Despite advances in sequencing technologies and computational methods in the past decade, researchers have uncovered genomes for just a small fraction of Earth’s microbial diversity. Because most microbes cannot be cultivated under laboratory conditions, their genomes can't be sequenced using traditional approaches. Identifying and characterizing the planet’s microbial diversity is key to understanding the roles of microorganisms in regulating nutrient cycles, as well as gaining insights into potential applications they may have in a wide range of research fields.

A public repository of 52,515 microbial draft genomes generated from environmental samples around the world, expanding the known diversity of bacteria and archaea by 44%, is now available and described November 9, 2020 in Nature Biotechnology. Known as the GEM (Genomes from Earth’s Microbiomes) catalog, this work results from a collaboration involving more than 200 scientists, researchers at the U.S.  Department of Energy (DOE) Joint Genome Institute (JGI), a DOE Office of Science  User Facility located at Lawrence Berkeley National Laboratory (Berkeley Lab), and the  DOE Systems Biology Knowledgebase (KBase).

Metagenomics is the study of the microbial communities in the environmental samples without needing to isolate individual organisms, using various methods for processing,  sequencing and analysis. "Using a technique called metagenome binning, we were able  to reconstruct thousands of metagenome-assembled genomes (MAGs) directly from  sequenced environmental samples without needing to cultivate the microbes in the lab,"  noted Stephen Nayfach, the study's first author and research scientist in Nikos Kyrpides' Microbiome Data Science group. “What makes this study really stand out from previous  efforts is the remarkable environmental diversity of the samples we analyzed.” 

Emiley Eloe-Fadrosh, head of the JGI Metagenome Program and senior author on the  study elaborated on Nayfach’s comments. “This study was designed to encompass the  broadest and most diverse range of samples and environments, including natural and  agricultural soils, human- and animal-host associated, and ocean and other aquatic  environments - that's pretty remarkable.” 

Adding Value Beyond Genome Sequences

Much of the data had been generated from environmental samples sequenced by the  JGI through the Community Science Program and was already available on the JGI’s  Integrated Microbial Genomes & Microbiomes (IMG/M) platform. Eloe-Fadrosh noted  that it was a nice example of “big-data” mining to gain a deeper understanding of the 

data and enhancing the value by making data publicly available. 

To acknowledge the efforts of the investigators who had done the sampling, Eloe Fadrosh reached out to more than 200 researchers around the world in accordance with the JGI data use policy. “I felt it is important to acknowledge the significant efforts to  collect and extract DNA from these samples, many of which come from unique, difficult  to access environments, and invited these researchers to be co-authors as part of IMG  data consortium,” she said. 

Using this massive dataset, Nayfach clustered the MAGs into 18,000 candidate species  groups, 70% of which were novel compared over 500,000 existing genomes available at  that time. "Looking across the tree of life, it's striking how many uncultivated lineages  are only represented by MAGs,” he said. “While these draft genomes are imperfect,  they can still reveal a lot about the biology and diversity of uncultured microbes."

Teams of researchers worked on multiple analyses harnessing the genome repository,  and the IMG/M team developed several updates and features to mine the GEM catalog. One group mined the  dataset for novel secondary metabolites of secondary metabolite biosynthetic gene  clusters (BGCs), increasing these BGCs in IMG/ABC (Atlas of Biosynthetic Gene  Clusters) by 31%. Nayfach also worked with another team on predicting host-virus  connections between all viruses in IMG/VR (Virus) and the GEM catalog, associating  81,000 viruses – 70% of which had not already been associated with a host - with  23,000 MAGs. 

Modeling A New Path for Metagenomics Researchers

Building upon these resources, KBase, a multi-institutional collaborative knowledge  creation and discovery environment designed for biologists and bioinformaticians, developed metabolic models for thousands of MAGs. The models are now available in a  public Narrative, which provides shareable, reproducible workflows. “Metabolic  modeling is a routine analysis for isolate genomes, but has not been done at scale for  uncultivated microbes,” said Eloe-Fadrosh, “and we felt that the collaboration with  KBase would add value beyond clustering and analysis of these MAGs.” 

“Just bringing this dataset into KBase has immediate value because people can find the high-quality MAGs and use them to inform future analyses,” said José P. Faria, a KBase  computational biologist at Argonne National Laboratory. “The process of building a  metabolic model is simple: you just select a genome or MAG and press a button to build  a model from our database of mappings between biochemical reactions and annotations. We look at what was annotated in the genome and at the resulting model  to assess the metabolic capabilities of the organism.” 

KBase User Engagement lead Elisha Wood-Charlson added that by demonstrating the  ease with which metabolic models were generated from the GEM dataset,  metagenomics researchers might consider branching into this space. “Most  metagenomics researchers might not be willing to dive into an entirely new research  field [metabolic modeling], but they might be interested in how biochemistry impacts  what they work on. The genomics community can now explore metabolism using  KBase’s easy path from genomes or MAGs to modeling that may not have been  considered,” she said. 

A Community Resource for Facilitating Research

Kostas Konstantinidis of Georgia Institute of Technology, one of the co-authors whose  data were part of the catalog, “I don't think there are many institutions that can do this  kind of large-scale metagenomics and that have the capacity for large scale analyses. 

The beauty of this study is that it’s done at this scale that individual labs cannot do, and  it gives us new insights into microbial diversity and function.” 

He is already finding ways to utilize the catalog in his own research on how microbes  respond to climate change. “With this dataset I can see where every microbe is found,  and how abundant it is. That's very useful for my work and for others doing similar  research.” Additionally, he’s interested in expanding the diversity of the reference  database he’s developing called the Microbial Genomes Atlas to allow for more robust  analyses by adding the MAGs. 

“This is a great resource for the community,” Konstantinidis added. “It’s a dataset that is  going to facilitate many more studies subsequently. And I hope JGI and other  institutions continue to do this kind of projects.”

The work also used resources of the National Energy Research Scientific Computing  Center (NERSC), another DOE Office of Science User Facility located at Berkeley Lab.

Publication: Nayfach S et al. A Genomic Catalog of Earth’s Microbiomes. Nature  Biotechnology. 2020 Nov 9.