香京julia种子在线播放

The EBP was envisaged as a hubs-and-spokes organization of regional nodes and taxon-focused projects (1). The human division of the planet into nation-states does not overlap with the ecosystems, biomes, and bioregions that pattern biodiversity. Stewardship of biodiversity is similarly localized, and individuals and groups, including Indigenous peoples and local communities, have a deep local understanding of species diversity (5). The EBP will have the greatest impact if we build on these strong, local foundations. Here, we present a model for EBP regional nodes, based on building autonomous capacity for genomics, from sample acquisition to genome analysis. We emphasize that, in addition to collecting locally, we envision regional nodes that will also sequence, assemble, and analyze locally.

The throughput required to meet EBP Phase II goals could be delivered by 25 regional nodes, each collecting an average of 12,000 species and sequencing and assembling at least 6,000 species over a 4-year period. The inception of regional nodes will be driven by local initiative, availability of funding, and assessment of accessible biodiversity. Regional nodes will build on existing local scientific collaborations and knowledge. Sustainable regional nodes will require local skills, capacity, and funding (see the Workforce section below) and will likely take 2 to 3 years to implement.

Sample acquisition relies heavily on human capital and local skills. In contrast to expected savings in sequencing and assembly, as harder-to-source species are targeted, the costs of collecting will be relatively static per species. We envisage only a 40% reduction between Phase I and Phase II, even with the implementation of novel technologies. Much of the required expertise resides in local learned societies, taxon interest groups, national and local biological collections, and Indigenous peoples and local communities. We propose formal recruitment of collector allies to each regional node, who will bring specific taxonomic or habitat expertise and local user-community agendas. To promote sustainable careers, allies could agree to provide specified sets of legally and ethically sourced species, receive guaranteed compensation to recover staff and other costs, and be awarded explicit scientific credit for their work. Species acquisition for EBP sequencing through allies will support the currently underfunded expert taxonomy community, build capacity, and promote engagement with conservationists and other practitioners with the goals and outputs of the EBP.

It is essential that regional nodes should be established in biodiverse regions, especially in the low- and middle-income nations of the Global South, which have historically been underrepresented in or excluded from the global scientific commons. New regional nodes should be strongly supported by existing biodiversity genomics centers. Building on existing installed capacity and interest, this will establish a legacy of genomics expertise that can be leveraged for a range of post-genomic work, including FIF projects. One way such capacity could be achieved is through the installation of a complete “genomes from a box” (gBox), specimen-to-sequence laboratory (see Box 1 ), equipped for EBP data production at scale. A gBox install would be accompanied by support from other established nodes through a system of mutual aid and buy-in from technology companies for reagents and support.

For Phase III, both of these models (regional nodes and biodiversity genomics allies) will have to be expanded to ensure collection from all biomes in an inclusive, just, and ethical manner. Regional nodes will serve as focal points to usher in a post-EBP world of genome-enabled science for conservation, medicine, and bioindustry.

EBP specimens contributing to reference genome assemblies should be accompanied by rich metadata. From sample acquisition to genome publication, the EBP will enforce use of the GBIF Darwin Core standards, which define information that must accompany any globally aggregated biodiversity data record (https://www.gbif.org/standards) (39). This will be coordinated via Darwin Core-compatible metadata management systems such as Symbiota (40). The EBP will redouble efforts to make specimen metadata, genomic data, and analyses compatible with the (sometimes conflicting) demands of the FAIR [Findable, Accessible, Interoperable, and Reusable (41)] and CARE (Collective benefit, Authority to control, Responsible, Ethical) (https://www.gida-global.org/care) (42) principles of data governance. Wherever possible, a Traditional Knowledge and Biocultural Label or Notice (https://localcontexts.org/) should be attached.

Methods for high-quality, three-dimensional (3D) imaging of specimens compatible with the use of specimens for genomics are sorely needed. These images would provide an essential digital voucher for specimens subsequently consumed for sequencing. Imaging will be critical as the project progresses, as confidence in species identity may be lower, and sequencing may occur before full taxonomic identification. EBP specimen images could contribute to training resources for artificial intelligence (AI) and other computer-aided species identifications (28, 43, 44). The development of open-data systems and smartphone applications, similar to or allied with the popular iNaturalist platform (https://www.inaturalist.org/), would provide significant benefits.

Biobanking is critical for storing materials for future analyses. Detailed recommendations and protocols for biobanking in the age of genomics are available (45). Regional nodes must put in place secure biobanking of samples, ideally in collaboration with national or regional museums, botanical gardens, and other collections. Expansion of biobanking to explicitly support additional modes of analysis, such as proteomics, metabolomics, and single-cell atlasing, is recommended. Whenever possible, and especially for endangered species, cell lines should be created for future conservation efforts (46, 47).

Currently, DNA and RNA extractions maximally compatible with high-quality genomics are achieved from fresh or ultra-low-temperature flash-frozen material. Best practices for sample acquisition and shipping currently rely on live transport or an unbroken cold chain from collection to extraction. These practices are unsustainable on a global scale based on logistical, welfare, cost, and environmental grounds. Approaches that preserve specimens at ambient temperatures will be game-changing in terms of expanding sample collection. These are being explored with some success (48) but remain an urgent development area.

Delivery of EBP goals will require contemporaneous processing of several thousands of species in a single laboratory. Robust laboratory information management systems and electronic lab notebooks are essential. Live aggregation of sample process data from these tools in a workspace such as GoaT (24) would enhance shared learning of best practices. EBP members are already using open platforms to share best practices for collection, storage, and extraction (e.g., through protocols.io; see https://www.protocols.io/workspaces/earth-biogenome-project). EBP Phase II partners can enhance the content of these platforms with protocols modified to work at scale across diverse species.

Following the lead of the VGP (16), the EBP has established exacting but achievable quality metrics for reference assemblies (https://www.earthbiogenome.org/report-on-assembly-standards). We note that it is currently not technically possible to generate assemblies that meet these metrics for some species, largely because of small organism size and, consequently, minimal yield of DNA. Assemblies of such species will be attempted and submitted to the public databases, aiming to meet the EBP representative metric (0.1 Mb contig N50, chromosomal level, and 6.C.Q40). On the other end of the quality spectrum, we expect complete and near error-free, i.e., telomere-to-telomere, assemblies for a growing number of species, where contig N50 is the same as chromosomal N50, and all chromosomes are complete (C.C.Q40) (16, 52, 53).

The current recipes for genome sequencing to meet EBP metrics involve a mix of cutting-edge technologies. Three data types are currently used: single-molecule long-read data for contig building, long-range data from Hi-C for scaffolding, and transcriptomic data for accurate annotation. Close collaboration with technology providers will be essential to generate these data types at reduced per-genome costs ( Table 1 , see section, Costing the new Phase II strategy ). These savings must be available worldwide and include equity-based price reductions.

In the future, it may be possible to simplify sequencing so that high-quality genome assemblies can be generated from a single data type, and workflows can be simplified by running long-read, long-range, and transcriptome libraries together on a single platform. Applying this paradigm (one sample, one library, one run, one genome) across biodiversity would put the EBP in a very strong position to deliver Phase III genomes to reference quality. New data types may also prove to be useful. It is already clear that ultralong reads (>100 kb) can be used to deliver much more contiguous, true telomere-to-telomere assemblies (54, 55). The generation of such data for a significant fraction of EBP target species would elevate the quality and value of the genomes produced.

Sequencing and assembly procedures for most taxa are robust and ready for Phase II implementation (16, 26, 56), but challenges remain. We estimate that for about half of extant species, less than 1 ng of DNA can be isolated from a single specimen, orders of magnitude less than the input requirements for many current long-read processes. We need to develop robust, transferable protocols that generate genomic sequencing libraries from minuscule inputs without compromising assembly quality. Some success in this area has already been reported (8, 49, 50). More challenging still is the sequencing of single-celled eukaryotes, including the paraphyletic “protists” and some fungi, several of which have surprisingly large genomes. Some will be sequenced from clonal cultures, but most species are not in culture. EBP-standard genome sequencing of single cells from environmental sources is currently very challenging. Approaches that combine bulk and single-cell data are promising (57–60) but need further development. EBP-affiliated projects are also exploring the problems presented by polyploid genomes, where varying levels of rediploidization make the assembly of distinct sets of homeologous chromosomes difficult (52, 61).

Another issue that impacts biodiversity genomics based on sampling from the wild is that target organisms may be accompanied by mutualist or parasitic symbionts, components of the host microbiome, or by accidentally co-isolated organisms (51, 62). Robust separation of the genomes of these cobionts from that of the target species is essential to avoid misattribution of biological capacity. This work has been facilitated by the recent development of highly sensitive and specific decontamination workflows such as Foreign Contamination Screen (FCS)-GX (63). Nevertheless, close attention to this aspect is essential for the future.

Informatics workflows covering primary assembly, haplotypic duplication removal, scaffolding, decontamination, and pre-curation processing, readying for EBP Phase II, are already openly available in workflow management systems such as Galaxy (64) and Nextflow nf-core (e.g., https://pipelines.tol.sanger.ac.uk/pipelines). The use of AI and ML toolkits to intelligently automate decision-making processes, such as raw data quality control and assembly curation, will make the flow of genomes more efficient and improve output genome quality. Equally important will be the open sharing of process and quality control information, so that issues can be foregrounded and solutions found rapidly for all EBP nodes.

Delivery of these advances will require extensive, focused research and development in the academic and commercial sectors to develop better methods of acquisition and shipping of specimens, extraction of nucleic acids, sequencing, and assembly. We have included an estimate of US$100M for these activities in each of Phases II and III.

The minimal annotation product for every EBP species should be the annotation of protein-coding and conserved non-coding gene types (https://www.earthbiogenome.org/report-on-annotation-standards) (6), accompanied by repeat finding using curated repeat libraries and de novo discovery. Many transcription units, perhaps most in lineages such as vertebrates, give rise to more than one mature transcript, and defining the diversity of these isoforms is essential to unpicking the true diversity of genes. Currently, all annotation approaches rely fundamentally on alignment to the genome of transcriptome or protein data, and statistical models of genomic features. For EBP Phase II, the generation of at least 50 million read pairs of short-read transcriptomic data from a single library for every species is the absolute minimum for high-quality annotation. Long-read transcriptomic data are an attractive alternative to standard short-read RNA sequencing as they robustly reveal the diversity of transcript isoforms (74), but are currently expensive compared to short-read data. Concatenation sequencing of full-length complementary DNAs (cDNAs) on the Pacific BioSciences (PacBio) high-fidelity (HiFi) long-read sequencing platform promises to deliver sequence reads for annotation at a reasonable cost (75), and transcript normalization may maximize the utility of these data for annotation. Development and community benchmarking of these and other techniques may usher in an expectation of long-read transcriptomics as standard. High-quality annotation, such as that currently built for “model” species, relies on transcriptome data covering multiple tissues, developmental stages and conditions, and additional functional genomic data. This is unlikely to be achievable for most species, where developmental stages are not collected and dissection into tissue types is impractical. It will be important to build tools that recognize the diversity of alternative splicing, perhaps using information from related species.

Cells and organisms read their DNA code using a complex mix of sequence-based and epigenetic signals. The development of AI toolkits to predict genes and their likely activity will need significant functional data beyond just deep sampling of mature messenger RNA (mRNA) transcripts. The EBP encourages the generation of additional modalities of functional genomic data for which high-throughput methodologies are already available (such as sequencing non-polyadenylated RNAs and small RNAs, identifying transcription start sites, determining cytosine and adenine DNA methylation, mapping open chromatin, and defining the patterns of chromatin histone modification), especially for species representative of families. High-quality annotations driven by rich multimodal datasets for a diverse set of carefully chosen species across biodiversity would provide a platform for new comparative approaches that leverage whole-genome alignments between well-annotated and newly sequenced species to simultaneously annotate coding genes and infer orthologous gene loci (72). AI offers an emergent opportunity to achieve high-quality annotation, particularly of protein-coding loci, by “learning” the embedded transcriptional code from well-studied taxa (73). Sharing existing and new high-quality, curated, and transcriptome-validated gene predictions as dense training data for deep learning approaches will be imperative to promote these developments.

Annotation of EBP genomes should aim to meet FAIR and CARE principles and be made publicly available through submission to INSDC databases. FAIR principles demand that annotations should be accompanied by defined metadata, including methods (software tools and parameters), external data used, and agreed-upon quality metrics. Tools such as Benchmarking Using Single Copy Orthologs (BUSCO) (76), compleasm (77), and OMArk (78, 79) that exploit the expectation of the presence of a curated set of single-copy orthologs to assess coding-gene annotation completeness will need to be dynamically updated to maintain precision as Phase II ramps up. Additional quality assessment metrics require development, such as descriptors of ancestral linkage group retention, gene structure congruence, the proportion of genes with transcriptome support, and the number of proteins containing known domains. A standardized tool for multidimensional metric computation from a genome and associated annotation files would contribute to streamlining the entire process of producing reference-quality genomes across the eukaryotic tree of life.

For comparative genomics analyses, the EBP will have to ensure that products derived from genome sequencing are available for open use. Products will include large-scale whole genome alignments, ancestral linkage group inference (10, 80, 81), repeat and mobile element family data aggregated across species (82), up-to-date and comprehensive gene orthology calls (78), genome-anchored descriptions of conserved functional elements (83–85), and genome-wide description of the 3D structure of each genome (86). To generate these at the new scale of Phase II will require active development of the toolkits used, many of which can currently only scale to tens or hundreds of genomes. The EBP must foster and, where possible, sponsor the exploration of new algorithms and computer architecture for comparative analyses, promoting the inclusion of all data-generating communities and nations.

Globally accessible resources are particularly important for species and ecosystems under threat, where targeted investment in additional data, such as population genomics data, will be crucial in estimating extinction risk, managing wild populations, and understanding the genetic underpinnings of adaptation to local environments. For individual species, expanding population genetic approaches to the whole genome will reveal large-scale structure and illuminate critical details of population interconnectedness (9). At the landscape and ecosystem levels, the availability of complete genomes for many species in an ecosystem will make approaches for investigating species presence, species interaction, or functional capacity using environmentally sourced DNA (eDNA) from water (87), sediments (88), or the atmosphere (89). These approaches will rely on the ability to map eDNA reads to large databases of well-annotated sequences. Metagenomic approaches to exploring biotic diversity diversity, such as the Tara Oceans initiative (90, 91), will be transformed by EBP genome data, again via large-scale read mapping to rich, open datasets. The EBP should promote significant pilot projects that explore the use of genomic resources in assessing the diversity, functional capacity, and temporal dynamics of selected ecosystems in “genomic observatories”, preferably in the Global South, through the FIF.

The Encyclopedia of DNA Elements (ENCODE) project has shown the power of deep, coordinated multimodal genomic assays in discovering the function and regulation of genomes in humans and model species (92–94). Similar deep-dive functional genomics analyses of species selected for their phylogenetic disparity or for their potential to illuminate particular evolutionary transitions would be very powerful. We can imagine additional species being selected for a Diversity ENCODE program, developing new approaches to permit multimodal data generation from diverse systems.

Similarly, cell atlasing projects focused on humans and other model species have illuminated the cellular diversity of tissues and systems, revealing the genetic underpinning of complex traits such as immunity and development (95, 96). The Biodiversity Cell Atlas (BCA) initiative proposes expanding the list of species assayed at the single-cell level to explore the dynamics of development and environmental response across eukaryotes (97). The BCA will coordinate with the EBP to sample across diversity to illuminate the evolution and diversification of cell types across life and, in turn, BCA data will enrich genome annotation in targeted species and their relatives, for example, by enhancing understanding of co-expression networks and of the links between the diversity of genes present in a genome and species phenotypes.

The EBP must promote consistent frameworks for collecting and accessing metadata, the information needed to track provenance, attribution, data processing activity, and public distribution, and to integrate EBP activity with global biodiversity and sequence information data services. The EBP has already taken concrete steps in planning for meeting Phase II requirements (see https://www.earthbiogenome.org/it-and-informatics-standards). Close coordination with INSDC members will be necessary to manage the archiving and sharing of public raw and analyzed data, and the EBP will promote the use of INSDC databases for all outputs. Metadata for public release must be readily combinable and interoperable, based on defined ontologies accepted in the field, and accessible through application programming interfaces. For example, information about the genetic code likely to be used by a species is needed to establish parameters for genome annotation, and sample descriptors including Darwin Core-compatible collection location, time, voucher identifiers, collector, and provenance are needed when submitting the sequence data to public repositories and when using the reference genome in the context of time- and space-resolved population datasets.

In the fast-moving EBP Phase II (over 3,000 genomes released per month, with 30,000 species “in flight” at any one time), real-time, trusted data sharing and integration will be critical. In a global project with overlapping ownership, jurisdictions, and interests, the EBP will strive to ensure the highest standards of explicit sharing of ongoing data generation. For example, GenomeArk (https://www.genomeark.org/) has been built to provide pre-publication access to and sharing of high-quality reference genomes. GoaT (https://goat.genomehubs.org) offers an integrative view of the EBP and affiliated project activities. Data partnerships among research institutions, governments, funding agencies, and the private sector will be needed to ensure the EBP delivers to its full potential. We envision long-term thinking, coordinated action, and committed funding to ensure that data sources created by the EBP will be a lasting legacy.

The complexity of integrative analysis across thousands of genomes generally scales with the square of the number of species analyzed. The cost of computation (98–100) and consideration of the implied carbon footprint of the EBP favor approaches that generate shared analytic products for wide reuse. The EBP will work toward a “compute once, reuse many” approach, where core analytic products are precomputed for all to reuse. For example, whole-genome, reference-free alignment (101) is costly, with final products best shared rather than regenerated. Similarly, phylogenetic analyses of species and genes requires significant computation, and dynamically updated phylogeny and gene orthology assignments can be generated once and reused many times (102). Refactoring algorithms to support incremental updates when new species' genomes are released—rather than re-running full analyses—can avoid costly whole-dataset recomputations. For example, in phylogenetics, heuristic placement of new taxa updates trees without recalculating from scratch (103).

Workflow management systems are critical to ensure the highest quality of data products, improve automation and scaling, reduce costs and carbon footprint, and meet FAIR and CARE principles. We envisage shared development of open resources widely distributed through workflow hubs. It is clear that AI methods will become widespread in the coming decade and, for the EBP, immediate applications in data tracking and annotation are evident. However, AI is expensive (104). We will need to ensure that EBP data are maximally AI-ready on deposition by providing detailed metadata and extensive quality-controlled training sets.

EBP projects will need to pay close attention to cybersecurity best practices in software, workflows, data storage, and management to protect data integrity and data privacy (e.g., under the Nagoya Protocol). Ensuring equitable access to all EBP data effectively requires access to global or regional cloud-based storage. Utilizing these resources effectively means that the EBP will have to maximize the compression of raw and analyzed data while ensuring carbon-neutral operation of the chosen data storage providers. Redesigning the EBP informatics workflow to minimize carbon emissions, and partnering with vendors and facilities that demonstrably reduce and offset CO₂ from storage and computing, will lower the project's projected carbon footprint (105).

Obtaining the US$3.9 billion required to collect, sequence, and annotate 1.67 million eukaryotic species is a considerable challenge. Phase I—sequencing 10,000 species—is decentralized to individual affiliated projects, and the larger projects have raised upwards of US$200 million. Securing funding for the completion of Phase I and initiating Phase II is a high priority and has almost been realized. Successes to date have leveraged the vision of the global project to generate considerable enthusiasm from public and private funding sources, and the EBP is pursuing multiple strategies to achieve funding goals. Ideally, the EBP requires pooled funding from multiple geographic regions to deliver improved coordination and outreach, thereby maximizing scientific and societal benefits. We recognize that funding may come with important stipulations, such as open and free access to data, considerations relating to intellectual property, benefit-sharing, capacity development and building, and partnership with Indigenous peoples and local communities, that may complicate achieving the project’s goals.

Attractive possibilities for funding include pre-competitive consortia as well as pooling resources from public agencies, major research universities and institutions, and private companies. Non-governmental organizations, not-for-profits, and the general public have shown interest in funding EBP activities. Crowdfunding among scientists has been effective in raising funds (e.g., for the VGP) and could be expanded to fill important taxonomic gaps. We anticipate that individual philanthropy will play an important role in achieving the project’s end goals, and much effort is already underway to work with visionary philanthropists who appreciate the planet-critical nature of the EBP (e.g., the Minderoo Foundation’s “OceanOmics” initiative; https://www.minderoo.org/oceanomics).

As noted above, the EBP’s impact can be maximized through rich rewards delivered by other modalities of analysis, such as deep functional genomics, proteomics, metabolomics, or single-cell atlasing, applied to a wide diversity of species. The biobanks established or enhanced for the EBP could provide essential support for these additional programs of work, which would need significant additional funding if attempted on a large scale.

Including the FIF, the total of US$4.42 billion required to fulfill the EBP goal of sequencing and analyzing 1.67 million species in 10 years is very reasonable for a global effort with such a lasting impact. The EBP offers extraordinary value for money if one compares the project cost to the US$3 billion Human Genome Project (nearly US$6 billion in inflation-adjusted dollars; see 107), the US$10 billion cost of the Webb Telescope (108), or the US$13 billion cost of discovering the Higgs boson at the Large Hadron Collider (109).