Sponsors

• Shion Hosoda, Waseda University, Japan
• Tsukasa Fukunaga, The University of Tokyo, Japan
• Michiaki Hamada, Waseda University, Japan

Motivation:
Accumulating evidence has highlighted the importance of microbial interaction networks. Methods have been developed for estimating microbial interaction networks,
of which the generalized Lotka-Volterra equation (gLVE)-based method can estimate a directed interaction network. The previous gLVE-based method for estimating microbial interaction networks did not consider time-varying interactions.
Results:
In this study, we developed unsupervised learning based microbial interaction inference method using Bayesian estimation (Umibato), a method for estimating time-varying microbial interactions. The Umibato algorithm comprises Gaussian process regression (GPR) and a new Bayesian probabilistic model, the continuous-time regression hidden Markov model (CTRHMM). Growth rates are estimated by GPR, and interaction networks are estimated by CTRHMM. CTRHMM can estimate time-varying interaction networks using interaction states, which are defined as hidden variables. Umibato outperformed the existing methods on synthetic datasets. In addition, it yielded reasonable estimations in experiments on a mouse gut microbiota dataset, thus providing novel insights into the relationship between consumed diets and the gut microbiota.

• Beatriz García-Jiménez, Biome Makers Inc., United States
• Rüdiger Ortiz-Álvarez, Biome Makers Inc., United States
• Héctor Ortega-Arranz, Biome Makers Inc., United States
• Vicente J. Ontiveros, Centre for Advanced Studies of Blanes (CEAB-CSIC), Spain
• Miguel de Celis, Complutense University of Madrid, Spain
• Charles Ravarani, Biome Makers Inc., United States
• Alberto Acedo, Biome Makers Inc., United States
• Ignacio Belda, Biome Makers Inc., United States

Motivation: Soil fungal communities play a key role in agroecosystem sustainability. Understanding the ecology behind the assembly and dynamics of soil microbiome is a fruitful way to improve management practices and plant productivity. Thus, monitoring soil health would benefit from the use of metrics that arise from ecological explanations. Beyond traditional biodiversity descriptors, network community-level properties have the potential of informing about particular ecological situations.

Results: We assess the impact of different farming practices in a survey of 350 vineyard soils from the United States and Spain by estimating microbiome network properties. Further than traditional methodologies in communities, we propose a novel analytical approach to infer ecological properties from local networks. We conclude that low-intervention practices (organic and biodynamic managements) promoted densely clustered networks, describing an equilibrium state. In contrast, conventionally managed vineyards had highly modular sparser communities. Thus, we hypothesize that network properties at the community level may help to understand how the environment and land use can affect community structure and ecological processes in agroecosystems. This methodological strategy could be widely applied to understand the effect of environmental disturbances in both natural and human-managed ecosystems and in other fields of interest such as food production or human health.

• Naveen Sivadasan, TCS Research, India
• Rajgopal Srinivasan, TCS Research, India
• Vidushi Walia, TCS Research, India

Metagenomic profiling from sequencing data aims to disentangle a microbial sample at the lower ranks of the taxonomy such as species and strains with precise measurement of abundances. State-of-the-art tools for metagenomic profiling often rely on homology based identification of marker sequences using highly complex and compute intensive methods. Such approaches often have difficulties in accurate strain level profiling due to limited read support. Furthermore, it makes incorporation of updates to the reference data expensive. We develop MAGE, which is a non homology based and alignment free strain level profiling tool for accurate abundance estimation. MAGE works with a k-mer based index of the reference data. MAGE uses a combination of the well-known FM index and R-index, together with an efficient meta index, for constructing a compact and comprehensive index. After read mapping, MAGE performs MLE based strain level relative abundance estimation. We performed several experiments for species/strain level profiling, where MAGE showed superior performance compared to the state-of-the-art on a wide range of performance measures, especially for strain level profiling.

• Aitor Blanco-Miguez, University of Trento, Italy
• Francesco Beghini, University of Trento, Italy
• Fabio Cumbo, University of Trento, Italy
• Moreno Zolfo, University of Trento, Italy
• Francesco Asnicar, University of Trento, Italy
• Paolo Manghi, University of Trento, Italy
• Leonard Dubois, University of Trento, Italy
• Kun D. Huang, University of Trento, Italy
• Gianmarco Piccinno, University of Trento, Italy
• Adrian Tett, University of Vienna, Austria
• Mireia Valles-Colomer, University of Trento, Italy
• Edoardo Pasolli, University of Naples, Italy
• Curtis Huttenhower, Chan School of Public Health, United States
• Nicola Segata, University of Trento, Italy

Metagenomic assembly is an effective tool for high-throughput novel organism discovery from communities. However, it can only accurately capture the most abundant organisms in complex communities. Here, we present one of the first methods to integrate both metagenomic assembly- and reference-based information for taxonomic profiling, MetaPhlAn 4beta. From a collection of 1.01M bacterial and archeal reference and metagenome-assembled genomes recovered from ~45k metagenomes, we defined unique marker genes for 26,970 species-level genome bins (SGBs), 4,992 of them taxonomically unidentified at the species level. MetaPhlAn 4beta doubles the number of species quantifiable in the human gut microbiome relative to previous versions and increases it by >10x in other less-characterized environments. Synthetic evaluations showed that MetaPhlAn is more accurate than available alternatives while reliably quantifying novel organisms represented as unknown SGBs (uSGBs). Large-scale profiling of ~20k metagenomes revealed that some such uSGBs are promising novel biomarkers for associations between microbiome and host diet, cardiometabolic markers, and gut-related disease. MetaPhlAn 4beta provides the first steps in integrated assembly- and reference-based metagenome analysis, provides efficient metagenomic profiling of uncharacterized species, and enables deeper and more comprehensive microbiome feature and biomarker detection.

• Milot Mirdita, Max Planck Institute for Biophysical Chemistry, Germany
• Martin Steinegger, Seoul National University, South Korea
• Florian Breitwieser, Johns Hopkins School of Medicine, United States
• Johannes Soeding, soeding@mpibpc.mpg.de, Germany
• Eli Levy Karin, Max Planck Institute for Biophysical Chemistry, Germany

Summary: MMseqs2 taxonomy is a new tool to assign taxonomic labels to metagenomic contigs. It extracts all pos- sible protein fragments from each contig, quickly retains those that can contribute to taxonomic annotation, assigns them with robust labels and determines the contig’s taxonomic identity by weighted voting. Its fragment extraction step is suitable for the analysis of all domains of life. MMseqs2 taxonomy is 2-18x faster than state-of-the-art tools and also contains new modules for creating and manipulating taxonomic reference databases as well as reporting and visualizing taxonomic assignments.
Availability and implementation: MMseqs2 taxonomy is part of the MMseqs2 free open-source software package available for Linux, macOS and Windows at https://mmseqs.com.

• Alice McHardy

• Mihai Pop

Determining the identity of an organism in a sample and determining the function of a gene within an organism are key steps in analytical pipelines used in both basic research and clinical applications. A range of computational techniques are used to perform these tasks, including: database searches, machine learning, protein structure prediction, etc. The majority of the approaches used today aim to provide a definitive answer, perhaps with an associated confidence estimate. In my talk, I will argue that it is often valuable to report a broader set of plausible answers, approach that can reveal information about the structure of reference databases and can provide a more nuanced analysis of sequences that have not been previously characterized. My presentation will focus on taxonomic annotation using our software Atlas that relies on a new approach for searching biological databases.

• Thilo Muth

Metaproteomics has substantially grown over the past years and supplements other omics approaches by bringing valuable functional information, enabling genotype- phenotype linkages and connections to metabolic outputs. Currently, a wide variety of metaproteomic workflows is available, yet their impact on the results remains to be thoroughly assessed.
Here, we carried out the first community-driven, multi-lab comparison in metaproteomics: the critical assessment of metaproteome investigation (CAMPI) study. Based on well-established workflows, we evaluated the influence of sample preparation, mass spectrometry acquisition, and bioinformatic analysis using two samples: a simplified, lab-assembled human intestinal model and a human fecal sample.
Although bioinformatic pipelines contributed to variability in peptide identification, wet-lab workflows were the most important source of differences between analyses. Overall, these peptide-level differences largely disappeared at the protein group level. Differences were observed between peptide- and protein-centric approaches for the predicted community composition but similar functional profiles were found across workflows.
The CAMPI findings demonstrate the robustness of current metaproteomics research and provide a perspective for future benchmarking studies.

• Wei Wei

• Alexander Sczyrba

• Yuzhen Ye

Advances of experimental and computational techniques have enabled the study of microbiomes (communities of microorganisms) that are related to almost every aspect of human beings. We have been developing new algorithms and computational tools for microbiome research, to address arising computational demands and challenges, and to make new use of microbiome data. In this talk, I will focus on our recently developed approaches for analyzing metaproteomic data and demonstrate their use in the inference of gut microbial signatures that are likely expressed and are predictive of host phenotypes, and for studying the proteome of human-associated bacterial species. In addition, I will share our research on the CRISPR–Cas adaptive immune systems, focusing on the tools we developed for discovery and characterization of CRISPR–Cas systems using microbiome data.

• Yancong Zhang, Harvard T. H. Chan School of Public Health, United States
• Kelsey Thompson, Harvard T. H. Chan School of Public Health, United States
• Curtis Huttenhower, Harvard T. H. Chan School of Public Health, United States
• Eric Franzosa, Harvard T. H. Chan School of Public Health, United States

Motivation: Metatranscriptomics (MTX) has become an increasingly practical way to profile the functional activity of microbial communities in situ. However, MTX remains underutilized due to experimental and computational limitations. The latter are complicated by non-independent changes in both RNA transcript levels and their underlying genomic DNA copies (as microbes simultaneously change their overall abundance in the population and regulate individual transcripts), genetic plasticity (as whole loci are frequently gained and lost in microbial lineages), and measurement compositionality and zero-inflation. Here, we present a systematic evaluation of and recommendations for differential expression (DE) analysis in MTX.

Results: We designed and assessed six statistical models for DE discovery in MTX that incorporate different combinations of DNA and RNA normalization and assumptions about the underlying changes of gene copies or species abundance within communities. We evaluated these models on multiple simulated and real multi-omic datasets. Models adjusting transcripts relative to their encoding gene copies as a covariate were significantly more accurate in identifying DE from MTX in both simulated and real datasets. Moreover, we show that when paired DNA measurements (metagenomic data, MGX) are not available, models normalizing MTX measurements within-species while also adjusting for total-species RNA balance sensitivity, specificity, and interpretability of DE detection, as does filtering likely technical zeros. The efficiency and accuracy of these models pave the way for more effective MTX-based DE discovery in microbial communities.

Availability: The analysis code and synthetic datasets used in this evaluation are available online at http://huttenhower.sph.harvard.edu/mtx2021.

• Jiayu Shang, City Univeristy of Hong Kong, Hong Kong
• Jingzhe Jiang, South China Sea Fisheries Research Institute, Chinese Academy of Fishery Sciences, Guangzhou, Guangdong Province, China, China
• Yanni Sun, City University of Hong Kong, Hong Kong

Motivation: Bacteriophages (aka phages), which mainly infect bacteria, play key roles in the biology of microbes. As the most abundant biological entities on the planet, the number of discovered phages is only the tip of the iceberg. Recently, many new phages have been revealed using high throughput sequencing, particularly metagenomic sequencing. Compared to the fast accumulation of phage-like sequences, there is a serious lag in the taxonomic classification of phages. High diversity, abundance, and limited known phages pose great challenges for taxonomic analysis. In particular, alignment-based tools have difficulty in classifying fast accumulating contigs assembled from metagenomic data.

Result: In this work, we present a novel semi-supervised learning model, named PhaGCN, to conduct taxonomic classification for phage contigs. In this learning model, we construct a knowledge graph by combining the DNA sequence features learned by convolutional neural network (CNN) and protein sequence similarity gained from gene-sharing network. Then we apply graph convolutional network (GCN) to utilize both the labeled and unlabeled samples in training to enhance the learning ability. We tested PhaGCN on both simulated and real sequencing data. The results clearly show that our method competes favorably against available phage classification tools.

• Ruoshi Zhang, Max Planck Institute for Biophysical Chemistry, Germany
• Milot Mirdita, Max Planck Institute for Biophysical Chemistry, Germany
• Eli Levy Karin, Max Planck Institute for Biophysical Chemistry, Germany
• Clovis Norroy, Max Planck Institute for Biophysical Chemistry, Germany
• Clovis Galiez, University of Grenoble Alpes, France
• Johannes Soeding, Max Planck Institute for Biophysical Chemistry, Germany

Summary: SpacePHARER (CRISPR Spacer Phage-Host Pair Finder) is a sensitive and fast tool for de novo prediction of phage–host relationships via identifying phage genomes that match CRISPR spacers in genomic or metagenomic data. SpacePHARER gains sensitivity by comparing spacers and phages at the protein level, optimizing its scores for matching very short sequences, and combining evidence from multiple matches, while controlling for false positives. We demonstrate that SpacePHARER has 1.4 to 4 times higher sensitivity than the state-of-the-art BLASTN-based method, while being around 47 times faster, by searching a comprehensive spacer list against all available complete phage genomes on NCBI. We hope that SpacePHARER will help uncovering many unknown phage-host relationships and facilitate further understanding of phage-host co-evolution.
Availability and implementation: SpacePHARER is available as an open-source (GPLv3), user-friendly command-line software for Linux and macOS: https://github.com/soedinglab/spacepharer.

• Kangjin Kim, School of Public Health, Seoul National University, Korea, South Korea
• Sungho Won, Seoul National University, South Korea
• Sanghun Lee, Institute, South Korea
• Sang-Chul Park, 3Brigham and Women's Hospital, Harvard Medical School, Boston, MA, USA, South Korea
• Nam-Eun Kim, Department of Public Health Science, Graduate School of Public Health, Seoul National University, Seoul, South Korea, South Korea
• Youngae Jung, Integrated Metabolomics Research Group, Western Seoul Center, Korea Basic Science Institute, Seoul, South Korea, South Korea
• Dankyu Yoon, Division of Allergy and Chronic Respiratory Diseases, Center for Biomedical Sciences, National, South Korea
• Hyeonjeong Kim, Korea Medical Institute, Seoul, South Korea, South Korea
• Sanghyun Kim, Korea Medical Institute, Seoul, South Korea, South Korea
• Geum-Sook Hwang, Integrated Metabolomics Research Group, Western Seoul Center, Korea Basic Science Institute, Seoul, South Korea, South Korea
• Chol Shin, 5Institute of Human Genomic Study, College of Medicine, Korea University Ansan Hospital, Ansan, South Korea, South Korea
• Seung Ku Lee, 5Institute of Human Genomic Study, College of Medicine, Korea University Ansan Hospital, Ansan, South Korea, South Korea

Background: We investigated the microorganisms potentially associated with T2D by tracking changes in microbiota-derived EV in urine samples collected three times. We also revealed how microbes and their metabolites interact with the human body and affect the development of T2D. Mendelian randomization analysis was conducted to evaluate the causal relationships among microbial organisms, metabolites, and clinical measures.
Results: We analyzed EV-derived metagenomic (N = 393), clinical (N = 5032), genomic (N = 8842), and metabolite (N = 574) data from a prospective and longitudinal Korean community-based cohort (KARE). Our data revealed that GU174097_g, an unclassified Lachnospiraceae, was associated with T2D (β = −189.13; p = 0.00006) and with the ketone bodies acetoacetate and 3-hydroxybutyrate (r = −0.0938 and −0.0829, respectively; p = 0.0022 and 0.0069, respectively), and we identified a causal relationship between acetoacetate and HbA1c levels (β = 0.0002; p = 0.0154). GU174097_g reduced the levels of ketone bodies, thus decreasing the HbA1c level and the risk of T2D.
Conclusions: Our findings indicate that GU174097_g may lower the risk of T2D by reducing the ketone body levels. This finding warrants verification through large-scale longitudinal studies and in-vivo and in-vitro studies.

• Eran Segal

Accumulating evidence supports a causal role for the human gut microbiome in obesity, diabetes, metabolic disorders, cardiovascular disease, and numerous other conditions. I will present our research on the role of the human microbiome in health and disease, ultimately aimed at developing personalized medicine approaches that combine human genetics, microbiome, and nutrition.
In one project, we tackled the subject of personalization of human nutrition, using a cohort of over 1,000 people in which we measured blood glucose response to >50,000 meals, lifestyle, medical and food frequency questionnaires, blood tests, genetics, and gut microbiome. We showed that blood glucose responses to meals greatly vary between people even when consuming identical foods; devised the first algorithm for accurately predicting personalized glucose responses to food based on clinical and microbiome data; and showed that personalized diets based on our algorithm successfully balanced blood glucose levels in prediabetic individuals.
Using the same cohort, we also studied the set of metabolites circulating in the human blood, termed the serum metabolome, which contain a plethora of biomarkers and causative agents. With the goal of identifying factors that determine levels of these metabolites, we devised machine learning algorithms that predict metabolite levels in held-out subjects. We show that a large number of these metabolites are significantly predicted by the microbiome and unravel specific bacteria that likely modulate particular metabolites. These findings pave the way towards microbiome-based therapeutics aimed at manipulating circulating metabolite levels for improved health.
Finally, I will present a novel metagenome-wide association study (MWAS) framework that we devised for analyzing the microbiome at the level of single nucleotide polymorphisms (SNPs). When applied to a large-scale cohort of ~50,000 microbiome samples, we identified ~100,000 statistically significant associations between a bacterial SNP and several traits such as BMI, diabetes, and acute coronary syndrome. MWAS benefits from the functional redundancy of orthologous genes across different bacterial species, as they can provide independent validation of SNP associations. Indeed, we found several such cases, such as 227 non-synonymous SNPs in orthologs for beta-galactosidase across dozens of bacterial species that were associated with BMI and other host phenotypes. Overall, we demonstrate that microbiome SNPs and host health have numerous robust associations with large effect sizes, paving the way towards a better understanding of the mechanisms underlying microbiome-disease associations.

• Sofia Forslund

Hypertension, as a component of the metabolic syndrome, is a risk factor for multiple further cardiovascular endpoints. Anecdotally, episodic fasting may improve components of the metabolic syndrome, and it is both seen in many traditional practices and likely to reflect living conditions throughout our evolutionary past, whereas it is absent from the "Western lifestyle". That also incorporates other dietary risk factors, many of which are reversed in the Dietary Approaches to Stop Hypertension (DASH) diet, one of few with substantial evidence backing it. However, efficacy of dietary interventions remains variable, and mechanisms underlying the anecdotal health benefits of fasting are underexplored.

Hypothesizing that 1) fasting operates in part through microbiome changes mediating immune system effects and 2) that episodic fasting can potentiate and prime for a shift to a DASH diet, thereby improving its efficacy, we recruited a cohort of metabolic syndrome patients randomly assigned to either begin a DASH diet following a 5-day Buchinger fast or to immediately begin a DASH diet. Probands were followed for three months under deep phenotyping, gut microbiome sequencing and immune cell cytometry, followed by bioinformatics analysis to account for changes in antihypertensive medication occurring during the study as proband health improved.

The analysis shows conclusively that episodic fasting induces pervasive microbiome and immune system shifts, which largely reverse after three months on a DASH diet, whereas a DASH diet alone has little impact, supporting a priming influence. Fasting probands substantially reduced blood pressure, need for antihypertensive medication, or both. Long-term responders also show more persistent changes reflecting bloom of gut bacteria with potential to produce anti-inflammatory short-chain fatty acids either during fasting or during refeeding, with concomitant immune system changes.

Responders are possible to identify already pre-intervention through machine learning, and are characterized pre-intervention by reduced abundance of these commensal bacteria, suggesting nutritional interventions can be personalized for maximal efficacy through identifying those with microbiotal and immune deficits the intervention can ameliorate. The DZHK researchers further validated the effects of fasting in a healthy cohort previously described, and will move forward with fasting intervention studies in cohorts for other health conditions.

• Dario Pescini, Department of Statistics and Quantitative Methods, University of Milano-Bicocca, Milan (Italy), Italy
• Giulia Agostinetto, Department of Biotechnology and Biosciences, University of Milano-Bicocca, Milan (Italy), Italy
• Davide Bozzi, Department of Biotechnology and Biosciences, University of Milano-Bicocca, Milan (Italy), Italy
• Massimo Labra, Department of Biotechnology and Biosciences, University of Milano-Bicocca, Milan (Italy), Italy
• Maurizio Casiraghi, Department of Biotechnology and Biosciences, University of Milano-Bicocca, Milan (Italy), Italy
• Antonia Bruno, Department of Biotechnology and Biosciences, University of Milano-Bicocca, Milan (Italy), Italy

Directly in contact with the environment, the skin microbiome is an intricate ecosystem that interacts with both the host and its surroundings, revealing its double-edged function of barrier and modulator of skin. Thanks to sequencing technology advances, initiatives such as Human Microbiome and Earth Microbiome Projects favored the collection of skin microbiome data. However, urges the need for data accessibility and reusability, according to FAIR (Findable, Accessible, Interoperable, and Reusable) principles, as supported by National Microbiome Data Collaborative and FAIR Microbiome.

To tackle the challenge of accelerating discovery and advances in skin microbiome research, we collected, integrated, organized, harmonized, and analyzed existing microbiome data resources from skin biome, categorized considering research goals. In parallel, we evaluated the set of skin data available in the most used public databases, identifying the main characteristics of data collection type, marker region used, sequencing approach and post-processing analysis (i.e. clustering method, reference databases).

Our work resulted in a state-of-the-art assessment focused on human skin microbiome data and projects, with special attention on data collection, generation and analysis. Further, we generated an enriched collection of harmonized datasets ready to be integrated into microbiome projects and advanced post-processing analysis, such as machine learning strategies.

• David Pellow, Tel Aviv University, Israel
• Alvah Zorea, Ben Gurion University of the Negev, Israel
• Maraike Probst, University of Innsbruck, Austria
• Ori Furman, Ben Gurion University, Israel
• Arik Segal, Soroka University Medical Center, Israel
• Itzhak Mizrahi, Ben Gurion University, Israel
• Ron Shamir, Tel Aviv University, Israel

Background: Metagenomic sequencing has led to the identification and assembly of many new bacterial genome sequences. These bacteria often contain plasmids: usually small, circular double-stranded DNA molecules that may transfer across bacterial species and confer antibiotic resistance and are less studied and understood. Part of the reason for this is the lack of computational tools enabling the analysis of plasmids in metagenomic samples.

Results: In order to assist in the study of plasmids we developed SCAPP (Sequence Contents-Aware Plasmid Peeler) - an algorithm and tool to assemble plasmid sequences from metagenomic sequencing. We compared the performance of SCAPP to Recycler and metaplasmidSPAdes on simulated metagenomes, real human gut microbiome samples, and a human gut plasmidome dataset that we generated. We also created plasmidome and metagenome data from the same cow rumen sample and used the parallel sequencing data to create a novel assessment procedure. In almost all cases SCAPP outperformed Recycler and metaplasmidSPAdes across this wide range of datasets.

Availability: SCAPP is an easy to use Python package that enables the assembly of full plasmid sequences from metagenomic samples. SCAPP is open-source software available from: https://github.com/Shamir-Lab/SCAPP.

This study is in press in BMC Microbiome.

• Peter Karp, SRI International, United States
• Suzanne Paley, SRI International, United States
• Markus Krummenacker, SRI International, United States

Pathway Tools (PTools) provides a large suite of capabilities for storing and analyzing integrated collections of genomic, metabolic, and regulatory information in the form of organism-specific Pathway/Genome Databases (PGDBs). A microbial community is represented in PTools by transforming each metagenome-assembled genome (MAG) into a PGDB. PTools can compute a metabolic reconstruction for each organism, and predict its operons. The properties of individual MAGs can be investigated using the many search and visualization operations within PTools. PTools also enables the user to investigate the properties of the microbial community by issuing searches across the full community, by visualizing and exploring a metabolic network diagram for the community, by overlaying community omics datasets on that network diagram, and by performing comparative operations across genome and pathway information. PTools also provides a tool for searching for metabolic transformation routes across an organism community.

• Donovan Parks

A fundamental question in prokaryotic microbiology is whether species exist and, if so, what criteria should be used to delineate species. Placing bounds on the genomic similarity of bacterial and archaeal strains is a practical method for delineating species that is largely congruent with existing species classifications defined using a polyphasic approach. Here, I discuss recent evidence suggesting a genetic discontinuity that would provide support for using genomic similarity for defining prokaryotic species, but argue that the evidence for this discontinuity is weak. Given the lack of strong evidence for discrete species clusters, I will discuss the use of genomic similarity as a pragmatic choice for delineating species within the Genome Taxonomy Database (GTDB) and highlight the benefits and challenges of this approach to taxonomic classification. I conclude with a proposal to reclassify Shigella species as later heterotypic synonyms of Escherichia coli within the GTDB as this classification is most congruent with the evolutionary relationship of these strains and the species definition adopted by the GTDB.