Investigation of machine learning algorithms for taxonomic classification of marine metagenomes.

Microbial communities play key roles in ocean ecosystems through regulation of biogeochemical processes such as carbon and nutrient cycling, food web dynamics, and gut microbiomes of invertebrates, fish, reptiles, and mammals. Assessments of marine microbial diversity are therefore critical to understanding spatiotemporal variations in microbial community structure and function in ocean ecosystems. With recent advances in DNA shotgun sequencing for metagenome samples and computational analysis, it is now possible to access the taxonomic and genomic content of ocean microbial communities to study their structural patterns, diversity, and functional potential. However, existing taxonomic classification tools depend upon manually curated phylogenetic trees, which can create inaccuracies in metagenomes from less well-characterized communities, such as from ocean water. Herein, we explore the utility of deep learning tools-DeepMicrobes and a novel Residual Network architecture-that leverage natural language processing and convolutional neural network architectures to map input sequence data (k-mers) to output labels (taxonomic groups) without reliance on a curated taxonomic tree. We trained both models using metagenomic reads simulated from marine microbial genomes in the MarRef database. The performance of both models (accuracy, precision, and percent microbe predicted) was compared with the standard taxonomic classification tool Kraken2 using 10 complex metagenomic data sets simulated from MarRef. Our results demonstrate that time, compute power, and microbial genomic diversity still pose challenges for machine learning (ML). Moreover, our results suggest that high genome coverage and rectification of class imbalance are prerequisites for a well-trained model, and therefore should be a major consideration in future ML work. IMPORTANCE Taxonomic profiling of microbial communities is essential to model microbial interactions and inform habitat conservation. This work develops approaches in constructing training/testing data sets from publicly available marine metagenomes and evaluates the performance of machine learning (ML) approaches in read-based taxonomic classification of marine metagenomes. Predictions from two models are used to test accuracy in metagenomic classification and to guide improvements in ML approaches. Our study provides insights on the methods, results, and challenges of deep learning on marine microbial metagenomic data sets. Future machine learning approaches can be improved by rectifying genome coverage and class imbalance in the training data sets, developing alternative models, and increasing the accessibility of computational resources for model training and refinement.

Decoding dissolved information: environmental DNA sequencing at global scale to monitor a changing ocean.

The use of environmental DNA (eDNA) technology for environmental monitoring is rapidly expanding, with applications for fisheries, coral reefs, harmful algal blooms, invasive and endangered species, and biodiversity monitoring. By enabling detection of species over space and time, eDNA fulfills a fundamental need of environmental surveys. Traditional surveys are expensive, require significant capital expenditure, and can be destructive; eDNA offers promise for cheaper, less invasive, and higher-resolution (i.e. genetic) assessments of environments and stocks. However, challenges in quantification, detection limits, biobanking capacity, reference databases, and data management and integration remain significant hurdles to efficient eDNA monitoring at global and decadal scale. Here, we consider the current state of eDNA technology and its suitability for the problems for which it is being used. We explore the current best practices, the logistical and social challenges that prevent eDNA from widespread adoption and benefit, and the emerging technologies that may address those challenges.

Standardized multi-omics of Earth's microbiomes reveals microbial and metabolite diversity.

Despite advances in sequencing, lack of standardization makes comparisons across studies challenging and hampers insights into the structure and function of microbial communities across multiple habitats on a planetary scale. Here we present a multi-omics analysis of a diverse set of 880 microbial community samples collected for the Earth Microbiome Project. We include amplicon (16S, 18S, ITS) and shotgun metagenomic sequence data, and untargeted metabolomics data (liquid chromatography-tandem mass spectrometry and gas chromatography mass spectrometry). We used standardized protocols and analytical methods to characterize microbial communities, focusing on relationships and co-occurrences of microbially related metabolites and microbial taxa across environments, thus allowing us to explore diversity at extraordinary scale. In addition to a reference database for metagenomic and metabolomic data, we provide a framework for incorporating additional studies, enabling the expansion of existing knowledge in the form of an evolving community resource. We demonstrate the utility of this database by testing the hypothesis that every microbe and metabolite is everywhere but the environment selects. Our results show that metabolite diversity exhibits turnover and nestedness related to both microbial communities and the environment, whereas the relative abundances of microbially related metabolites vary and co-occur with specific microbial consortia in a habitat-specific manner. We additionally show the power of certain chemistry, in particular terpenoids, in distinguishing Earth's environments (for example, terrestrial plant surfaces and soils, freshwater and marine animal stool), as well as that of certain microbes including Conexibacter woesei (terrestrial soils), Haloquadratum walsbyi (marine deposits) and Pantoea dispersa (terrestrial plant detritus). This Resource provides insight into the taxa and metabolites within microbial communities from diverse habitats across Earth, informing both microbial and chemical ecology, and provides a foundation and methods for multi-omics microbiome studies of hosts and the environment.

Tourmaline: A containerized workflow for rapid and iterable amplicon sequence analysis using QIIME 2 and Snakemake.

BACKGROUND: Amplicon sequencing (metabarcoding) is a common method to survey diversity of environmental communities whereby a single genetic locus is amplified and sequenced from the DNA of whole or partial organisms, organismal traces (e.g., skin, mucus, feces), or microbes in an environmental sample. Several software packages exist for analyzing amplicon data, among which QIIME 2 has emerged as a popular option because of its broad functionality, plugin architecture, provenance tracking, and interactive visualizations. However, each new analysis requires the user to keep track of input and output file names, parameters, and commands; this lack of automation and standardization is inefficient and creates barriers to meta-analysis and sharing of results. FINDINGS: We developed Tourmaline, a Python-based workflow that implements QIIME 2 and is built using the Snakemake workflow management system. Starting from a configuration file that defines parameters and input files-a reference database, a sample metadata file, and a manifest or archive of FASTQ sequences-it uses QIIME 2 to run either the DADA2 or Deblur denoising algorithm; assigns taxonomy to the resulting representative sequences; performs analyses of taxonomic, alpha, and beta diversity; and generates an HTML report summarizing and linking to the output files. Features include support for multiple cores, automatic determination of trimming parameters using quality scores, representative sequence filtering (taxonomy, length, abundance, prevalence, or ID), support for multiple taxonomic classification and sequence alignment methods, outlier detection, and automated initialization of a new analysis using previous settings. The workflow runs natively on Linux and macOS or via a Docker container. We ran Tourmaline on a 16S ribosomal RNA amplicon data set from Lake Erie surface water, showing its utility for parameter optimization and the ability to easily view interactive visualizations through the HTML report, QIIME 2 viewer, and R- and Python-based Jupyter notebooks. CONCLUSION: Automated workflows like Tourmaline enable rapid analysis of environmental amplicon data, decreasing the time from data generation to actionable results. Tourmaline is available for download at github.com/aomlomics/tourmaline.

Dominance of Sulfurospirillum in Metagenomes Associated with the Methane Ice Worm (Sirsoe methanicola).

Sirsoe methanicola, commonly known as the methane ice worm, is the only macrofaunal species known to inhabit the Gulf of Mexico methane hydrates. Little is known about this elusive marine polychaete that can colonize rich carbon and energy reserves. Metagenomic analysis of gut contents and worm fragments predicted diverse metabolic capabilities with the ability to utilize a range of nitrogen, sulfur, and organic carbon compounds through microbial taxa affiliated with Campylobacterales, Desulfobacterales, Enterobacterales, SAR324, Alphaproteobacteria, and Mycoplasmatales. Entomoplasmatales and Chitinivibrionales were additionally identified from extracted full-length 16S rRNA sequences, and read analysis identified 196 bacterial families. Overall, the microbial community appeared dominated by uncultured Sulfurospirillum, a taxon previously considered free-living rather than host-associated. Metagenome-assembled genomes (MAGs) classified as uncultured Sulfurospirillum predicted thiosulfate disproportionation and the reduction of tetrathionate, sulfate, sulfide/polysulfide, and nitrate. Microbial amino acid and vitamin B12 biosynthesis genes were identified in multiple MAGs, suggesting nutritional value to the host. Reads assigned to aerobic or anaerobic methanotrophic taxa were rare. IMPORTANCE Methane hydrates represent vast reserves of natural gas with roles in global carbon cycling and climate change. This study provided the first analysis of metagenomes associated with Sirsoe methanicola, the only polychaete species known to colonize methane hydrates. Previously unrecognized participation of Sulfurospirillum in a gut microbiome is provided, and the role of sulfur compound redox reactions within this community is highlighted. The comparative biology of S. methanicola is of general interest given research into the adverse effects of sulfide production in human gut microbiomes. In addition, taxonomic assignments are provided for nearly 200 bacterial families, expanding our knowledge of microbiomes in the deep sea.

MIxS-SA: a MIxS extension defining the minimum information standard for sequence data from symbiont-associated micro-organisms.

The symbiont-associated (SA) environmental package is a new extension to the minimum information about any (x) sequence (MIxS) standards, established by the Parasite Microbiome Project (PMP) consortium, in collaboration with the Genomics Standard Consortium. The SA was built upon the host-associated MIxS standard, but reflects the nestedness of symbiont-associated microbiota within and across host-symbiont-microbe interactions. This package is designed to facilitate the collection and reporting of a broad range of metadata information that apply to symbionts such as life history traits, association with one or multiple host organisms, or the nature of host-symbiont interactions along the mutualism-parasitism continuum. To better reflect the inherent nestedness of all biological systems, we present a novel feature that allows users to co-localize samples, to nest a package within another package, and to identify replicates. Adoption of the MIxS-SA and of the new terms will facilitate reports of complex sampling design from a myriad of environments.

Microbiome Metadata Standards: Report of the National Microbiome Data Collaborative's Workshop and Follow-On Activities.

Microbiome samples are inherently defined by the environment in which they are found. Therefore, data that provide context and enable interpretation of measurements produced from biological samples, often referred to as metadata, are critical. Important contributions have been made in the development of community-driven metadata standards; however, these standards have not been uniformly embraced by the microbiome research community. To understand how these standards are being adopted, or the barriers to adoption, across research domains, institutions, and funding agencies, the National Microbiome Data Collaborative (NMDC) hosted a workshop in October 2019. This report provides a summary of discussions that took place throughout the workshop, as well as outcomes of the working groups initiated at the workshop.

Subsurface automated samplers for eDNA (SASe) for biological monitoring and research.

Sampling of environmental DNA (eDNA) in seawater is an increasingly common approach to non-invasively assess marine biodiversity, detect cryptic or invasive species, and monitor specific groups of organisms. Despite this remarkable utility, collection and filtration of eDNA samples in the field still requires considerable time and effort. Recent advancements in automated water samplers have standardized the eDNA collection process, allowing researchers to collect eDNA day or night, sample in locations that are difficult to access, and remove the need for highly trained personnel to perform sampling. However, the high cost of purchasing or building these samplers represents a financial hurdle to widespread application. To overcome this difficulty, we have designed and built a low-cost subsurface automated sampler for eDNA (SASe). Each sampler is submersible to 55 m, can filter a pre-programmable volume of water, and preserves eDNA at the site of collection. SASe samplers have replaceable filters and a low build cost (∼280 USD vs. >100,000 USD for other eDNA samplers), which facilitates repeated field sampling at fine spatial and temporal scales. Lab testing has shown the SASe to be as effective as a standard desktop peristaltic pump for sampling, preserving, and recovering marine eDNA. SASe design files and operating code are open-source, promoting the use of this tool to meet a range of future eDNA research applications, including project-specific customizations to the current design.

Two hundred and fifty-four metagenome-assembled bacterial genomes from the bank vole gut microbiota.

Vertebrate gut microbiota provide many essential services to their host. To better understand the diversity of such services provided by gut microbiota in wild rodents, we assembled metagenome shotgun sequence data from a small mammal, the bank vole Myodes glareolus (Rodentia, Cricetidae). We were able to identify 254 metagenome assembled genomes (MAGs) that were at least 50% (n = 133 MAGs), 80% (n = 77 MAGs) or 95% (n = 44 MAGs) complete. As typical for a rodent gut microbiota, these MAGs are dominated by taxa assigned to the phyla Bacteroidetes (n = 132 MAGs) and Firmicutes (n = 80), with some Spirochaetes (n = 15) and Proteobacteria (n = 11). Based on coverage over contigs, Bacteroidetes were estimated to be most abundant group, followed by Firmicutes, Spirochaetes and Proteobacteria. These draft bacterial genomes can be used freely to determine the likely functions of gut microbiota community composition in wild rodents.

MetaMiner: A Scalable Peptidogenomics Approach for Discovery of Ribosomal Peptide Natural Products with Blind Modifications from Microbial Communities.

Ribosomally synthesized and post-translationally modified peptides (RiPPs) are an important class of natural products that contain antibiotics and a variety of other bioactive compounds. The existing methods for discovery of RiPPs by combining genome mining and computational mass spectrometry are limited to discovering specific classes of RiPPs from small datasets, and these methods fail to handle unknown post-translational modifications. Here, we present MetaMiner, a software tool for addressing these challenges that is compatible with large-scale screening platforms for natural product discovery. After searching millions of spectra in the Global Natural Products Social (GNPS) molecular networking infrastructure against just eight genomic and metagenomic datasets, MetaMiner discovered 31 known and seven unknown RiPPs from diverse microbial communities, including human microbiome and lichen microbiome, and microorganisms isolated from the International Space Station.

Red Sea SAR11 and Prochlorococcus Single-Cell Genomes Reflect Globally Distributed Pangenomes.

Evidence suggests many marine bacteria are cosmopolitan, with widespread but sparse strains poised to seed abundant populations under conducive growth conditions. However, studies supporting this "microbial seed bank" hypothesis have analyzed taxonomic marker genes rather than whole genomes/metagenomes, leaving open the possibility that disparate ocean regions harbor endemic gene content. The Red Sea is isolated geographically from the rest of the ocean and has a combination of high irradiance, high temperature, and high salinity that is unique among the oceans; we therefore asked whether it harbors endemic gene content. We sequenced and assembled single-cell genomes of 21 SAR11 (subclades Ia, Ib, Id, and II) and 5 Prochlorococcus (ecotype HLII) samples from the Red Sea and combined them with globally sourced reference genomes to cluster genes into ortholog groups (OGs). Ordination of OG composition could distinguish clades, including phylogenetically cryptic Prochlorococcus ecotypes LLII and LLIII. Compared with reference genomes, 1% of Prochlorococcus and 17% of SAR11 OGs were unique to the Red Sea genomes (RS-OGs). Most (83%) RS-OGs had no annotated function, but 65% of RS-OGs were expressed in diel Red Sea metatranscriptomes, suggesting they are functional. Searching Tara Oceans metagenomes, RS-OGs were as likely to be found as non-RS-OGs; nevertheless, Red Sea and other warm samples could be distinguished from cooler samples using the relative abundances of OGs. The results suggest that the prevalence of OGs in these surface ocean bacteria is largely cosmopolitan, with differences in population metagenomes manifested by differences in relative abundance rather than complete presence/absence of OGs.IMPORTANCE Studies have shown that as we sequence seawater from a selected environment deeper and deeper, we approach finding every bacterial taxon known for the ocean as a whole. However, such studies have focused on taxonomic marker genes rather than on whole genomes, raising the possibility that the lack of endemism results from the method of investigation. We took a geographically isolated water body, the Red Sea, and sequenced single cells from it. We compared those single-cell genomes to available genomes from around the ocean and to ocean-spanning metagenomes. We showed that gene ortholog groups found in Red Sea genomes but not in other genomes are nevertheless common across global ocean metagenomes. These results suggest that Baas Becking's hypothesis "everything is everywhere, but the environment selects" also applies to gene ortholog groups. This widely dispersed functional diversity may give oceanic microbial communities the functional capacity to respond rapidly to changing conditions.

A Novel Sparse Compositional Technique Reveals Microbial Perturbations.

The central aims of many host or environmental microbiome studies are to elucidate factors associated with microbial community compositions and to relate microbial features to outcomes. However, these aims are often complicated by difficulties stemming from high-dimensionality, non-normality, sparsity, and the compositional nature of microbiome data sets. A key tool in microbiome analysis is beta diversity, defined by the distances between microbial samples. Many different distance metrics have been proposed, all with varying discriminatory power on data with differing characteristics. Here, we propose a compositional beta diversity metric rooted in a centered log-ratio transformation and matrix completion called robust Aitchison PCA. We demonstrate the benefits of compositional transformations upstream of beta diversity calculations through simulations. Additionally, we demonstrate improved effect size, classification accuracy, and robustness to sequencing depth over the current methods on several decreased sample subsets of real microbiome data sets. Finally, we highlight the ability of this new beta diversity metric to retain the feature loadings linked to sample ordinations revealing salient intercommunity niche feature importance. IMPORTANCE By accounting for the sparse compositional nature of microbiome data sets, robust Aitchison PCA can yield high discriminatory power and salient feature ranking between microbial niches. The software to perform this analysis is available under an open-source license and can be obtained at https://github.com/biocore/DEICODE; additionally, a QIIME 2 plugin is provided to perform this analysis at https://library.qiime2.org/plugins/deicode/.

Evolutionary trends in host physiology outweigh dietary niche in structuring primate gut microbiomes.

Over the past decade several studies have reported that the gut microbiomes of mammals with similar dietary niches exhibit similar compositional and functional traits. However, these studies rely heavily on samples from captive individuals and often confound host phylogeny, gut morphology, and diet. To more explicitly test the influence of host dietary niche on the mammalian gut microbiome we use 16S rRNA gene amplicon sequencing and shotgun metagenomics to compare the gut microbiota of 18 species of wild non-human primates classified as either folivores or closely related non-folivores, evenly distributed throughout the primate order and representing a range of gut morphological specializations. While folivory results in some convergent microbial traits, collectively we show that the influence of host phylogeny on both gut microbial composition and function is much stronger than that of host dietary niche. This pattern does not result from differences in host geographic location or actual dietary intake at the time of sampling, but instead appears to result from differences in host physiology. These findings indicate that mammalian gut microbiome plasticity in response to dietary shifts over both the lifespan of an individual host and the evolutionary history of a given host species is constrained by host physiological evolution. Therefore, the gut microbiome cannot be considered separately from host physiology when describing host nutritional strategies and the emergence of host dietary niches.

Species-level functional profiling of metagenomes and metatranscriptomes.

Functional profiles of microbial communities are typically generated using comprehensive metagenomic or metatranscriptomic sequence read searches, which are time-consuming, prone to spurious mapping, and often limited to community-level quantification. We developed HUMAnN2, a tiered search strategy that enables fast, accurate, and species-resolved functional profiling of host-associated and environmental communities. HUMAnN2 identifies a community's known species, aligns reads to their pangenomes, performs translated search on unclassified reads, and finally quantifies gene families and pathways. Relative to pure translated search, HUMAnN2 is faster and produces more accurate gene family profiles. We applied HUMAnN2 to study clinal variation in marine metabolism, ecological contribution patterns among human microbiome pathways, variation in species' genomic versus transcriptional contributions, and strain profiling. Further, we introduce 'contributional diversity' to explain patterns of ecological assembly across different microbial community types.

The genome of a novel isolate of Prochlorococcus from the Red Sea contains transcribed genes for compatible solute biosynthesis.

Marine microbes possess genomic and physiological adaptations to cope with varying environmental conditions. So far, the effects of high salinity on the most abundant marine photoautotrophic organism, Prochlorococcus, in marine oligotrophic environments, are mostly unknown. Here, we report the isolation of a new Prochlorococcus strain (RSP50) belonging to high-light (HL) clade II from the Red Sea, one of the warmest and most saline bodies of water in the global oceans. A comparative genomic analysis identified a set of 59 genes that were exclusive to RSP50 relative to currently available Prochlorococcus genomes, the majority of which (70%) encode for hypothetical proteins of unknown function. However, three of the unique genes encode for a complete pathway for the biosynthesis of the compatible solute glucosylglycerol, and are homologous to enzymes found in the sister lineage Synechococcus. Metatranscriptomic analyses of this metabolic pathway in the water column of the Red Sea revealed that the corresponding genes were constitutively transcribed, independent of depth and light, suggesting that osmoregulation using glucosylglycerol is a general feature of HL II Prochlorococcus in the Red Sea.

Environmental radiation alters the gut microbiome of the bank vole Myodes glareolus.

Gut microbiota composition depends on many factors, although the impact of environmental pollution is largely unknown. We used amplicon sequencing of bacterial 16S rRNA genes to quantify whether anthropogenic radionuclides at Chernobyl (Ukraine) impact the gut microbiome of the bank vole Myodes glareolus. Exposure to elevated levels of environmental radionuclides had no detectable effect on the gut community richness but was associated with an almost two-fold increase in the Firmicutes:Bacteroidetes ratio. Animals inhabiting uncontaminated areas had remarkably similar gut communities irrespective of their proximity to the nuclear power plant. Hence, samples could be classified to high-radiation or low-radiation sites based solely on microbial community with >90% accuracy. Radiation-associated bacteria had distinct inferred functional profiles, including pathways involved in degradation, assimilation and transport of carbohydrates, xenobiotics biodegradation, and DNA repair. Our results suggest that exposure to environmental radionuclides significantly alters vertebrate gut microbiota.

Best practices for analysing microbiomes.

Complex microbial communities shape the dynamics of various environments, ranging from the mammalian gastrointestinal tract to the soil. Advances in DNA sequencing technologies and data analysis have provided drastic improvements in microbiome analyses, for example, in taxonomic resolution, false discovery rate control and other properties, over earlier methods. In this Review, we discuss the best practices for performing a microbiome study, including experimental design, choice of molecular analysis technology, methods for data analysis and the integration of multiple omics data sets. We focus on recent findings that suggest that operational taxonomic unit-based analyses should be replaced with new methods that are based on exact sequence variants, methods for integrating metagenomic and metabolomic data, and issues surrounding compositional data analysis, where advances have been particularly rapid. We note that although some of these approaches are new, it is important to keep sight of the classic issues that arise during experimental design and relate to research reproducibility. We describe how keeping these issues in mind allows researchers to obtain more insight from their microbiome data sets.

Taxon-specific aerosolization of bacteria and viruses in an experimental ocean-atmosphere mesocosm.

Ocean-derived, airborne microbes play important roles in Earth's climate system and human health, yet little is known about factors controlling their transfer from the ocean to the atmosphere. Here, we study microbiomes of isolated sea spray aerosol (SSA) collected in a unique ocean-atmosphere facility and demonstrate taxon-specific aerosolization of bacteria and viruses. These trends are conserved within taxonomic orders and classes, and temporal variation in aerosolization is similarly shared by related taxa. We observe enhanced transfer into SSA of Actinobacteria, certain Gammaproteobacteria, and lipid-enveloped viruses; conversely, Flavobacteriia, some Alphaproteobacteria, and Caudovirales are generally under-represented in SSA. Viruses do not transfer to SSA as efficiently as bacteria. The enrichment of mycolic acid-coated Corynebacteriales and lipid-enveloped viruses (inferred from genomic comparisons) suggests that hydrophobic properties increase transport to the sea surface and SSA. Our results identify taxa relevant to atmospheric processes and a framework to further elucidate aerosolization mechanisms influencing microbial and viral transport pathways.