSMB8 gene, is increased and regulates the inflammatory response in this disease. In humans, the Nakajo-Nishimura Syndrome is caused by a mutation in the gene PSMB8, a part of the immunoproteasome subunit. Therefore, special attention to wild and mutant (G210V) PSMB8 protein is imperative. In this study, we performed a 100 ns molecular dynamics (MD) simulation for wild-type PSMB8 and the mutant G210V. Then, we analyzed the fundamental and essential simulation results using another Google Colab system. The energy analysis ensures the structural deviation due to point mutation. The trajectory of the fundamental simulation (RMSD, RMSF, and Rg) describes that the G210V mutated protein is more flexible and less stable than the wild type. We observed the conformational changes due to mutation by analyzing the RMSD average linkage hierarchical clustering, total SASA, and SASA autocorrelation. The differences in the protein's overall motion and the atoms' precise location are identified by the principal component analysis, showing that the overall motion and location of the atoms are different. Our study provides valuable insights into the dynamics and structure of this protein, which can aid in further understanding its biological functions and potential implications for disease.

CATH v4.4: major expansion of CATH by experimental and predicted structural data

CATH (https://www.cathdb.info) is a structural classification database that assigns domains to the structures in the Protein Data Bank (PDB) and AlphaFold Protein Structure Database (AFDB) and adds layers of biological information, including homology and functional annotation. This article covers developments in the CATH classification since 2021. We report the significant expansion of structural information (180-fold) for CATH superfamilies through classification of PDB domains and predicted domain structures from the Encyclopedia of Domains (TED) resource. TED provides information on predicted domains in AFDB. CATH v4.4 represents an expansion of ∼64 844 experimentally determined domain structures from PDB. We also present a mapping of ∼90 million predicted domains from TED to CATH superfamilies. New PDB and TED data increases the number of superfamilies from 5841 to 6573, folds from 1349 to 2078 and architectures from 41 to 77. TED data comprises predicted structures, so these new folds and architectures remain hypothetical until experimentally confirmed. CATH also classifies domains into functional families (FunFams) within a superfamily. We have updated sequences in FunFams by scanning FunFam-HMMs against UniProt release 2024_02, giving a 276% increase in FunFams coverage. The mapping of TED structural domains has resulted in a 4-fold increase in FunFams with structural information.

Bilingual language model for protein sequence and structure

Adapting language models to protein sequences spawned the development of powerful protein language models (pLMs). Concurrently, AlphaFold2 broke through in protein structure prediction. Now we can systematically and comprehensively explore the dual nature of proteins that act and exist as three-dimensional (3D) machines and evolve as linear strings of one-dimensional (1D) sequences. Here, we leverage pLMs to simultaneously model both modalities in a single model. We encode protein structures as token sequences using the 3Di-alphabet introduced by the 3D-alignment method Foldseek. For training, we built a non-redundant dataset from AlphaFoldDB and fine-tuned an existing pLM (ProtT5) to translate between 3Di and amino acid sequences. As a proof-of-concept for our novel approach, dubbed Protein 'structure-sequence' T5 (ProstT5), we showed improved performance for subsequent, structure-related prediction tasks, leading to three orders of magnitude speedup for deriving 3Di. This will be crucial for future applications trying to search metagenomic sequence databases at the sensitivity of structure comparisons. Our work showcased the potential of pLMs to tap into the information-rich protein structure revolution fueled by AlphaFold2. ProstT5 paves the way to develop new tools integrating the vast resource of 3D predictions and opens new research avenues in the post-AlphaFold2 era.

Semantic search using protein large language models detects class II microcins in bacterial genomes

Class II microcins are antimicrobial peptides that have shown some potential as novel antibiotics. However, to date, only 10 class II microcins have been described, and the discovery of novel microcins has been hampered by their short length and high sequence divergence. Here, we ask if we can use numerical embeddings generated by protein large language models to detect microcins in bacterial genome assemblies and whether this method can outperform sequence-based methods such as BLAST. We find that embeddings detect known class II microcins much more reliably than does BLAST and that any two microcins tend to have a small distance in embedding space even though they typically are highly diverged at the sequence level. In data sets of Escherichia coli, Klebsiella spp., and Enterobacter spp. genomes, we further find novel putative microcins that were previously missed by sequence-based search methods.

IMPORTANCE

Antibiotic resistance is becoming an increasingly serious problem in modern medicine, but the development pipeline for conventional antibiotics is not promising. Therefore, alternative approaches to combat bacterial infections are urgently needed. One such approach may be to employ naturally occurring antibacterial peptides produced by bacteria to kill competing bacteria. A promising class of such peptides are class II microcins. However, only a small number of class II microcins have been discovered to date, and the discovery of further such microcins has been hampered by their high sequence divergence and short length, which can cause sequence-based search methods to fail. Here, we demonstrate that a more robust method for microcin discovery can be built on the basis of a protein large language model, and we use this method to identify several putative novel class II microcins.

Assessing the role of evolutionary information for enhancing protein language model embeddings

Embeddings from protein Language Models (pLMs) are replacing evolutionary information from multiple sequence alignments (MSAs) as the most successful input for protein prediction. Is this because embeddings capture evolutionary information? We tested various approaches to explicitly incorporate evolutionary information into embeddings on various protein prediction tasks. While older pLMs (SeqVec, ProtBert) significantly improved through MSAs, the more recent pLM ProtT5 did not benefit. For most tasks, pLM-based outperformed MSA-based methods, and the combination of both even decreased performance for some (intrinsic disorder). We highlight the effectiveness of pLM-based methods and find limited benefits from integrating MSAs.

Artificial Intelligence Learns Protein Prediction

From AlphaGO over StableDiffusion to ChatGPT, the recent decade of exponential advances in artificial intelligence (AI) has been altering life. In parallel, advances in computational biology are beginning to decode the language of life: AlphaFold2 leaped forward in protein structure prediction, and protein language models (pLMs) replaced expertise and evolutionary information from multiple sequence alignments with information learned from reoccurring patterns in databases of billions of proteins without experimental annotations other than the amino acid sequences. None of those tools could have been developed 10 years ago; all will increase the wealth of experimental data and speed up the cycle from idea to proof. AI is affecting molecular and medical biology at giant steps, and the most important might be the leap toward more powerful protein design.

Clustering protein functional families at large scale with hierarchical approaches

Proteins, fundamental to cellular activities, reveal their function and evolution through their structure and sequence. CATH functional families (FunFams) are coherent clusters of protein domain sequences in which the function is conserved across their members. The increasing volume and complexity of protein data enabled by large-scale repositories like MGnify or AlphaFold Database requires more powerful approaches that can scale to the size of these new resources. In this work, we introduce MARC and FRAN, two algorithms developed to build upon and address limitations of GeMMA/FunFHMMER, our original methods developed to classify proteins with related functions using a hierarchical approach. We also present CATH-eMMA, which uses embeddings or Foldseek distances to form relationship trees from distance matrices, reducing computational demands and handling various data types effectively. CATH-eMMA offers a highly robust and much faster tool for clustering protein functions on a large scale, providing a new tool for future studies in protein function and evolution.

CATH 2024: CATH-AlphaFlow Doubles the Number of Structures in CATH and Reveals Nearly 200 New Folds

CATH (https://www.cathdb.info) classifies domain structures from experimental protein structures in the PDB and predicted structures in the AlphaFold Database (AFDB). To cope with the scale of the predicted data a new NextFlow workflow (CATH-AlphaFlow), has been developed to classify high-quality domains into CATH superfamilies and identify novel fold groups and superfamilies. CATH-AlphaFlow uses a novel state-of-the-art structure-based domain boundary prediction method (ChainSaw) for identifying domains in multi-domain proteins. We applied CATH-AlphaFlow to process PDB structures not classified in CATH and AFDB structures from 21 model organisms, expanding CATH by over 100%. Domains not classified in existing CATH superfamilies or fold groups were used to seed novel folds, giving 253 new folds from PDB structures (September 2023 release) and 96 from AFDB structures of proteomes of 21 model organisms. Where possible, functional annotations were obtained using (i) predictions from publicly available methods (ii) annotations from structural relatives in AFDB/UniProt50. We also predicted functional sites and highly conserved residues. Some folds are associated with important functions such as photosynthetic acclimation (in flowering plants), iron permease activity (in fungi) and post-natal spermatogenesis (in mice). CATH-AlphaFlow will allow us to identify many more CATH relatives in the AFDB, further characterising the protein structure landscape.

PLMSearch: Protein language model powers accurate and fast sequence search for remote homology

Homologous protein search is one of the most commonly used methods for protein annotation and analysis. Compared to structure search, detecting distant evolutionary relationships from sequences alone remains challenging. Here we propose PLMSearch (Protein Language Model), a homologous protein search method with only sequences as input. PLMSearch uses deep representations from a pre-trained protein language model and trains the similarity prediction model with a large number of real structure similarity. This enables PLMSearch to capture the remote homology information concealed behind the sequences. Extensive experimental results show that PLMSearch can search millions of query-target protein pairs in seconds like MMseqs2 while increasing the sensitivity by more than threefold, and is comparable to state-of-the-art structure search methods. In particular, unlike traditional sequence search methods, PLMSearch can recall most remote homology pairs with dissimilar sequences but similar structures. PLMSearch is freely available at https://dmiip.sjtu.edu.cn/PLMSearch .

Machine learning-based prediction of proteins' architecture using sequences of amino acids and structural alphabets

In addition to the growth of protein structures generated through wet laboratory experiments and deposited in the PDB repository, AlphaFold predictions have significantly contributed to the creation of a much larger database of protein structures. Annotating such a vast number of structures has become an increasingly challenging task. CATH is widely recognized as one the most common platforms for addressing this challenge, as it classifies proteins based on their structural and evolutionary relationships, offering the scientific community an invaluable resource for uncovering various properties, including functional annotations. While CATH annotation involves - to some extent - human intervention, keeping up with the classification of the rapidly expanding repositories of protein structures has become exceedingly difficult. Therefore, there is a pressing need for a fully automated approach. On the other hand, the abundance of protein sequences stemming from next generation sequencing technologies, lacking structural annotations, presents an additional challenge to the scientific community. Consequently, 'pre-annotating' protein sequences with structural features, ensuring a high level of precision, could prove highly advantageous. In this paper, after a thorough investigation, we introduce a novel machine-learning model capable of classifying any protein domain, whether it has a known structure or not, into one of the 40 main CATH Architectures. We achieve an F1 Score of 0.92 using only the amino acid sequence and a score of 0.94 using both the sequence of amino acids and the sequence of structural alphabets.

SumoPred-PLM: human SUMOylation and SUMO2/3 sites Prediction using Pre-trained Protein Language Model

SUMOylation is an essential post-translational modification system with the ability to regulate nearly all aspects of cellular physiology. Three major paralogues SUMO1, SUMO2 and SUMO3 form a covalent bond between the small ubiquitin-like modifier with lysine residues at consensus sites in protein substrates. Biochemical studies continue to identify unique biological functions for protein targets conjugated to SUMO1 versus the highly homologous SUMO2 and SUMO3 paralogues. Yet, the field has failed to harness contemporary AI approaches including pre-trained protein language models to fully expand and/or recognize the SUMOylated proteome. Herein, we present a novel, deep learning-based approach called SumoPred-PLM for human SUMOylation prediction with sensitivity, specificity, Matthew's correlation coefficient, and accuracy of 74.64%, 73.36%, 0.48% and 74.00%, respectively, on the CPLM 4.0 independent test dataset. In addition, this novel platform uses contextualized embeddings obtained from a pre-trained protein language model, ProtT5-XL-UniRef50 to identify SUMO2/3-specific conjugation sites. The results demonstrate that SumoPred-PLM is a powerful and unique computational tool to predict SUMOylation sites in proteins and accelerate discovery.

Integrating Embeddings from Multiple Protein Language Models to Improve Protein O-GlcNAc Site Prediction

O-linked β-N-acetylglucosamine (O-GlcNAc) is a distinct monosaccharide modification of serine (S) or threonine (T) residues of nucleocytoplasmic and mitochondrial proteins. O-GlcNAc modification (i.e., O-GlcNAcylation) is involved in the regulation of diverse cellular processes, including transcription, epigenetic modifications, and cell signaling. Despite the great progress in experimentally mapping O-GlcNAc sites, there is an unmet need to develop robust prediction tools that can effectively locate the presence of O-GlcNAc sites in protein sequences of interest. In this work, we performed a comprehensive evaluation of a framework for prediction of protein O-GlcNAc sites using embeddings from pre-trained protein language models. In particular, we compared the performance of three protein sequence-based large protein language models (pLMs), Ankh, ESM-2, and ProtT5, for prediction of O-GlcNAc sites and also evaluated various ensemble strategies to integrate embeddings from these protein language models. Upon investigation, the decision-level fusion approach that integrates the decisions of the three embedding models, which we call LM-OGlcNAc-Site, outperformed the models trained on these individual language models as well as other fusion approaches and other existing predictors in almost all of the parameters evaluated. The precise prediction of O-GlcNAc sites will facilitate the probing of O-GlcNAc site-specific functions of proteins in physiology and diseases. Moreover, these findings also indicate the effectiveness of combined uses of multiple protein language models in post-translational modification prediction and open exciting avenues for further research and exploration in other protein downstream tasks. LM-OGlcNAc-Site's web server and source code are publicly available to the community.

Machine Learning-Guided Protein Engineering

Recent progress in engineering highly promising biocatalysts has increasingly involved machine learning methods. These methods leverage existing experimental and simulation data to aid in the discovery and annotation of promising enzymes, as well as in suggesting beneficial mutations for improving known targets. The field of machine learning for protein engineering is gathering steam, driven by recent success stories and notable progress in other areas. It already encompasses ambitious tasks such as understanding and predicting protein structure and function, catalytic efficiency, enantioselectivity, protein dynamics, stability, solubility, aggregation, and more. Nonetheless, the field is still evolving, with many challenges to overcome and questions to address. In this Perspective, we provide an overview of ongoing trends in this domain, highlight recent case studies, and examine the current limitations of machine learning-based methods. We emphasize the crucial importance of thorough experimental validation of emerging models before their use for rational protein design. We present our opinions on the fundamental problems and outline the potential directions for future research.

AlphaFold2 reveals commonalities and novelties in protein structure space for 21 model organisms

Deep-learning (DL) methods like DeepMind's AlphaFold2 (AF2) have led to substantial improvements in protein structure prediction. We analyse confident AF2 models from 21 model organisms using a new classification protocol (CATH-Assign) which exploits novel DL methods for structural comparison and classification. Of ~370,000 confident models, 92% can be assigned to 3253 superfamilies in our CATH domain superfamily classification. The remaining cluster into 2367 putative novel superfamilies. Detailed manual analysis on 618 of these, having at least one human relative, reveal extremely remote homologies and further unusual features. Only 25 novel superfamilies could be confirmed. Although most models map to existing superfamilies, AF2 domains expand CATH by 67% and increases the number of unique 'global' folds by 36% and will provide valuable insights on structure function relationships. CATH-Assign will harness the huge expansion in structural data provided by DeepMind to rationalise evolutionary changes driving functional divergence.