PhANNs, a fast and accurate tool and web server to classify phage structural proteins

Cantu, Adrian; Salamon, Peter; Seguritan, Victor; Redfield, Jackson; Salamon, David; Edwards, Robert A.; Segall, Anca M.

doi:10.1101/2020.04.03.023523

Cited by 17 publications

(9 citation statements)

References 32 publications

(21 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Within the 24 main target PCs, a majority of sequences (71%) did not display any significant sequence similarity to any known protein domain outside of the C-terminal VR region, even when using highly-sensitive annotation tools such as HHblits 23 . Hence, we instead classified targets into broad functional classes, namely “structural proteins” vs “unknown” for viral-encoded DGRs and “membrane-bound” proteins vs “unknown” for cellular-encoded DGRs, using non-similarity-based protein annotation approaches 24 , 25 (Supplementary Note 8 ).…”

Section: Resultsmentioning

confidence: 99%

“…Annotations were derived from hits with a score of ≥50 in hmmsearch or ≥90% probability in hhblits, except for hits overlapping the prediction VR region for which these cutoffs were lowered to ≥30 on score and ≥80% probability, in order to enable the identification of distantly related C-lectin folds. In addition, individual target sequences were also searched for transmembrane domains and signal peptides using TMHMM 63 v2.0c (default parameters) and SignalP 64 v4.1 (score D ≥ Dmaxcut), and searched for potential Caudovirales structural proteins (capsid or tail proteins) using DeepCapTail v3038c4d 24 (version downloaded Jan. 2020) and PhANNs v1.0.0 25 with thresholds of ≥0.9 and ≥0.2 on the score, respectively. The same clustering and annotation pipeline were applied to predicted cds from NCBI RefSeq Caudovirales genomes, after having dereplicated these protein sequences at 99% using cd-hit 44 v4.8.1 ( n = 250,209 proteins), in order to evaluate the functional annotation across all Caudovirales of (i) sequences containing an Ig-like domain (ii) sequences predicted as “capsid” or “tail” via DeepCapTail (see Supplementary Note 8 ).…”

Section: Methodsmentioning

confidence: 99%

See 1 more Smart Citation

Ecology and molecular targets of hypermutation in the global microbiome

et al. 2021

View full text Add to dashboard Cite

Changes in the sequence of an organism’s genome, i.e., mutations, are the raw material of evolution. The frequency and location of mutations can be constrained by specific molecular mechanisms, such as diversity-generating retroelements (DGRs). DGRs have been characterized from cultivated bacteria and bacteriophages, and perform error-prone reverse transcription leading to mutations being introduced in specific target genes. DGR loci were also identified in several metagenomes, but the ecological roles and evolutionary drivers of these DGRs remain poorly understood. Here, we analyze a dataset of >30,000 DGRs from public metagenomes, establish six major lineages of DGRs including three primarily encoded by phages and seemingly used to diversify host attachment proteins, and demonstrate that DGRs are broadly active and responsible for >10% of all amino acid changes in some organisms. Overall, these results highlight the constraints under which DGRs evolve, and elucidate several distinct roles these elements play in natural communities.

show abstract

Section: Resultsmentioning

confidence: 99%

Section: Methodsmentioning

confidence: 99%

Ecology and molecular targets of hypermutation in the global microbiome

et al. 2021

View full text Add to dashboard Cite

show abstract

“…Diversity generating retroelements were identified in members of cluster1819 by using both the myDGR web server [67] and MetaCSST(v1.0) [68] tools. To predict whether the target genes identified were putative tail fibre genes, as has previously been suggested [69], the proteins from all members were analysed with PhANNs(v1.0.0) [70] and the most significant hit to a tail fibre gene was carried forward. These were then compared with the results from the previous tools.…”

Section: Identification Of Diversity Generating Retroelements In the Most Abundant Prophage Clustermentioning

confidence: 99%

Prophages in the infant gut are largely induced, and may be functionally relevant to their hosts

Redgwell

Thorsen

Petit

et al. 2021

Preprint

View full text Add to dashboard Cite

Background: Bacteriophages are the most abundant biological entity on the planet, and are key components of any ecosystem they are present in. The gut virome is increasingly being implicated in disease states although these studies largely focus on lytic phages in adults. Here we identify prophages from a large infant cohort and investigate their potential functions. Results: We identified 10645 vOTUs from 662 metagenomes. No core virome was found: the most prevalent vOTU was identified in 70% of the samples. The most abundant and prevalent group of phages are a novel group closely related to Bacteroides phage Hanky p00. Functional annotation of this group revealed the presence of genes in the dDTP-L-rhamnose pathway, possibly involved in the production of capsular polysaccharides. We also found an abundance of diversity generating retroelements in the phages. Additionally, paired virome data allowed us to show that the majority of prophages are induced in at least one sample and that this is not affected by the use of antibiotics in the 4 weeks prior to sampling. Conclusions: Prophages in the infant gut are largely unique to the individual and not shared. Most of them appear to be induced and so may be key drivers in shaping the bacterial microbiome. The most abundant group of phages are novel, and possess elements that may allow them to maintain differentially susceptible subpopulations of their host bacterium; whilst also containing diversity generating retroelements that could expand their host range. Prophages are important components of the infant gut that may have far reaching influences on the composition and function of the microbiome.

show abstract

“…However, among the limitations of our findings is the use of database-dependent analyses. In the future, tools such as artificial neural networks (Mendez et al, 2019;Cantu et al, 2020) can be used in bioinformatic and chemoinformatic approaches to shed light on sequence and mass-spectrometry data that is absent in repositories. Furthermore, less invasive sampling techniques could be used to capture molecular dynamics over relevant temporal scales (e.g., before and after stress events).…”

Section: Current Limitations and Future Directionsmentioning

confidence: 99%

Three-Dimensional Molecular Cartography of the Caribbean Reef-Building Coral Orbicella faveolata

et al. 2021

View full text Add to dashboard Cite

All organisms host a diversity of associated viruses, bacteria, and protists, collectively defined as the holobiont. While scientific advancements have enhanced the understanding of the functional roles played by various components of the holobiont, there is a growing need to integrate multiple types of molecular data into spatially and temporally resolved frameworks. To that end, we mapped 16S and 18S rDNA metabarcoding, metatranscriptomics, and metabolomic data onto three-dimensional reconstructions of coral colonies to examine microbial diversity, microbial gene expression, and biochemistry on two colonies of the ecologically important, reef-building coral, Orbicella faveolata and their competitors (i.e., adjacent organisms interacting with the corals: fleshy algae, turf algae, hydrozoans, and other corals). Overall, no statistically significant spatial patterns were observed among the samples for any of the data types; instead, strong signatures of the macroorganismal hosts (e.g., coral, algae, hydrozoa) were detected, in the microbiome, the transcriptome, and the metabolome. The 16S rDNA analysis demonstrated higher abundance of Firmicutes in the coral microbiome than in its competitors. A single bacterial amplicon sequence variant from the genus Clostridium was found exclusively in all O. faveolata samples. In contrast to microbial taxa, a portion of the functionally annotated bacterial RNA transcripts (6.86%) and metabolites (1.95%) were ubiquitous in all coral and competitor samples. Machine learning analysis of microbial transcripts revealed elevated T7-like cyanophage-encoded photosystem II transcripts in O. faveolata samples, while sequences involved in bacterial cell division were elevated in turf algal and interface samples. Similar analysis of metabolites revealed that bacterial-produced antimicrobial and antifungal compounds were highly enriched in coral samples. This study provides insight into the spatial and biological patterning of the coral microbiome, transcriptome, and metabolome.

show abstract

PhANNs, a fast and accurate tool and web server to classify phage structural proteins

Cited by 17 publications

References 32 publications

Ecology and molecular targets of hypermutation in the global microbiome

Ecology and molecular targets of hypermutation in the global microbiome

Prophages in the infant gut are largely induced, and may be functionally relevant to their hosts

Three-Dimensional Molecular Cartography of the Caribbean Reef-Building Coral Orbicella faveolata

Contact Info

Product

Resources

About