Protein homology detection by HMM-HMM comparison

Redefining Chlorobotryaceae as one of the principal and most diverse lineages of eustigmatophyte algae

Eustigmatophyceae is one of the ∼17 classes of the vast algal phylum Ochrophyta. Over the last decade, the eustigmatophytes emerged as an expansive group that has grown from the initially recognized handful of species to well over 200 genetically distinct entities (potential species). Yet the majority of eustigs, remain represented by unidentified strains, or even only metabarcode sequences obtained from environmental samples. Moreover, the formal classification of the group has not yet been harmonized with the recently uncovered diversity and phylogenetic relationships within the class. Here we make a major step towards resolving this issue by addressing the diversity, phylogeny and classification of one of the most prominent eustigmatophyte clades previously informally called the "Eustigmataceae group". We obtained 18S rDNA and rbcL gene sequences from four new strains from the "Eustigmataceae group", and from several additional eustig strains, and performed the most comprehensive phylogenetic analyses of Eustigmatophyceae to date. Our results of these analyses confirm the monophyly of the "Eustigmataceae group" and define its major subclades. We also sequenced plastid genomes of five "Eustigmataceae group" strains to not only improve our understanding of the plastid gene content evolution in eustigs, but also to obtain a robustly resolved eustigmatophyte phylogeny. With this new genomic data, we have solidified the view of the "Eustigmataceae group" as a well-defined family level clade. Crucially, we also have firmly established the genus Chlorobotrys as a member of the "Eustigmataceae group". This new molecular evidence, together with a critical analysis of the literature going back to the 19th century, provided the basis to radically redefine the historical concept of the family Chlorobotryaceae as the formal taxonomic rubric corresponding to the "Eustigmataceae group". With this change, the family names Eustigmataceae and Characiopsidaceae are reduced to synonymy with the Chlorobotryaceae, with the latter having taxonomic priority. We additionally studied in detail the morphology and ultrastructure of two Chlorobotryaceae members, which we describe as Neustupella aerophytica gen. et sp. nov. and Lietzensia polymorpha gen. et sp. nov. Finally, our analyses of partial genomic data from several Chlorobotryaceae representatives identified genes for hallmark flagellar proteins in all of these strains. The presence of the flagellar proteins strongly suggests that zoosporogenesis is a common trait of the family and also occurs in the members never observed to produce flagellated stages. Altogether, our work paints a rich picture of one of the most diverse principal lineages of eustigmatophyte algae.

Opportunities and Challenges of Data-Driven Virus Discovery

Virus discovery has been fueled by new technologies ever since the first viruses were discovered at the end of the 19th century. Starting with mechanical devices that provided evidence for virus presence in sick hosts, virus discovery gradually transitioned into a sequence-based scientific discipline, which, nowadays, can characterize virus identity and explore viral diversity at an unprecedented resolution and depth. Sequencing technologies are now being used routinely and at ever-increasing scales, producing an avalanche of novel viral sequences found in a multitude of organisms and environments. In this perspective article, we argue that virus discovery has started to undergo another transformation prompted by the emergence of new approaches that are sequence data-centered and primarily computational, setting them apart from previous technology-driven innovations. The data-driven virus discovery approach is largely uncoupled from the collection and processing of biological samples, and exploits the availability of massive amounts of publicly and freely accessible data from sequencing archives. We discuss open challenges to be solved in order to unlock the full potential of data-driven virus discovery, and we highlight the benefits it can bring to classical (mostly molecular) virology and molecular biology in general.

Analyzing structural features of proteins from deep-sea organisms

Protein adaptations to extreme environmental conditions are drivers in biotechnological process optimization and essential to unravel the molecular limits of life. Most proteins with such desirable adaptations are found in extremophilic organisms inhabiting extreme environments. The deep sea is such an environment and a promising resource that poses multiple extremes on its inhabitants. Conditions like high hydrostatic pressure and high or low temperature are prevalent and many deep-sea organisms tolerate multiple of these extremes. While molecular adaptations to high temperature are comparatively good described, adaptations to other extremes like high pressure are not well-understood yet. To fully unravel the molecular mechanisms of individual adaptations it is probably necessary to disentangle multifactorial adaptations. In this study, we evaluate differences of protein structures from deep-sea organisms and their respective related proteins from nondeep-sea organisms. We created a data collection of 1281 experimental protein structures from 25 deep-sea organisms and paired them with orthologous proteins. We exhaustively evaluate differences between the protein pairs with machine learning and Shapley values to determine characteristic differences in sequence and structure. The results show a reasonable discrimination of deep-sea and nondeep-sea proteins from which we distinguish correlations previously attributed to thermal stability from other signals potentially describing adaptions to high pressure. While some distinct correlations can be observed the overall picture appears intricate.

Multi-color RNA imaging with CRISPR-Cas13b systems in living cells

Visualizing RNA dynamics is important for understanding RNA function. Catalytically dead (d) CRISPR-Cas13 systems have been established to image and track RNAs in living cells, but efficient dCas13 for RNA imaging is still limited. Here, we analyzed metagenomic and bacterial genomic databases to comprehensively screen Cas13 homologies for their RNA labeling capabilities in living mammalian cells. Among eight previously unreported dCas13 proteins that can be used for RNA labeling, dHgm4Cas13b and dMisCas13b displayed comparable, if not higher, efficiencies to the best-known ones when targeting endogenous MUC4 and NEAT1_2 by single guide (g) RNAs. Further examination of the labeling robustness of different dCas13 systems using the GCN4 repeats revealed that a minimum of 12 GCN4 repeats was required for dHgm4Cas13b and dMisCas13b imaging at the single RNA molecule level, while >24 GCN4 repeats were required for reported dLwaCas13a, dRfxCas13d and dPguCas13b. Importantly, by silencing pre-crRNA processing activity of dMisCas13b (ddMisCas13b) and further incorporating RNA aptamers including PP7, MS2, Pepper or BoxB to individual gRNAs, a CRISPRpalette system was developed to successfully achieve multi-color RNA visualization in living cells.

ProtGPT2 is a deep unsupervised language model for protein design

Protein design aims to build novel proteins customized for specific purposes, thereby holding the potential to tackle many environmental and biomedical problems. Recent progress in Transformer-based architectures has enabled the implementation of language models capable of generating text with human-like capabilities. Here, motivated by this success, we describe ProtGPT2, a language model trained on the protein space that generates de novo protein sequences following the principles of natural ones. The generated proteins display natural amino acid propensities, while disorder predictions indicate that 88% of ProtGPT2-generated proteins are globular, in line with natural sequences. Sensitive sequence searches in protein databases show that ProtGPT2 sequences are distantly related to natural ones, and similarity networks further demonstrate that ProtGPT2 is sampling unexplored regions of protein space. AlphaFold prediction of ProtGPT2-sequences yields well-folded non-idealized structures with embodiments and large loops and reveals topologies not captured in current structure databases. ProtGPT2 generates sequences in a matter of seconds and is freely available.

Methodology-Centered Review of Molecular Modeling, Simulation, and Prediction of SARS-CoV-2

Despite tremendous efforts in the past two years, our understanding of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), virus-host interactions, immune response, virulence, transmission, and evolution is still very limited. This limitation calls for further in-depth investigation. Computational studies have become an indispensable component in combating coronavirus disease 2019 (COVID-19) due to their low cost, their efficiency, and the fact that they are free from safety and ethical constraints. Additionally, the mechanism that governs the global evolution and transmission of SARS-CoV-2 cannot be revealed from individual experiments and was discovered by integrating genotyping of massive viral sequences, biophysical modeling of protein-protein interactions, deep mutational data, deep learning, and advanced mathematics. There exists a tsunami of literature on the molecular modeling, simulations, and predictions of SARS-CoV-2 and related developments of drugs, vaccines, antibodies, and diagnostics. To provide readers with a quick update about this literature, we present a comprehensive and systematic methodology-centered review. Aspects such as molecular biophysics, bioinformatics, cheminformatics, machine learning, and mathematics are discussed. This review will be beneficial to researchers who are looking for ways to contribute to SARS-CoV-2 studies and those who are interested in the status of the field.

Bacteria deplete deoxynucleotides to defend against bacteriophage infection

DNA viruses and retroviruses consume large quantities of deoxynucleotides (dNTPs) when replicating. The human antiviral factor SAMHD1 takes advantage of this vulnerability in the viral lifecycle, and inhibits viral replication by degrading dNTPs into their constituent deoxynucleosides and inorganic phosphate. Here, we report that bacteria use a similar strategy to defend against bacteriophage infection. We identify a family of defensive bacterial deoxycytidine triphosphate (dCTP) deaminase proteins that convert dCTP into deoxyuracil nucleotides in response to phage infection. We also identify a family of phage resistance genes that encode deoxyguanosine triphosphatase (dGTPase) enzymes, which degrade dGTP into phosphate-free deoxyguanosine and are distant homologues of human SAMHD1. Our results suggest that bacterial defensive proteins deplete specific deoxynucleotides (either dCTP or dGTP) from the nucleotide pool during phage infection, thus starving the phage of an essential DNA building block and halting its replication. Our study shows that manipulation of the dNTP pool is a potent antiviral strategy shared by both prokaryotes and eukaryotes.

Discovery of archaeal fusexins homologous to eukaryotic HAP2/GCS1 gamete fusion proteins

Sexual reproduction consists of genome reduction by meiosis and subsequent gamete fusion. The presence of genes homologous to eukaryotic meiotic genes in archaea and bacteria suggests that DNA repair mechanisms evolved towards meiotic recombination. However, fusogenic proteins resembling those found in gamete fusion in eukaryotes have so far not been found in prokaryotes. Here, we identify archaeal proteins that are homologs of fusexins, a superfamily of fusogens that mediate eukaryotic gamete and somatic cell fusion, as well as virus entry. The crystal structure of a trimeric archaeal fusexin (Fusexin1 or Fsx1) reveals an archetypical fusexin architecture with unique features such as a six-helix bundle and an additional globular domain. Ectopically expressed Fusexin1 can fuse mammalian cells, and this process involves the additional globular domain and a conserved fusion loop. Furthermore, archaeal fusexin genes are found within integrated mobile elements, suggesting potential roles in cell-cell fusion and gene exchange in archaea, as well as different scenarios for the evolutionary history of fusexins.

Conserved and lineage-specific hypothetical proteins may have played a central role in the rise and diversification of major archaeal groups

BACKGROUND

Archaea play fundamental roles in the environment, for example by methane production and consumption, ammonia oxidation, protein degradation, carbon compound turnover, and sulfur compound transformations. Recent genomic analyses have profoundly reshaped our understanding of the distribution and functionalities of Archaea and their roles in eukaryotic evolution.

RESULTS

Here, 1179 representative genomes were selected from 3197 archaeal genomes. The representative genomes clustered based on the content of 10,866 newly defined archaeal protein families (that will serve as a community resource) recapitulates archaeal phylogeny. We identified the co-occurring proteins that distinguish the major lineages. Those with metabolic roles were consistent with experimental data. However, two families specific to Asgard were determined to be new eukaryotic signature proteins. Overall, the blocks of lineage-specific families are dominated by proteins that lack functional predictions.

CONCLUSIONS

Given that these hypothetical proteins are near ubiquitous within major archaeal groups, we propose that they were important in the origin of most of the major archaeal lineages. Interestingly, although there were clearly phylum-specific co-occurring proteins, no such blocks of protein families were shared across superphyla, suggesting a burst-like origin of new lineages early in archaeal evolution.

Viral diversity and biogeochemical potential revealed in different prawn-culture sediments by virus-enriched metagenome analysis

As the most numerous biological entities on Earth, viruses affect the microbial dynamics, metabolism and biogeochemical cycles in the aquatic ecosystems. Viral diversity and functions in ocean have been relatively well studied, but our understanding of viruses in mariculture systems is limited. To fill this knowledge gap, we studied viral diversity and potential biogeochemical impacts of sediments from four different prawn-mariculture ecosystems (mono-culture of prawn and poly-culture of prawn with jellyfish, sea cucumber, and clam) using a metagenomic approach with prior virus-like particles (VLPs) separation. We found that the order Caudovirales was the predominant viral category and accounted for the most volume (78.39% of classified viruses). Sediment viruses were verified to have a high diversity by using the construct phylogenetic tree of terL gene, with three potential novel clades being identified. Meanwhile, compared with viruses inhabiting other ecosystems based on gene-sharing network, our results revealed that mariculture sediments harbored considerable unexplored viral diversity and that maricultural species were potentially important drivers of the viral community structure. Notably, viral auxiliary metabolic genes were identified and suggested that viruses influence carbon and sulfur cycling, as well as cofactors/vitamins and amino acid metabolism, which indirectly participate in biogeochemical cycling. Overall, our findings revealed the genomic diversity and ecological function of viral communities in prawn mariculture sediments, and suggested the role of viruses in microbial ecology and biogeochemistry.

S51 Family Peptidases Provide Resistance to Peptidyl-Nucleotide Antibiotic McC

Microcin C (McC)-like compounds are natural Trojan horse peptide-nucleotide antibiotics produced by diverse bacteria. The ribosomally synthesized peptide parts of these antibiotics are responsible for their facilitated transport into susceptible cells. Once inside the cell, the peptide part is degraded, releasing the toxic payload, an isoaspartyl-nucleotide that inhibits aspartyl-tRNA synthetase, an enzyme essential for protein synthesis. Bacteria that produce microcin C-like compounds have evolved multiple ways to avoid self-intoxication. Here, we describe a new strategy through the action of S51 family peptidases, which we name MccG. MccG cleaves the toxic isoaspartyl-nucleotide, rendering it inactive. While some MccG homologs are encoded by gene clusters responsible for biosynthesis of McC-like compounds, most are encoded by standalone genes whose products may provide a basal level of resistance to peptide-nucleotide antibiotics in phylogenetically distant bacteria.

New β-Propellers Are Continuously Amplified From Single Blades in all Major Lineages of the β-Propeller Superfamily

β-Propellers are toroidal folds, in which consecutive supersecondary structure units of four anti-parallel β-strands-called blades-are arranged radially around a central axis. Uniquely among toroidal folds, blades span the full range of sequence symmetry, from near identity to complete divergence, indicating an ongoing process of amplification and differentiation. We have proposed that the major lineages of β-propellers arose through this mechanism and that therefore their last common ancestor was a single blade, not a fully formed β-propeller. Here we show that this process of amplification and differentiation is also widespread within individual lineages, yielding β-propellers with blades of more than 60% pairwise sequence identity in most major β-propeller families. In some cases, the blades are nearly identical, indicating a very recent amplification event, but even in cases where such recently amplified β-propellers have more than 80% overall sequence identity to each other, comparison of their DNA sequence shows that the amplification occurred independently.

Point (S-to-G) Mutations in the W(S/G)GE Motif in Red/Green Cyanobacteriochrome GAF Domains Enhance Thermal Reversion Rates

Cyanobacteriochromes (CBCRs) are photoreceptors consisting of single or tandem GAF (cGMP-phosphodiesterase/adenylate cyclase/FhlA) domains that bind bilin chromophores. Canonical red/green CBCR GAF domains are a well-characterized subgroup of the expanded red/green CBCR GAF domain family that binds phycocyanobilin (PCB) and converts between a thermally stable red-absorbing Pr state and a green-absorbing Pg state. The rate of thermal reversion from Pg to Pr varies widely among canonical red/green CBCR GAF domains, with half-lives ranging from days to seconds. Since the thermal reversion rate is an important parameter for the application of CBCR GAF domains as optogenetic tools, the molecular factors controlling the thermal reversion rate are of particular interest. Here, we report that point mutations in a well-conserved W(S/G)GE motif alter reversion rates in canonical red/green CBCR GAF domains in a predictable manner. Specifically, S-to-G mutations enhance thermal reversion rates, while the reverse, G-to-S mutations slow thermal reversion. Despite the distance (>10 Å) of the mutation site from the chromophore, molecular dynamics simulations and nuclear magnetic resonance (NMR) analyses suggest that the presence of a glycine residue allows the formation of a water bridge that alters the conformational dynamics of chromophore-interacting residues, leading to enhanced Pg to Pr thermal reversion.

AI-Based Protein Structure Prediction in Drug Discovery: Impacts and Challenges

Proteins are the molecular machinery of the human body, and their malfunctioning is often responsible for diseases, making them crucial targets for drug discovery. The three-dimensional structure of a protein determines its biological function, its conformational state determines substrates, cofactors, and protein binding. Rational drug discovery employs engineered small molecules to selectively interact with proteins to modulate their function. To selectively target a protein and to design small molecules, knowing the protein structure with all its specific conformation is critical. Unfortunately, for a large number of proteins relevant for drug discovery, the three-dimensional structure has not yet been experimentally solved. Therefore, accurately predicting their structure based on their amino acid sequence is one of the grant challenges in biology. Recently, AlphaFold2, a machine learning application based on a deep neural network, was able to predict unknown structures of proteins with an unprecedented accuracy. Despite the impressive progress made by AlphaFold2, nature still challenges the field of structure prediction. In this Perspective, we explore how AlphaFold2 and related methods help make drug design more efficient. Furthermore, we discuss the roles of predicting domain-domain orientations, all relevant conformational states, the influence of posttranslational modifications, and conformational changes due to protein binding partners. We highlight where further improvements are needed for advanced machine learning methods to be successfully and frequently used in the pharmaceutical industry.

Encapsulins

Subcellular compartmentalization is a defining feature of all cells. In prokaryotes, compartmentalization is generally achieved via protein-based strategies. The two main classes of microbial protein compartments are bacterial microcompartments and encapsulin nanocompartments. Encapsulins self-assemble into proteinaceous shells with diameters between 24 and 42 nm and are defined by the viral HK97-fold of their shell protein. Encapsulins have the ability to encapsulate dedicated cargo proteins, including ferritin-like proteins, peroxidases, and desulfurases. Encapsulation is mediated by targeting sequences present in all cargo proteins. Encapsulins are found in many bacterial and archaeal phyla and have been suggested to play roles in iron storage, stress resistance, sulfur metabolism, and natural product biosynthesis. Phylogenetic analyses indicate that they share a common ancestor with viral capsid proteins. Many pathogens encode encapsulins, and recent evidence suggests that they may contribute toward pathogenicity. The existing information on encapsulin structure, biochemistry, biological function, and biomedical relevance is reviewed here.

The OceanDNA MAG catalog contains over 50,000 prokaryotic genomes originated from various marine environments

Marine microorganisms are immensely diverse and play fundamental roles in global geochemical cycling. Recent metagenome-assembled genome studies, with particular attention to large-scale projects such as Tara Oceans, have expanded the genomic repertoire of marine microorganisms. However, published marine metagenome data is still underexplored. We collected 2,057 marine metagenomes covering various marine environments and developed a new genome reconstruction pipeline. We reconstructed 52,325 qualified genomes composed of 8,466 prokaryotic species-level clusters spanning 59 phyla, including genomes from the deep-sea characterized as deeper than 1,000 m (n = 3,337), low-oxygen zones of <90 μmol O2 per kg water (n = 7,884), and polar regions (n = 7,752). Novelty evaluation using a genome taxonomy database shows that 6,256 species (73.9%) are novel and include genomes of high taxonomic novelty, such as new class candidates. These genomes collectively expanded the known phylogenetic diversity of marine prokaryotes by 34.2%, and the species representatives cover 26.5-42.0% of prokaryote-enriched metagenomes. Thoroughly leveraging accumulated metagenomic data, this genome resource, named the OceanDNA MAG catalog, illuminates uncharacterized marine microbial 'dark matter' lineages.

Transcriptional Basis for Haustorium Formation and Host Establishment in Hemiparasitic Psittacanthus schiedeanus Mistletoes

The mistletoe Psittacanthus schiedeanus, a keystone species in interaction networks between plants, pollinators, and seed dispersers, infects a wide range of native and non-native tree species of commercial interest. Here, using RNA-seq methodology we assembled the whole circularized quadripartite structure of P. schiedeanus chloroplast genome and described changes in the gene expression of the nuclear genomes across time of experimentally inoculated seeds. Of the 140,467 assembled and annotated uniGenes, 2,000 were identified as differentially expressed (DEGs) and were classified in six distinct clusters according to their expression profiles. DEGs were also classified in enriched functional categories related to synthesis, signaling, homoeostasis, and response to auxin and jasmonic acid. Since many orthologs are involved in lateral or adventitious root formation in other plant species, we propose that in P. schiedeanus (and perhaps in other rootless mistletoe species), these genes participate in haustorium formation by complex regulatory networks here described. Lastly, and according to the structural similarities of P. schiedeanus enzymes with those that are involved in host cell wall degradation in fungi, we suggest that a similar enzymatic arsenal is secreted extracellularly and used by mistletoes species to easily parasitize and break through tissues of the host.

CRFalign: A Sequence-Structure Alignment of Proteins Based on a Combination of HMM-HMM Comparison and Conditional Random Fields

Sequence-structure alignment for protein sequences is an important task for the template-based modeling of 3D structures of proteins. Building a reliable sequence-structure alignment is a challenging problem, especially for remote homologue target proteins. We built a method of sequence-structure alignment called CRFalign, which improves upon a base alignment model based on HMM-HMM comparison by employing pairwise conditional random fields in combination with nonlinear scoring functions of structural and sequence features. Nonlinear scoring part is implemented by a set of gradient boosted regression trees. In addition to sequence profile features, various position-dependent structural features are employed including secondary structures and solvent accessibilities. Training is performed on reference alignments at superfamily levels or twilight zone chosen from the SABmark benchmark set. We found that CRFalign method produces relative improvement in terms of average alignment accuracies for validation sets of SABmark benchmark. We also tested CRFalign on 51 sequence-structure pairs involving 15 FM target domains of CASP14, where we could see that CRFalign leads to an improvement in average modeling accuracies in these hard targets (TM-CRFalign ≃42.94%) compared with that of HHalign (TM-HHalign ≃39.05%) and also that of MRFalign (TM-MRFalign ≃36.93%). CRFalign was incorporated to our template search framework called CRFpred and was tested for a random target set of 300 target proteins consisting of Easy, Medium and Hard sets which showed a reasonable template search performance.

Complete Genome Sequence of a Phage Infecting Sphingomonadaceae

We isolated a phage infecting a member of the Sphingomonadaceae family from a freshwater lake. The phage has a DNA genome of 41,771 bp, with a GC content of 61.7%. The genome harbors 50 predicted protein-coding genes and an auxiliary metabolic gene, which encodes a protein belonging to the radical S-adenosylmethionine superfamily.

Widespread stop-codon recoding in bacteriophages may regulate translation of lytic genes

Bacteriophages (phages) are obligate parasites that use host bacterial translation machinery to produce viral proteins. However, some phages have alternative genetic codes with reassigned stop codons that are predicted to be incompatible with bacterial translation systems. We analysed 9,422 phage genomes and found that stop-codon recoding has evolved in diverse clades of phages that infect bacteria present in both human and animal gut microbiota. Recoded stop codons are particularly over-represented in phage structural and lysis genes. We propose that recoded stop codons might function to prevent premature production of late-stage proteins. Stop-codon recoding has evolved several times in closely related lineages, which suggests that adaptive recoding can occur over very short evolutionary timescales.

Contrastive learning on protein embeddings enlightens midnight zone

Experimental structures are leveraged through multiple sequence alignments, or more generally through homology-based inference (HBI), facilitating the transfer of information from a protein with known annotation to a query without any annotation. A recent alternative expands the concept of HBI from sequence-distance lookup to embedding-based annotation transfer (EAT). These embeddings are derived from protein Language Models (pLMs). Here, we introduce using single protein representations from pLMs for contrastive learning. This learning procedure creates a new set of embeddings that optimizes constraints captured by hierarchical classifications of protein 3D structures defined by the CATH resource. The approach, dubbed ProtTucker, has an improved ability to recognize distant homologous relationships than more traditional techniques such as threading or fold recognition. Thus, these embeddings have allowed sequence comparison to step into the 'midnight zone' of protein similarity, i.e. the region in which distantly related sequences have a seemingly random pairwise sequence similarity. The novelty of this work is in the particular combination of tools and sampling techniques that ascertained good performance comparable or better to existing state-of-the-art sequence comparison methods. Additionally, since this method does not need to generate alignments it is also orders of magnitudes faster. The code is available at https://github.com/Rostlab/EAT.

Using deep learning to annotate the protein universe

Understanding the relationship between amino acid sequence and protein function is a long-standing challenge with far-reaching scientific and translational implications. State-of-the-art alignment-based techniques cannot predict function for one-third of microbial protein sequences, hampering our ability to exploit data from diverse organisms. Here, we train deep learning models to accurately predict functional annotations for unaligned amino acid sequences across rigorous benchmark assessments built from the 17,929 families of the protein families database Pfam. The models infer known patterns of evolutionary substitutions and learn representations that accurately cluster sequences from unseen families. Combining deep models with existing methods significantly improves remote homology detection, suggesting that the deep models learn complementary information. This approach extends the coverage of Pfam by >9.5%, exceeding additions made over the last decade, and predicts function for 360 human reference proteome proteins with no previous Pfam annotation. These results suggest that deep learning models will be a core component of future protein annotation tools.

Subcellular localization of Type VI secretion system assembly in response to cell–cell contact

Bacteria require a number of systems, including the type VI secretion system (T6SS), for interbacterial competition and pathogenesis. The T6SS is a large nanomachine that can deliver toxins directly across membranes of proximal target cells. Since major reassembly of T6SS is necessary after each secretion event, accurate timing and localization of T6SS assembly can lower the cost of protein translocation. Although critically important, mechanisms underlying spatiotemporal regulation of T6SS assembly remain poorly understood. Here, we used super‐resolution live‐cell imaging to show that while Acinetobacter and Burkholderia thailandensis can assemble T6SS at any site, a significant subset of T6SS assemblies localizes precisely to the site of contact between neighboring bacteria. We identified a class of diverse, previously uncharacterized, periplasmic proteins required for this dynamic localization of T6SS to cell–cell contact (TslA). This precise localization is also dependent on the outer membrane porin OmpA. Our analysis links transmembrane communication to accurate timing and localization of T6SS assembly as well as uncovers a pathway allowing bacterial cells to respond to cell–cell contact during interbacterial competition.

Folding and Evolution of a Repeat Protein on the Ribosome

Life on earth is the result of the work of proteins, the cellular nanomachines that fold into elaborated 3D structures to perform their functions. The ribosome synthesizes all the proteins of the biosphere, and many of them begin to fold during translation in a process known as cotranslational folding. In this work we discuss current advances of this field and provide computational and experimental data that highlight the role of ribosome in the evolution of protein structures. First, we used the sequence of the Ankyrin domain from the Drosophila Notch receptor to launch a deep sequence-based search. With this strategy, we found a conserved 33-residue motif shared by different protein folds. Then, to see how the vectorial addition of the motif would generate a full structure we measured the folding on the ribosome of the Ankyrin repeat protein. Not only the on-ribosome folding data is in full agreement with classical in vitro biophysical measurements but also it provides experimental evidence on how folded proteins could have evolved by duplication and fusion of smaller fragments in the RNA world. Overall, we discuss how the ribosomal exit tunnel could be conceptualized as an active site that is under evolutionary pressure to influence protein folding.

Genome-Wide Identification, Classification, Expression and Duplication Analysis of bZIP Family Genes in Juglans regia L

Basic leucine zipper (bZIP), a conserved transcription factor widely found in eukaryotes, has important regulatory roles in plant growth. To understand the information related to the bZIP gene family in walnut, 88 JrbZIP genes were identified at the genome-wide level and classified into 13 subfamilies (A, B, C, D, E, F, G, H, I, J, K, M, and S) using a bioinformatic approach. The number of exons in JrbZIPs ranged from 1 to 12, the number of amino acids in JrbZIP proteins ranged from 145 to 783, and the isoelectric point ranged from 4.85 to 10.05. The majority of JrbZIP genes were localized in the nucleus. The promoter prediction results indicated that the walnut bZIP gene contains a large number of light-responsive and jasmonate-responsive action elements. The 88 JrbZIP genes were involved in DNA binding and nucleus and RNA biosynthetic processes of three ontological categories, molecular functions, cellular components and biological processes. The codon preference analysis showed that the bZIP gene family has a stronger bias for AGA, AGG, UUG, GCU, GUU, and UCU than other codons. Moreover, the transcriptomic data showed that JrbZIP genes might play an important role in floral bud differentiation. The results of a protein interaction network map and kegg enrichment analysis indicated that bZIP genes were mainly involved in phytohormone signaling, anthocyanin synthesis and flowering regulation. qRT-PCR demonstrated the role of the bZIP gene family in floral bud differentiation. Co-expression network maps were constructed for 29 walnut bZIP genes and 6 flowering genes, and JrCO (a homolog of AtCO) was significantly correlated (p < 0.05) with 13 JrbZIP genes in the level of floral bud differentiation expression, including JrbZIP31 (homolog of AtFD), and JrLFY was significantly and positively correlated with JrbZIP10,11,51,59,67 (p < 0.05), and the above results suggest that bZIP family genes may act together with flowering genes to regulate flower bud differentiation in walnut. This study was the first genome-wide report of the walnut bZIP gene family, which could improve our understanding of walnut bZIP proteins and provide a solid foundation for future cloning and functional analyses of this gene family.

DNA-binding protein prediction based on deep transfer learning

The study of DNA binding proteins (DBPs) is of great importance in the biomedical field and plays a key role in this field. At present, many researchers are working on the prediction and detection of DBPs. Traditional DBP prediction mainly uses machine learning methods. Although these methods can obtain relatively high pre-diction accuracy, they consume large quantities of human effort and material resources. Transfer learning has certain advantages in dealing with such prediction problems. Therefore, in the present study, two features were extracted from a protein sequence, a transfer learning method was used, and two classical transfer learning algorithms were compared to transfer samples and construct data sets. In the final step, DBPs are detected by building a deep learning neural network model in a way that uses attention mechanisms.

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.

An analysis of protein language model embeddings for fold prediction

The identification of the protein fold class is a challenging problem in structural biology. Recent computational methods for fold prediction leverage deep learning techniques to extract protein fold-representative embeddings mainly using evolutionary information in the form of multiple sequence alignment (MSA) as input source. In contrast, protein language models (LM) have reshaped the field thanks to their ability to learn efficient protein representations (protein-LM embeddings) from purely sequential information in a self-supervised manner. In this paper, we analyze a framework for protein fold prediction using pre-trained protein-LM embeddings as input to several fine-tuning neural network models, which are supervisedly trained with fold labels. In particular, we compare the performance of six protein-LM embeddings: the long short-term memory-based UniRep and SeqVec, and the transformer-based ESM-1b, ESM-MSA, ProtBERT and ProtT5; as well as three neural networks: Multi-Layer Perceptron, ResCNN-BGRU (RBG) and Light-Attention (LAT). We separately evaluated the pairwise fold recognition (PFR) and direct fold classification (DFC) tasks on well-known benchmark datasets. The results indicate that the combination of transformer-based embeddings, particularly those obtained at amino acid level, with the RBG and LAT fine-tuning models performs remarkably well in both tasks. To further increase prediction accuracy, we propose several ensemble strategies for PFR and DFC, which provide a significant performance boost over the current state-of-the-art results. All this suggests that moving from traditional protein representations to protein-LM embeddings is a very promising approach to protein fold-related tasks.

LOMETS3: integrating deep learning and profile alignment for advanced protein template recognition and function annotation

Deep learning techniques have significantly advanced the field of protein structure prediction. LOMETS3 (https://zhanglab.ccmb.med.umich.edu/LOMETS/) is a new generation meta-server approach to template-based protein structure prediction and function annotation, which integrates newly developed deep learning threading methods. For the first time, we have extended LOMETS3 to handle multi-domain proteins and to construct full-length models with gradient-based optimizations. Starting from a FASTA-formatted sequence, LOMETS3 performs four steps of domain boundary prediction, domain-level template identification, full-length template/model assembly and structure-based function prediction. The output of LOMETS3 contains (i) top-ranked templates from LOMETS3 and its component threading programs, (ii) up to 5 full-length structure models constructed by L-BFGS (limited-memory Broyden–Fletcher–Goldfarb–Shanno algorithm) optimization, (iii) the 10 closest Protein Data Bank (PDB) structures to the target, (iv) structure-based functional predictions, (v) domain partition and assembly results, and (vi) the domain-level threading results, including items (i)–(iii) for each identified domain. LOMETS3 was tested in large-scale benchmarks and the blind CASP14 (14th Critical Assessment of Structure Prediction) experiment, where the overall template recognition and function prediction accuracy is significantly beyond its predecessors and other state-of-the-art threading approaches, especially for hard targets without homologous templates in the PDB. Based on the improved developments, LOMETS3 should help significantly advance the capability of broader biomedical community for template-based protein structure and function modelling.

3DLigandSite: structure-based prediction of protein-ligand binding sites

3DLigandSite is a web tool for the prediction of ligand-binding sites in proteins. Here, we report a significant update since the first release of 3DLigandSite in 2010. The overall methodology remains the same, with candidate binding sites in proteins inferred using known binding sites in related protein structures as templates. However, the initial structural modelling step now uses the newly available structures from the AlphaFold database or alternatively Phyre2 when AlphaFold structures are not available. Further, a sequence-based search using HHSearch has been introduced to identify template structures with bound ligands that are used to infer the ligand-binding residues in the query protein. Finally, we introduced a machine learning element as the final prediction step, which improves the accuracy of predictions and provides a confidence score for each residue predicted to be part of a binding site. Validation of 3DLigandSite on a set of 6416 binding sites obtained 92% recall at 75% precision for non-metal binding sites and 52% recall at 75% precision for metal binding sites. 3DLigandSite is available at https://www.wass-michaelislab.org/3dligandsite. Users submit either a protein sequence or structure. Results are displayed in multiple formats including an interactive Mol* molecular visualization of the protein and the predicted binding sites.

Developments in Algorithms for Sequence Alignment: A Review

The continuous development of sequencing technologies has enabled researchers to obtain large amounts of biological sequence data, and this has resulted in increasing demands for software that can perform sequence alignment fast and accurately. A number of algorithms and tools for sequence alignment have been designed to meet the various needs of biologists. Here, the ideas that prevail in the research of sequence alignment and some quality estimation methods for multiple sequence alignment tools are summarized.

GWYRE: A resource for mapping variants onto experimental and modeled structures of human protein complexes

•

Structure of protein complexes is important for interpreting genetic variation.

•

Data on single amino acid variants is available from high-throughput sequencing.

•

Integrated modeling approach was applied to proteins and their complexes.

•

GWYRE resource incorporates predicted protein complexes with mapped mutations.

Rapid progress in structural modeling of proteins and their interactions is powered by advances in knowledge-based methodologies along with better understanding of physical principles of protein structure and function. The pool of structural data for modeling of proteins and protein–protein complexes is constantly increasing due to the rapid growth of protein interaction databases and Protein Data Bank. The GWYRE (Genome Wide PhYRE) project capitalizes on these developments by advancing and applying new powerful modeling methodologies to structural modeling of protein–protein interactions and genetic variation. The methods integrate knowledge-based tertiary structure prediction using Phyre2 and quaternary structure prediction using template-based docking by a full-structure alignment protocol to generate models for binary complexes. The predictions are incorporated in a comprehensive public resource for structural characterization of the human interactome and the location of human genetic variants. The GWYRE resource facilitates better understanding of principles of protein interaction and structure/function relationships. The resource is available at http://www.gwyre.org.

VHH Structural Modelling Approaches: A Critical Review

VHH, i.e., VH domains of camelid single-chain antibodies, are very promising therapeutic agents due to their significant physicochemical advantages compared to classical mammalian antibodies. The number of experimentally solved VHH structures has significantly improved recently, which is of great help, because it offers the ability to directly work on 3D structures to humanise or improve them. Unfortunately, most VHHs do not have 3D structures. Thus, it is essential to find alternative ways to get structural information. The methods of structure prediction from the primary amino acid sequence appear essential to bypass this limitation. This review presents the most extensive overview of structure prediction methods applied for the 3D modelling of a given VHH sequence (a total of 21). Besides the historical overview, it aims at showing how model software programs have been shaping the structural predictions of VHHs. A brief explanation of each methodology is supplied, and pertinent examples of their usage are provided. Finally, we present a structure prediction case study of a recently solved VHH structure. According to some recent studies and the present analysis, AlphaFold 2 and NanoNet appear to be the best tools to predict a structural model of VHH from its sequence.

EnsembleFam: towards more accurate protein family prediction in the twilight zone

Background

Current protein family modeling methods like profile Hidden Markov Model (pHMM), k-mer based methods, and deep learning-based methods do not provide very accurate protein function prediction for proteins in the twilight zone, due to low sequence similarity to reference proteins with known functions.

Results

We present a novel method EnsembleFam, aiming at better function prediction for proteins in the twilight zone. EnsembleFam extracts the core characteristics of a protein family using similarity and dissimilarity features calculated from sequence homology relations. EnsembleFam trains three separate Support Vector Machine (SVM) classifiers for each family using these features, and an ensemble prediction is made to classify novel proteins into these families. Extensive experiments are conducted using the Clusters of Orthologous Groups (COG) dataset and G Protein-Coupled Receptor (GPCR) dataset. EnsembleFam not only outperforms state-of-the-art methods on the overall dataset but also provides a much more accurate prediction for twilight zone proteins.

Conclusions

EnsembleFam, a machine learning method to model protein families, can be used to better identify members with very low sequence homology. Using EnsembleFam protein functions can be predicted  using just sequence information with better accuracy than state-of-the-art methods.

Construct a variable-length fragment library for de novo protein structure prediction

Although remarkable achievements, such as AlphaFold2, have been made in end-to-end structure prediction, fragment libraries remain essential for de novo protein structure prediction, which can help explore and understand the protein-folding mechanism. In this work, we developed a variable-length fragment library (VFlib). In VFlib, a master structure database was first constructed from the Protein Data Bank through sequence clustering. The hidden Markov model (HMM) profile of each protein in the master structure database was generated by HHsuite, and the secondary structure of each protein was calculated by DSSP. For the query sequence, the HMM-profile was first constructed. Then, variable-length fragments were retrieved from the master structure database through dynamically variable-length profile-profile comparison. A complete method for chopping the query HMM-profile during this process was proposed to obtain fragments with increased diversity. Finally, secondary structure information was used to further screen the retrieved fragments to generate the final fragment library of specific query sequence. The experimental results obtained with a set of 120 nonredundant proteins show that the global precision and coverage of the fragment library generated by VFlib were 55.04% and 94.95% at the RMSD cutoff of 1.5 Å, respectively. Compared with the benchmark method of NNMake, the global precision of our fragment library had increased by 62.89% with equivalent coverage. Furthermore, the fragments generated by VFlib and NNMake were used to predict structure models through fragment assembly. Controlled experimental results demonstrate that the average TM-score of VFlib was 16.00% higher than that of NNMake.

Surface-based protein domains retrieval methods from a SHREC2021 challenge

Proteins are essential to nearly all cellular mechanism and the effectors of the cells activities. As such, they often interact through their surface with other proteins or other cellular ligands such as ions or organic molecules. The evolution generates plenty of different proteins, with unique abilities, but also proteins with related functions hence similar 3D surface properties (shape, physico-chemical properties, …). The protein surfaces are therefore of primary importance for their activity. In the present work, we assess the ability of different methods to detect such similarities based on the geometry of the protein surfaces (described as 3D meshes), using either their shape only, or their shape and the electrostatic potential (a biologically relevant property of proteins surface). Five different groups participated in this contest using the shape-only dataset, and one group extended its pre-existing method to handle the electrostatic potential. Our comparative study reveals both the ability of the methods to detect related proteins and their difficulties to distinguish between highly related proteins. Our study allows also to analyze the putative influence of electrostatic information in addition to the one of protein shapes alone. Finally, the discussion permits to expose the results with respect to ones obtained in the previous contests for the extended method. The source codes of each presented method have been made available online.

Long-Read Sequencing Reveals Genetic Adaptation of Bartonella Adhesin A Among Different Bartonella henselae Isolates

Bartonella henselae is the causative agent of cat scratch disease and other clinical entities such as endocarditis and bacillary angiomatosis. The life cycle of this pathogen, with alternating host conditions, drives evolutionary and host-specific adaptations. Human, feline, and laboratory adapted B. henselae isolates often display genomic and phenotypic differences that are related to the expression of outer membrane proteins, for example the Bartonella adhesin A (BadA). This modularly-structured trimeric autotransporter adhesin is a major virulence factor of B. henselae and is crucial for the initial binding to the host via the extracellular matrix proteins fibronectin and collagen. By using next-generation long-read sequencing we demonstrate a conserved genome among eight B. henselae isolates and identify a variable genomic badA island with a diversified and highly repetitive badA gene flanked by badA pseudogenes. Two of the eight tested B. henselae strains lack BadA expression because of frameshift mutations. We suggest that active recombination mechanisms, possibly via phase variation (i.e., slipped-strand mispairing and site-specific recombination) within the repetitive badA island facilitate reshuffling of homologous domain arrays. The resulting variations among the different BadA proteins might contribute to host immune evasion and enhance long-term and efficient colonisation in the differing host environments. Considering the role of BadA as a key virulence factor, it remains important to check consistently and regularly for BadA surface expression during experimental infection procedures.

Bacterial developmental checkpoint that directly monitors cell surface morphogenesis

Bacillus subtilis spores are encased in two concentric shells: an outer proteinaceous "coat" and an inner peptidoglycan "cortex," separated by a membrane. Cortex assembly depends on coat assembly initiation, but how cells achieve this coordination across the membrane is unclear. Here, we report that the protein SpoVID monitors the polymerization state of the coat basement layer via an extension to a functional intracellular LysM domain that arrests sporulation when coat assembly is initiated improperly. Whereas extracellular LysM domains bind mature peptidoglycan, SpoVID LysM binds to the membrane-bound lipid II peptidoglycan precursor. We propose that improper coat assembly exposes the SpoVID LysM domain, which then sequesters lipid II and prevents cortex assembly. SpoVID defines a widespread group of firmicute proteins with a characteristic N-terminal domain and C-terminal peptidoglycan-binding domains that might combine coat and cortex assembly roles to mediate a developmental checkpoint linking the morphogenesis of two spatially separated supramolecular structures.

ProALIGN: Directly Learning Alignments for Protein Structure Prediction via Exploiting Context-Specific Alignment Motifs

Template-based modeling (TBM), including homology modeling and protein threading, is one of the most reliable techniques for protein structure prediction. It predicts protein structure by building an alignment between the query sequence under prediction and the templates with solved structures. However, it is still very challenging to build the optimal sequence-template alignment, especially when only distantly related templates are available. Here we report a novel deep learning approach ProALIGN that can predict much more accurate sequence-template alignment. Like protein sequences consisting of sequence motifs, protein alignments are also composed of frequently occurring alignment motifs with characteristic patterns. Alignment motifs are context-specific as their characteristic patterns are tightly related to sequence contexts of the aligned regions. Inspired by this observation, we represent a protein alignment as a binary matrix (in which 1 denotes an aligned residue pair) and then use a deep convolutional neural network to predict the optimal alignment from the query protein and its template. The trained neural network implicitly but effectively encodes an alignment scoring function, which reduces inaccuracies in the handcrafted scoring functions widely used by the current threading approaches. For a query protein and a template, we apply the neural network to directly infer likelihoods of all possible residue pairs in their entirety, which could effectively consider the correlations among multiple residues. We further construct the alignment with maximum likelihood, and finally build a structure model according to the alignment. Tested on three independent data sets with a total of 6688 protein alignment targets and 80 CASP13 TBM targets, our method achieved much better alignments and 3D structure models than the existing methods, including HHpred, CNFpred, CEthreader, and DeepThreader. These results clearly demonstrate the effectiveness of exploiting the context-specific alignment motifs by deep learning for protein threading.

Catalytic trajectory of a dimeric nonribosomal peptide synthetase subunit with an inserted epimerase domain

Nonribosomal peptide synthetases (NRPSs) are modular assembly-line megaenzymes that synthesize diverse metabolites with wide-ranging biological activities. The structural dynamics of synthetic elongation has remained unclear. Here, we present cryo-EM structures of PchE, an NRPS elongation module, in distinct conformations. The domain organization reveals a unique “H”-shaped head-to-tail dimeric architecture. The capture of both aryl and peptidyl carrier protein-tethered substrates and intermediates inside the heterocyclization domain and l-cysteinyl adenylate in the adenylation domain illustrates the catalytic and recognition residues. The multilevel structural transitions guided by the adenylation C-terminal subdomain in combination with the inserted epimerase and the conformational changes of the heterocyclization tunnel are controlled by two residues. Moreover, we visualized the direct structural dynamics of the full catalytic cycle from thiolation to epimerization. This study establishes the catalytic trajectory of PchE and sheds light on the rational re-engineering of domain-inserted dimeric NRPSs for the production of novel pharmaceutical agents.

The catalytic domains in nonribosomal peptide synthetases (NRPSs) are responsible for a choreography of events that elongates substrates into natural products. Here, the authors present cryo-EM structures of a siderophore-producing dimeric NRPS elongation module in multiple distinct conformations, which provides insight into the mechanisms of catalytic trajectory.

DisCovER: distance- and orientation-based covariational threading for weakly homologous proteins

Threading a query protein sequence onto a library of weakly homologous structural templates remains challenging, even when sequence-based predicted contact or distance information is used. Contact-assisted or distance-assisted threading methods utilize only the spatial proximity of the interacting residue pairs for template selection and alignment, ignoring their orientation. Moreover, existing threading methods fail to consider the neighborhood effect induced by the query-template alignment. We present a new distance- and orientation-based covariational threading method called DisCovER by effectively integrating information from inter-residue distance and orientation along with the topological network neighborhood of a query-template alignment. Our method first selects a subset of templates using standard profile-based threading coupled with topological network similarity terms to account for the neighborhood effect and subsequently performs distance- and orientation-based query-template alignment using an iterative double dynamic programming framework. Multiple large-scale benchmarking results on query proteins classified as weakly homologous from the continuous automated model evaluation experiment and from the current literature show that our method outperforms several existing state-of-the-art threading approaches, and that the integration of the neighborhood effect with the inter-residue distance and orientation information synergistically contributes to the improved performance of DisCovER. DisCovER is freely available at https://github.com/Bhattacharya-Lab/DisCovER.

Native Ion Mobility-Mass Spectrometry-Enabled Fast Structural Interrogation of Labile Protein Surface Modifications at the Intact Protein Level

Protein sialylation has been closely linked to many diseases including Alzheimer's disease (AD). It is also broadly implicated in therapeutics operating in a pattern-dependent (e.g., Neu5Ac vs Neu5Gc) manner. However, how the sialylation pattern affects the AD-associated, transferrin-assisted iron/Aβ cellular uptake process remains largely ill-defined. Herein, we report the use of native ion mobility-mass spectrometry (IM-MS)-based fast structural probing methodology, enabling well-controlled, synergistic, and in situ manipulation of mature glycoproteins and attached sialic acids. IM-MS-centered experiments enable the combinatorial interrogation of sialylation effects on Aβ cytotoxicity and the chemical, conformational, and topological stabilities of transferrin. Cell viability experiments suggest that Neu5Gc replacement enhances the transferrin-assisted, iron loading-associated Aβ cytotoxicity. Native gel electrophoresis and IM-MS reveal that sialylation stabilizes transferrin conformation but inhibits its dimerization. Collectively, IM-MS is adapted to capture key sialylation intermediates involved in fine-tuning AD-associated glycoprotein structural microheterogeneity. Our results provide the molecular basis for the importance of sustaining moderate TF sialylation levels, especially Neu5Ac, in promoting iron cellular transportation and rescuing iron-enhanced Aβ cytotoxicity.

Bacterial gasdermins reveal an ancient mechanism of cell death

[Figure: see text].

Rel/NF-κB Transcription Factors Emerged at the Onset of Opisthokonts

The Rel/NF-κB transcription factor family has myriad roles in immunity, development, and differentiation in animals, and was considered a key innovation for animal multicellularity. Rel homology domain-containing proteins were previously hypothesized to have originated in a last common ancestor of animals and some of their closest unicellular relatives. However, key taxa were missing from previous analyses, necessitating a systematic investigation into the distribution and evolution of these proteins. Here, we address this knowledge gap by surveying taxonomically broad data from eukaryotes, with a special emphasis on lineages closely related to animals. We report an earlier origin for Rel/NF-κB proteins than previously described, in the last common ancestor of animals and fungi, and show that even in the sister group to fungi, these proteins contain elements that in animals are necessary for the subcellular regulation of Rel/NF-κB.

Profiles of Natural and Designed Protein-Like Sequences Effectively Bridge Protein Sequence Gaps: Implications in Distant Homology Detection

Sequence-based approaches are fundamental to guide experimental investigations in obtaining structural and/or functional insights into uncharacterized protein families. Powerful profile-based sequence search methods rely on a sequence space continuum to identify non-trivial relationships through homology detection. The computational design of protein-like sequences that serve as "artificial linkers" is useful in identifying relationships between distant members of a structural fold. Such sequences act as intermediates and guide homology searches between distantly related proteins. Here, we describe an approach that represents natural intermediate sequences and designed protein-like sequences as HMM (Hidden Markov Models) profiles, to improve the sensitivity of existing search methods. Searches made within the "Profile database" were shown to recognize the parent structural fold for 90% of the search queries at query coverage better than 60%. For 1040 protein families with no available structure, fold associations were made through searches in the database of natural and designed sequence profiles. Most of the associations were made with the Alpha-alpha superhelix, Transmembrane beta-barrels, TIM barrel, and Immunoglobulin-like beta-sandwich folds. For 11 domain families of unknown functions, we provide confident fold associations using the profiles of designed sequences and a consensus from other fold recognition methods. For two DUFs (Domain families of Unknown Functions), we performed detailed functional annotation through comparisons with characterized templates of families of known function.

Predicting the capsid architecture of phages from metagenomic data

Tailed phages are viruses that infect bacteria and are the most abundant biological entities on Earth. Their ecological, evolutionary, and biogeochemical roles in the planet stem from their genomic diversity. Known tailed phage genomes range from 10 to 735 kilobase pairs thanks to the size variability of the protective protein capsids that store them. However, the role of tailed phage capsids’ diversity in ecosystems is unclear. A fundamental gap is the difficulty of associating genomic information with viral capsids in the environment. To address this problem, here, we introduce a computational approach to predict the capsid architecture (T-number) of tailed phages using the sequence of a single gene—the major capsid protein. This approach relies on an allometric model that relates the genome length and capsid architecture of tailed phages. This allometric model was applied to isolated phage genomes to generate a library that associated major capsid proteins and putative capsid architectures. This library was used to train machine learning methods, and the most computationally scalable model investigated (random forest) was applied to human gut metagenomes. Compared to isolated phages, the analysis of gut data reveals a large abundance of mid-sized (T = 7) capsids, as expected, followed by a relatively large frequency of jumbo-like tailed phage capsids (T ≥ 25) and small capsids (T = 4) that have been under-sampled. We discussed how to increase the method’s accuracy and how to extend the approach to other viruses. The computational pipeline introduced here opens the doors to monitor the ongoing evolution and selection of viral capsids across ecosystems.