SPECIAL FEATURE: SPECIAL FEATURE: PERSPECTIVE PERSPECTIVE Standards recommendations for the Earth BioGenome Project Mara K. N. Lawniczaka, Richard Durbina,b, Paul Fliceka,c, Kerstin Lindblad-Tohd,e , Xiaofeng Weif, John M. Archibaldg , William J. Bakerh , Katherine Belovi, Mark L. Blaxtera, Tomas Marques Bonetj,k,l,m , Anna K. Childersn , Jonathan A. Coddingtono , Keith A. Crandallp , Andrew J. Crawfordq, Robert P. Daveyr, Federica Di Palmas, Qi Fangt , Wilfried Haertyu, Neil Halls,u , Katharina J. Hoffv , Kerstin Howea , Erich D. Jarvisw,x, Warren E. Johnsony,z, Rebecca N. Johnsono , Paul J. Kerseyaa , Xin Liuf, Jose Victor Lopezbb , Eugene W. Myerscc , Olga Vinnere Petterssondd , Adam M. Phillippyee , Monica F. Poelchauff , Kim D. Pruittgg, Arang Rhieee , Juan Carlos Castilla-Rubiohh, Sunil Kumar Sahuf,t , Nicholas A. Salmona, Pamela S. Soltisii , David Swarbrecku , Françoise Thibaud-Nissengg, Sibo Wangf, Jill L. Wegrzynjj,kk , Guojie Zhangll,mm,nn,oo , He Zhangpp, Harris A. Lewinqq,rr , and Stephen Richardsqq,1 Edited by Gene Robinson, Entomology, University of Illinois at Urbana–Champaign, Urbana, IL; received September 10, 2021; accepted November 3, 2021 A global international initiative, such as the Earth BioGenome Project (EBP), requires both agreement and coordination on standards to ensure that the collective effort generates rapid progress toward its goals. To this end, the EBP initiated five technical standards committees comprising volunteer members from the global genomics scientific community: Sample Collection and Processing, Sequencing and Assembly, Anno- tation, Analysis, and IT and Informatics. The current versions of the resulting standards documents are avail- able on the EBP website, with the recognition that opportunities, technologies, and challenges may improve or change in the future, requiring flexibility for the EBP to meet its goals. Here, we describe some highlights from the proposed standards, and areas where additional challenges will need to be met. Earth BioGenome Project j genomics j ethics j genome assembly The Earth BioGenome Project (EBP) aims to quality reference genomes, suitable for use as sequence and characterize reference genomes for long-term resources faithfully representing the each known eukaryotic species (1, 2). The project ini- DNA sequence of each species in chromosomal tiated from early discussions including a working scaffolds with few gaps. One recent outcome of this group in 2017 (3), and formally launched at the Well- effort—led by A.R., Shane McCarthy, Olivier come foundation EBP/Vertebrate Genomes Project Fedrigo, and many members of the VGP—is the (VGP) workshop in London on November first, 2018. publication of the assembly lessons learned from the This ambitious vision builds on previous successes, generation of 16 very high-quality genome assem- starting with sequencing the human genome (4) and blies representing major vertebrate lineages (10). those of associated model organisms (5–9), through Companies such as Pacific Biosciences (PacBio) have projects completing tens and more recently hun- worked hand-in-hand with these efforts to improve dreds of genome sequences (1). A major positive the quality of long-read sequencing for fast and outcome of these efforts—especially those of the accurate genome assembly (e.g., ref. 11) and low- VGP—has been a renewed focus on producing high- input requirements to enable sequencing of species Author contributions: M.K.N.L.., R.D., P.F., K.L.-T., X.W., J.M.A., W.J.B., K.B., M.L.B., T.M.B., A.K.C., J.A,C., K.A.C., A.J.C., R.P.D., F.D.P., Q.F., W.H., N.H., K.J.H., K.H., E.D.J., W.E.J., R.N.J., P.J.K., X.L., J.V.L., E.W.M., O.V.P., A.M.P., M.F.P., K.D.P., A.R., J.C.C.-R., S.K.S., N.A.S., P.S.S., D.S., F.T.-N., S.W., J.L.W., G.Z., H.Z., H.A.L., and S.R. wrote the paper; M.K.N.L. chaired the Sample Collection and Processing Committee, and led the development of those standards; R.D. chaired the Assembly Standards Committee and led the development of assembly standards; P.F. chaired the Annotation Standards Committee and led the development of annotation standards; K.L.-T. chaired the Analysis Committee and led the development of analysis standards; X.W. chaired the IT Standards Committee and led the development and design of IT standards; J.M.A., W.J.B., K.B., M.L.B., T.M.B., A.K.C., J.A.C., K.A.C., A.J.C., R.P.D., F.D.P., Q.F., W.H., N.H., K.J.H., K.H., E.D.J., W.E.J., R.N.J., P.J.K., X.L., J.V.L., E.W.M., O.V.P., A.M.P., M.F.P., K.D.P., A.R., J.C.C.-R., S.K.S., N.A.S., P.S.S., D.S., F.T.-N., S.W., Downloaded at Universitat Autonoma de Barcelona (UAB) on January 24, 2022 J.L.W., G.Z., H.Z., and H.A.L. contributed to Earth BioGenome Project standards development; S.R. led standards development for all subgroups and was Chair of the Earth BioGenome Project International Scientific Committee for the development of these standards and contributed to Earth BioGenome Project standards development. Competing interest statement: P.F. is a member of the scientific advisory boards of Fabric Genomics, Inc. and Eagle Genomics, Ltd. This article is a PNAS Direct Submission. This article is distributed under Creative Commons Attribution-NonCommercial-NoDerivatives License 4.0 (CC BY-NC-ND). 1 To whom correspondence may be addressed. Email: srichards@ucdavis.edu. This article contains supporting information online at http://www.pnas.org/lookup/suppl/doi:10.1073/pnas.2115639118/-/DCSupplemental. Published January 18, 2022. PNAS 2022 Vol. 119 No. 4 e2115639118 https://doi.org/10.1073/pnas.2115639118 j 1 of 8 with minute individual size (12). The major lessons from these eukaryotic family (2). To provide greater flexibility in achieving efforts are the necessity of long reads, the need to sequence phase 1 goals, we suggest that multiple species for each family single individuals to reduce haplotype complexity, the neces- should be proposed and collected simultaneously. sity of long-range linked read technology, such as Hi-C to As much as possible, this process should be globally trans- place assembled sequences in a chromosomal context, high- parent and open to input from the wider community that may quality QC pipelines, and iterative improvement of assembly not be actively generating reference genomes but who will ben- pipelines for continued increases in quality and decreases in efit from their availability. To assist with this transparency and cost. the need for community input, we have created an open EBP The five EBP technical standards committees (Sample Collec- Family_Reference_Proposals spreadsheet to document pro- tion and Processing, Sequencing and Assembly, Annotation, posals for target taxa. Please note there are two tabs available Analysis, and IT and Informatics) have built on these and other for community entries: “Family Reference Suggestions” collates advances to recommend quality standards and organizational recommendations from the community on ideal family-level tar- principles at all steps of the process from ethical sample collec- get species, and “Family Reference Projects” records reference tion to sharing analyses, to guide forward progress (13). In genome sequencing projects already underway (more below) that most cases the committees have refrained from highly specific are targeting a relatively small and clearly defined set of taxa. For recommendations, as the rapid pace of research in all areas larger-scale projects that aim to sequence hundreds or thousands often changes the best approach within months. We have pro- of species, target lists are likely to undergo revision as projects duced recommendations rather than a definitive “how to” set proceed. Therefore, we have created two distinct approaches (for of commandments, and we will update these recommendations small defined projects or larger projects with funding) to facilitate annually as EBP projects progress and technologies improve. global transparency of target species lists depending on the size Here we describe these recommendations, also found on the of the project. In both cases, species and their associated projects EBP website (13). will be searchable and displayed on the “Genomes on a Tree” service (https://goat.genomehubs.org/). Results Considerations for selection of phase 1 family representatives. Sample Collection and Processing Recommendations. Ethical A variety of factors should be considered when proposing a collecting. It is of utmost importance that specimens and data species representative of a family, and these are listed here in contributing to the EBP are legally obtained and projects are order of importance: 1) permissions and availability: sampling is ethically conducted. All collection activities should carefully achievable taking into consideration permissions and legal follow institutional and national protocols, including but not requirements; 2) community value: of broad community use and limited to prior informed consent, compliance with the value (this could be assessed through surveys oriented toward Nagoya Protocol of the Convention for Biological Diversity (14), target communities, e.g., https://docs.google.com/forms/d/e/ and endangered species legislation. Sample collectors should 1FAIpQLSfOA4GRA3mkLLnN3fng9ZM8RdMIHusqPqfMDPePy8 ensure that all local and national permissions for collection are a0sE72Ug/viewform for the Darwin Tree of Life [DToL] project); in place, and that a record of these permissions is maintained 3) publicly registered: the species is registered with its current for referral in the case of questions on a specimen’s legality. name and taxonomy in a publicly available database (we recom- These permissions can vary widely among countries, so it is mend the National Center for Biotechnology Information [NCBI] beyond the scope of this document to summarize them. A com- Taxonomy Database https://www.ncbi.nlm.nih.gov/taxonomy) plicating factor is that some specimens will be collected and and assigned a Tree of Life ID (TOLID; see below) to assist with moved out of their country of origin for sequencing. Where the tracking name and taxonomy changes over time; 4) physical country of origin has implemented the Nagoya Protocol with size: considering today’s technology limits, we propose a local laws, they must be followed. Again, precise guidance on requirement for 10 samples, each weighing more than 10 mg how to follow the Nagoya Protocol is beyond the scope of this per 1 Gb of genome size, for animals, fungi, and protists, and document, as countries interpret the protocol differently. We 100 mg per 1 Gb of genome size for plants, including a mini- recommend that organizations receiving samples ensure they mum of three samples to support the three platforms of long have permissions within a Material Transfer Agreement to pass read, Hi-C, and transcriptome sequencing; 5) species represen- the samples on if there is any anticipation that such transfer tative: generally considered to be a “good” biological species might be required. The reader is also directed to additional (not from a known species complex) and if possible, sampled guidance presented by the EBP’s Ethical, Legal, and Social from or near the type locality, which will help make the sample Issues committee (15). Beyond conforming to rules and regula- robust to taxonomic ambiguities or changes; 6) genome size: tions, collecting methods must be ethical. For example, projects where genome sizes and ploidy are known, prioritizing species should consider sampling strategies that aim to avoid overcol- with smaller, diploid genomes (because costs of data genera- lection, such as lineage focused bio-blitzes supervised by a tion are projected to fall and our ability to assemble large high- group of taxonomic experts. repeat content genomes will improve in future EBP phases); 7) Identifying EBP phase 1 family-representative target species: taxonomic stability: not currently subject to disagreement Downloaded at Universitat Autonoma de Barcelona (UAB) on January 24, 2022 Open collation of community proposals and genomes underway. or revision. There are over 9,000 taxonomic families within Eukaryota, Metadata to place each genome in its biological and ecological including 6,470 Metazoa, 1,052 Chromista, 757 Fungi, 964 Plan- context. The assemblies for EBP species must be accompanied tae, and 221 Protozoa in the Catalogue of Life checklist (www. by robust and comprehensive metadata, including the collec- catalogueoflife.org/annual-checklist/2019). An important collec- tion event and collector identification in a common set of tive step toward EBP ambitions is to gather the list of target agreed metadata fields. Each contributed specimen should species proposed to be family representatives to fulfill the be identified to species level by a taxonomic expert and, when- phase 1 goal of a high-quality reference genome for every ever possible, material from the same specimen should be 2 of 8 j PNAS Lawniczak et al. https://doi.org/10.1073/pnas.2115639118 Standards recommendations for the Earth BioGenome Project independently DNA barcoded using appropriate markers and scope, this is unlikely to be true in years to come and the reten- the data deposited into BOLD (www.boldsystems.org) and the tion of relevant material to add additional layers of data to the International Nucleotide Sequence Database Collaboration (16) high-quality reference genomes would be prudent. Thus, for (INSDC, https://www.insdc.org) databases (GenBank, European specimens where material is available in excess, considerations Nucleotide Archive [ENA], DNA Databank of Japan [DDBJ]). should be given to appropriate storage of additional samples to These DNA barcodes will serve both to ensure that species with future-proof these specimens as much as possible. For speci- reference genomes have assembly-derived barcodes matching mens where all material is used in the data generation, perhaps independently generated DNA barcode data and that the refer- typically discarded supernatants should be retained for future ence genome corresponds to the named taxon (i.e., that no investigations. sample swaps have occurred during processing). Given the We also encourage activities that simplify and streamline importance of ensuring the specimen is truly representative of standard procedures, such that it becomes easier to “containerize” the species to which it is assigned, where these standards are extraction, sequencing, and assembly activities. Containerization not achievable it is advisable to substitute a different represen- means a world in which a shipping container or crate could house tative for the family. everything needed to go from sample to sequence and would Consensus data practices should be established early in the build capacity in low- and middle-income countries, often harbor- project planning to ensure that metadata fields and terms are ing the greatest biodiversity. This will allow all nations to deliver to standardized, and ideally that dedicated tools are provided to EBP goals while building capacity and expertise and enhancing assist with capture and brokering of metadata into INSDC and global science. project-level repositories [e.g., COPO, in use by the United Kingdom Darwin Tree of Life project (https://copo-project.org Sequencing and Assembly Recommendations. The primary goal (17)]. Coordination of best practice in metadata collection is is to obtain for each species a sufficiently complete, accurate, delivered by the EBP IT/Informatics Best Practices committee, and contiguous representative genome sequence that can pro- which draws on the efforts to standardize metadata collected by vide a reference for further genomic analysis. This represents a the DToL project (https://darwintreeoflife.org). We recommend single haplotype at each locus, and so can be provided by a sin- when possible that high-quality, informative images should gle haploid assembly from an inbred or outbred individual, accompany each contributed specimen, and these should be whatever the ploidy. Representation of alternate haplotypes made publicly available. Ideally, these images will be deposited from the sampled individual, if outbred, is desirable to illustrate in the BOLD database to accompany the DNA barcode (and in genetic variation, but secondary. Quantitative assembly standards. Where sufficient DNA and a yet-to-be designed portal for EBP that supports access to all material is available from a single individual (or clonal colony), metadata for each sequenced species). currently more than around 100 ng DNA per gigabase genome, Specimens used to generate reference genomes should be we propose a minimum reference standard of 6.C.Q40 (i.e., >1 vouchered. These vouchers should take as many complemen- Mb NG50 contig continuity and chromosomal scale NG50 scaf- tary forms as is practical, including museum voucher specimens folding, with less than 1/10,000 nucleotide error rate (see SI (unless the whole specimen is required for sequencing or legal/ Appendix, Table S1, taken from ref. 10, for notation and further ethical issues prevent sacrificing the source individual), tissue information). When the chromosome NG50 is smaller than a vouchers (discussed further below), viably frozen cell lines, megabase this will be C.C.Q40, with “C” denoting having image vouchers, and molecular vouchers of extracted RNA and achieved full chromosome lengths. This standard is now DNA. Vouchers should, wherever possible be deposited in pub- licly accessible collection facilities located in the country of ori- gin of the specimen, or where there is excess material, possibly spread across multiple repositories. It is recommended that sub- samples and sample derivatives be stored in a Global Genome Biodiversity Network member institution and linked to the voucher using a unique identifier. Fig. 1 shows an example of a collected species during physical processing on dry ice, which is recommended to maintain high molecular weight DNA and RNA integrity. EBP sample collection in the future. Current best-practice assembly guidelines are to generate a combination of data types, including long-read (PacBio HiFi, and Oxford Nanopore Technologies long-reads [ONT]), long-range (Hi-C paired reads and perhaps linked reads or Bionano optical maps), and tran- scriptomic (RNA-seq Illumina short read, PacBio Iso-seq, or ONT cDNA-PCR) data from the same specimen wherever possi- Downloaded at Universitat Autonoma de Barcelona (UAB) on January 24, 2022 ble, aiming for the heterogametic sex where relevant. EBP part- Fig. 1. An example of the documentation that should occur as a ners are developing protocols that support minimal extraction sample is being processed. The SPECIMEN_ID (the Natural History of material sufficient for any of these types of data generation. Museum, London [NHMUK] barcode under the fly) is photographed Furthermore, current standard operating procedures for nucleic alongside the specimen and the barcoded tubes into which different samples of that specimen will be placed. The metadata tracking acid extraction for genomics effectively discard material that sheet would have three entries for this fly, where collection-related may be useful in the future, such as proteins and metabolites, information would be identical, but tissue type and tissue size and do not explicitly preserve the microbiome or live cells. would vary (e.g., head, thorax, and abdomen, each in a separate While data generation from these materials is currently out of tube). Image credit: M.K.N.L. Lawniczak et al. PNAS j 3 of 8 Standards recommendations for the Earth BioGenome Project https://doi.org/10.1073/pnas.2115639118 achievable with reasonable effort using multiple alternative between the species (i.e., all chromosomes are in one-to-one cor- long-read–based strategies, and is comparable to the standard respondence, but may have within chromosome rearrangements). of the human GRCh37 (hg19) and mouse GRCm38 (mm10) INSDC project structure and nomenclature. For a reference references that have proven to be of high utility for many years. genome to count toward the EBP goals, it must be submitted to Alongside the contiguity and error-rate goals, we propose the the INSDC Genomes Division for open access use by the scien- following additional criteria (SI Appendix, Table S1): fewer than tific community. When this submission is made, the assembly is 5% false duplications, more than 90% kmer completeness, more associated with a “data” BioProject object, which can be part of than 90% of the sequence assigned to candidate chromosomal a hierarchy by being assigned to an “umbrella” BioProject. We sequences, more than 90% single copy conserved genes [e.g., suggest the structure shown in Fig. 2, with a data project per as inferred using BUSCO (18)] complete and single copy, and assembly, an umbrella project for the sample (note that under more than 90% of transcripts from the same species mappable. this there may be assemblies of separate symbiont or cobiont Suggestions of and links to standard tools for measuring these species), and then above an umbrella BioProject corresponding metrics are provided on the EBP website (13). to the overall project. When creating an umbrella project in We believe that these are achievable goals for most and per- INSDC, please request that it be linked to the EBP BioProject haps ultimately all species for which large enough high-quality object (PRJNA533106, https://www.ncbi.nlm.nih.gov/bioproject/ samples can be obtained. However, we recognize that for many 533106). If an assembly contributes to other larger-scale efforts reasons (e.g., sample quality, very large genomes, polyploidy, (e.g., Global Invertebrate Genomics Alliance or VGP), you can cost expediency) they may not be met in the first instance. request it also be linked to the respective umbrella BioProjects. Interim references that do not meet the standard can be very These requests go through an approval process. useful and should be valued. Nevertheless, there will be a con- To be submitted to the INSDC databases to a BioProject, an tinuing EBP goal to revisit these and bring them up to the target assembly needs to be assigned to a TaxonID entry in the NCBI standard as that becomes practical. Taxonomy database. Although TaxonID identifiers can be cre- For species for which there is more limited DNA available ated for informal taxa, such as Maylandia sp. “pearly”, we would per individual, additional technology development is required like EBP genomes to be associated with taxonomically valid both in generation of sequence from limiting input, and in the species names, and we urge EBP-affiliated projects to work with assembly of genomes from multiple individuals. We expect that appropriate taxonomists to identify samples to a species and, improved approaches will bring more and more “small” species where necessary, to establish the species name in the standard into group above. In the meantime, we propose the interim manner in the literature. standard outlined in the 4.5.Q40 column of SI Appendix, Table Furthermore, in addition to the numerical identifiers gener- S1. For as yet uncultured single-celled eukaryotes, we expect ated for assemblies by the public databases, we request proj- that a metagenomics-like standard will be determined, based ects to adopt the TOLID standard short nomenclature for on experience in the prokaryotic and emerging microeukaryote samples and assemblies, as used by the VGP and DToL. This metagenome fields. No specific recommendations are made at takes the form .. For example, ilAlcRepa1.1 identifies the first this time. assembly of an insect lepidopteran (two characters “il”), Alcis Additional requirements. In addition to the quantitative require- (three characters “Alc”), repandata (four characters “Repa”) ments above, our experience has shown that currently all (com- individual one (“1”), assembly version one (0.1). Unique TOLID binations of) automated processes generate assemblies with a designations have been generated to cover all ∼485,000 spe- cies that already have data in INSDC or are found in Britain and variety of remaining errors, some of which are relatively easy to Ireland. Other species can be added on request. Details of the address and should be corrected before submission to public two-letter prefix assignment, which partitions the tree of life, databases (19). The full recommendations also propose that and of the currently assigned identifiers, are at https://gitlab. projects meet a wider set of quality control criteria, including com/wtsi-grit/darwin-tree-of-life-sample-naming. A server to separation of the organelle genomes of target species from view and assign unique individual identifiers for specimens is their nuclear genomes, separation of the genomes of any sym- available at https://id.tol.sanger.ac.uk/, with instructions for reg- bionts or cobionts, explicit identification of primary and alter- istering new users and projects. nate haplotype assemblies, that the sequence includes only A, Ongoing EBP sequencing and assembly challenges. Given rap- C, G, T, and N base calls, and that sequences should not begin idly evolving technologies and bioinformatic tools, we have or end with Ns. The reader is referred to the full recommenda- deliberately avoided defining hard and fast rules about how tions on the EBP website for full details (13). genomes should be sequenced and assembled. Continued Identification and naming of chromosomal-scale scaffolds efforts are still required to utilize future sequencing technology can be achieved with Hi-C two-dimensional maps and compari- improvements by combining them with new assembly algo- son to existing karyotyping and linkage maps where available. If rithms to continue to improve both assembly quality and reduce chromosome naming already exists for a given species, it should costs for future phases of the EBP. Despite major achievements be reflected in the new assembly so long as the chromosomes (e.g., ref. 19), improved technologies are still required for the Downloaded at Universitat Autonoma de Barcelona (UAB) on January 24, 2022 match to those of the prior naming. If no previous names exist, large number of eukaryotic species with very small individuals, we recommend naming chromosomes by size, taking into and genomics of unculturable single-celled eukaryotes remains account scaffolds that can be assigned to belong to a certain a major challenge requiring innovation before large scale pro- chromosome, but could not be unambiguously placed (unlocal- duction is advisable. ized scaffolds). An alternative that is applicable in some cases is to name chromosomes after those in a closely related species Annotation Recommendations. High-quality annotation is required with an established nomenclature; we only recommend this if to transform reference genome sequences into actionable knowl- there are no major interchromosomal rearrangements identified edge. In the most expansive definition, genome annotation is an 4 of 8 j PNAS Lawniczak et al. https://doi.org/10.1073/pnas.2115639118 Standards recommendations for the Earth BioGenome Project The full annotation set for each feature class may consist of one or more feature sets. For example, a full protein-coding annota- tion set may have a generalized feature set annotated with a particular methodology and a specific feature set, such as immune genes, annotated with a different methodology. Appropriate descriptors thus must exist at the level of the anno- tation set, the feature set and, where possible, each annotated feature. Descriptors for annotation sets and feature sets are largely interchangeable and are reported together. Annotation set and feature set descriptors are: • Accession number(s) of all transcriptomic data used in the annotation. The use of each data within the annotation pro- cess must also be specified. These data may be specifically generated for the annotation or previously available. • Defined protein sequence data sets ideally specified by, for Fig. 2. Recommended EBP INSDC assembly BioProject submission structure. Alternatives on this theme could include maternal assem- example, UniProt release number or the annotation set trans- bly and paternal assembly at the DataBioProject levels, when com- lated specific by, for example, Ensembl release number. plete genome assemblies are generated for both haplotypes of a • Accession numbers of genome assemblies used for align- diploid sexually reproducing species. “Your Project” represents a ments and information about any annotation on those assem- project sequencing multiple genomes, such as the VGP or DToL. blies used for projection. • Release numbers of supporting information used for annota- accounting of the role and history of each base pair in the tion such as repeat libraries or hidden Markov model profiles. genome. Unlike genome assembly for which completeness and • Software versions and parameter settings used. accuracy can be exactly defined, for most eukaryotic species it is Annotation feature descriptors are: likely impossible that an annotation can be constructed that would account for all possible conditions and sequence variation. How- • Where annotated features are based on accessioned evidence ever, the completeness and quality of gene annotation are sub- sequences such as protein sequences, these accession num- stantially higher when transcriptomic data from the same species bers should be associated with the annotated object. (ideally the same individual that has had its genome sequenced) is • Where annotated features, such as genes, are named the pro- incorporated. For this reason, we recommend collection of tran- cess for naming must be described and the supporting infor- scriptomic data for all sequenced species. Indeed, some annota- mation used for the naming (e.g., Hugo Gene Nomenclature tion pipelines require such data. Committee access date, reference genome annotation, and Genome features to be annotated. We propose that required so forth) must be specified. annotation for all genomes include the feature classes listed in Annotation sharing and coordination. All annotation must be the left-hand column of SI Appendix, Table S2. Additionally, the freely available without any restrictions on further use (equiva- feature classes listed in the right-hand column may be anno- lent to public domain or CC0 licensing). tated in some genomes. Annotation must be mappable to sequence records con- For genes, the supporting evidence shall be provided at the tained in the INSDC, although the provider may use different level of gene or transcript annotation (see SI Appendix, Table nomenclature for a given sequence, such as “chromosome 1,” S3 for evidence types) if possible or at the level of feature set if as long as the corresponding INSDC sequence identifier is not. Note that hybrid annotation approaches are common using incorporated into the annotation release. Annotation sets will multiple evidence categories across a single transcript or gene. be named based on the formal assembly name. Annotation In general, all the evidence categories used for annotation shall be provided in standard file formats such as GFF3 with should be recorded. standard descriptive attribute tags. As described in SI Appendix, Table S3, evidence data from Requirements for annotation providers. Annotation meeting other species (i.e., homology information) is commonly used. EBP standards can be produced either by centralized annota- The value of this information is a function of the type of informa- tion services, such as those provided by Ensembl (20) at Euro- tion and the sequence divergence between the target genome pean Molecular Biology Laboratory’s European Bioinformatics and the source of the evidence. The use of transcriptomic data Institute (EMBL-EBI) and RefSEq. (21) at the NIH’s NCBI or by from other species for annotation is possible but should be other groups, either as part of the generation of the genome carefully evaluated because alignment errors across will nega- sequence or by groups provided distributed annotation services tively impact annotation accuracy. that are targeted by phylogenetic scope or other reason. As much as possible, coding and noncoding genes should For the centralized annotation services, genomes will only Downloaded at Universitat Autonoma de Barcelona (UAB) on January 24, 2022 be assigned a molecular function and, when assigned, the be annotated after they have been submitted to an INSDC method for assigning such function must be specified. Common database (GenBank, ENA, DDBJ), have been accessioned with methods for assigning function include homology, orthology to an GCA_XX identifier, and have been publicly released. a reference genome, matching to hidden Markov models, and Supporting transcriptome data for annotation (including tran- gene ontology terms. scriptome assemblies where they have been used) must also be Annotation description. For a given genome, similar features submitted, accessioned, and released via a recognized archival will be collected into feature sets that will be further described database, such as ArrayExpress, Gene Expression Omnibus, by the software and parameters used to create the annotation. or INSDC. Other groups proposing themselves as central Lawniczak et al. PNAS j 5 of 8 Standards recommendations for the Earth BioGenome Project https://doi.org/10.1073/pnas.2115639118 annotation services in the context of EBP are expected to follow 2. Repeat content and evolution: Catalogs of simple sequence these requirements of annotation and use of openly available repeats and mobile elements will be generated during genome sequences and evidence data. genome annotation. Repeat elements may be exapted into Estimating annotation quality. Annotation quality is an estimate novel regulatory elements or transcripts that may regulate of completeness and accuracy. It should be based on multiple gene function. Based on their similarity and distribution factors, including: 1) deep evolutionary conservation; 2) com- within and between genomes, it is possible to form and test parative data from closely related species in the same clade, hypotheses of the evolution of repeat content and their con- ideally with at least one gold standard reference for the clade to tributions to adaptation. compare to; and 3) expressed sequence evidence from the spe- 3. Partial or whole-genome duplication: Genome size is a func- cies or individual being annotated. Moreover, annotation qual- tion of gain and loss of sequence, including genes, repeats, ity assessment encompasses both the need to capture the loci segmental, and genome duplications. Many evolutionary lin- (i.e., the completeness of the gene set) as well as the details of eages have undergone whole or partial genome duplication. the transcript set. Such duplications allow the divergent evolution of duplicated For the foreseeable future, annotation quality will be mea- genes. Using synteny between species and internal similarity sured in a relative scale as compared to other genomes within of genome sections allows for analysis of gene gain and loss. the same clade. Finished genome annotation is unlikely to be 4. Updating the evolutionary tree: Species trees provide a com- possible for nearly all species. When the quality of annotation parative framework for multiple analyses, such as inferring for given species has fallen substantially below the relative level evolutionary constraint, detecting positive selection, delineat- for similar species, it should be prioritized for reannotation. ing species boundaries/hybridization events, and estimating Basic statistics for protein-coding and nonprotein-coding evolutionary relationships. The tree may be built using differ- genes should be produced, including the number of genes and, ent types of genomic sequence data, including ancestral for protein coding genes, the number of exons per gene. repeats, genes, and fourfold degenerate codons. While our Protein-coding gene annotation should also be compared to understanding of family-level relationships across the tree of gene-based assessments of the assembly quality. For example, life is fairly stable, more closely related species can vary sig- BUSCO results from the final gene predictions should be com- nificantly across genomic regions through processes of pared to the estimated BUSCO score on the genome assembly hybridization, incomplete lineage sorting, and horizontal to ensure that these metrics are comparable. Repeat annotation gene transfer. should be compared with a measure of repeat completeness in 5. Evolutionary constraint: To identify functional elements in the the genome assembly, such as the long terminal repeat index genome of each species, evolutionarily constrained regions value (22). need to be identified. These will include not only transcribed Ongoing annotation challenges. A number of basic and applied protein-coding and RNA genes, but also regulatory elements. genome annotation research challenges in genome annotation Given sufficient data (such as a large number of related spe- remain to be addressed, including: 1) the creation of robust, cies) constraint can be detected at the single-base level. community-defined measures of annotation quality; 2) increas- 6. Analysis of gene content, gene family expansion/contraction, ing portability and automation of current genome annotation and selection on protein-coding genes: Both gene family pipelines; and 3) developing training and providing funding to expansions and contractions as well as positive selection on access cloud-based or deployable annotation tools to diversify specific regions of a protein are important for species evolu- the community of scientists who can use and contribute to tion. Typically, a majority of genes are 1:1 orthologs between genome annotation and the development of annotation tools. larger species groups (∼70% of mammalian genes). To detect gene family expansions, whole-genome alignments and syn- Analysis Recommendations. One of the challenges of capturing teny can be used. Such studies can also detect horizontal (lat- analysis standards for the diverse needs of a genomics commu- eral) gene transfer. nity is that there are many scientific goals and practical uses for 7. Analysis of noncoding transcripts: Noncoding transcripts genome data. Substantial variation in genome size and com- have a key role in genome regulation and function, and plexity, life history, and other features means that analysis tools should be identified as part of the genome annotation pro- may not work on all species and groups of species. Here we cess. Noncoding transcripts typically evolve more rapidly highlight some key analysis areas, across a broad set of than protein-coding genes, so performing analyses to study questions related to genome evolution, population genomics, the evolution of noncoding transcripts in different species is and biodiversity, and list some currently available tools (SI important. Analysis can be performed either by direct com- Appendix, Table S4). Many of these tools will need to be altered parison of transcripts between related species or by compari- to handle larger datasets and to ask novel questions only son using the whole-genome alignment/synteny. answerable at the scale of the EBP. 8. Intraspecific variation, conservation, biodiversity, and adapta- Key types of analysis. Genome analysis targets a broad set of tion: As more and more species become endangered or criti- questions related to genome evolution, population genomics, cally endangered, the need for genomic information to guide and biodiversity. We outline some of the more common Downloaded at Universitat Autonoma de Barcelona (UAB) on January 24, 2022 conservation efforts increases. To generate data for these research areas: analyses, whole genome resequencing data from multiple 1. Alignments and synteny analysis of related species: An align- populations with 10 to 20 individuals each are ideal; however, ment forms the basis for comparative analysis across species, important information can still be gleaned from the two hap- allowing comparison of the same genome features in many lotypes present in a single diploid individual (i.e., the genome species. Multiple tools for sequence alignment exist, but reference). With multiple populations, not only can markers most need to be scaled up to handle the expected thousands for managing populations be generated, but signals of adap- of genomes. tation can also be identified. 6 of 8 j PNAS Lawniczak et al. https://doi.org/10.1073/pnas.2115639118 Standards recommendations for the Earth BioGenome Project 9. Environmental DNA and ecological sampling: eDNA analysis facilitates the identification of threatened and nonthreatened species within ecosystems. Currently, analyses typically use PCR amplicon metabarcoding, but the availability of whole genomes would allow unbiased sampling approaches. By generating a high-quality digital library, based on whole- genome sequences to underpin the identification of eDNA samples, the EBP can accelerate this work. Thus, it is impor- tant to deposit not just genome sequences into INSDC data- bases, but also derived barcodes to appropriate barcode public repositories (e.g., BOLD: www.boldsystems.org) to accelerate their growth and enable identification of species from nonbarcode sequences. Supporting the EBP analysis mission. To support the EBP analy- sis mission, we note four key needs: 1) enhancing analysis tools to handle large datasets; 2) having sufficient computer resour- ces and ways to share the data; 3) fostering collaboration between nongenomicists and genomicists; and 4) ensuring that the whole EBP community has access to tutorials for different types of genome analysis. Fig. 3. Recommended future EBP data service portal. Researchers would easily apply for the hardware resources that are shared by IT and Informatics Recommendations. Access and sharing poli- the EBP affiliated project networks. cies. EBP supports the FAIR principles (findability, accessibility, interoperability, and reusability) (23) and we recommend that of reference genomes will require active exploration and collec- no restrictions are placed on data access by submitting consen- tive learning once the phase 1 references are available. How- sus reference genomes and corresponding sample metadata to ever, coordination in the analysis space is critical to accelerate INSDC-managed data repositories. There are possible caveats returns to society and EBP members are committed to that may arise from the Convention on Biological Diversity continued efforts in this area. We will revisit and update recom- Access and Benefit Sharing protocols and national laws, but the mendations as the EBP progresses. For phases 2 and 3 we will EBP recommends open deposition in INSDC databases unless have gained experience and improved technologies in sam- local laws make this impossible. pling, sequencing, and analysis, to fill gaps where no recom- IT infrastructure. The EBP infrastructure requirements are large, mendations can be made at the current time. not only for the genome assembly data that need to be For individual groups or large consortia wishing to coordi- archived, but also the storage and computing resources needed nate with the EBP, there are two general ways to contribute. For for intermediate analyses. These infrastructures need to be an individual laboratory, where the goal is to generate a refer- interoperable with repositories and data-management plat- ence genome worthy of being a long-term resource, we believe forms that house EBP data and metadata. This will require the recommendations here are critical to assuring success. consensus across platforms to facilitate clear and transparent Coordination of target species via the mechanisms, such as the information exchange. We recommend the development of EBP Family_Reference_Proposals spreadsheet described above, a mechanism to share the infrastructure capabilities for each and rapid public release of references to the INSDC with affiliated project (Fig. 3) with the aim of ensuring global genome note manuscripts will accelerate all fields. For large participation. consortia, we recommend formally joining the EBP as described IT and informatics areas requiring future research. The IT and on the EBP website (13) and contributing groups of species informatics challenges are not conceptual but practical. While coordinated with other EBP members. the computational resources required are large, they are cer- Finally, we are excited that we are technically able to com- tainly not intractable on the global scale. Thus, the practicalities plete phase 1 reference genomes to an extremely high quality. of making pipelines and computational resources available to We hope that these guidelines will influence and improve the all is a significant organizational problem to be solved. A key quality of individual reference genome sequences, whomever factor for success will be the identification of distinct “modules” generates them, to create annotated, analyzed, world-wide within the data processing and outputs of EBP to enable parallel accessible references that will serve humanity for the foresee- research on key computational tasks without affecting overall progress. Some of these are very clear: for example, assembly able future. software can be improved without negatively affecting down- stream genomic analyses. Data Availability. There are no data underlying this work. Downloaded at Universitat Autonoma de Barcelona (UAB) on January 24, 2022 Discussion Acknowledgments This set of recommendations is the first from the EBP commu- The work of K.D.P. and F.T.-N. was supported by the Intramural Research Pro- nity, and specifically addresses phase 1 of the project. As might gram of the National Library of Medicine, NIH. K.L.-T. is a distinguished professor funded by the Swedish Research Council. J.A.C. is supported by the Global be expected, there is more clarity for the early steps in the pro- Genome Initiative of the National Museum of Natural History, Smithsonian Insti- cess (sample collection, sequencing, and assembly) than for the tution. E.D.J. is supported by the Howard Hughes Medical Institute (HHMI). P.F. later steps (annotation, analysis, global sharing at scale, and so is supported by Wellcome Grant WT108749/Z/15/Z and the European Molecular forth). The comparative genomics analysis of tens of thousands Biology Laboratory. J.L.W. is supported by National Science Foundation Grant Lawniczak et al. PNAS j 7 of 8 Standards recommendations for the Earth BioGenome Project https://doi.org/10.1073/pnas.2115639118 DBI:IIBR:CAREER #1943371. This work was supported in part by the US Depart- specific information and does not imply recommendation or endorsement by the ment of Agriculture, Agricultural Research Service. Mention of trade names or US Department of Agriculture. The US Department of Agriculture is an equal commercial products in this publication is solely for the purpose of providing opportunity provider and employer. a Wellcome Sanger Institute, Wellcome Genome Campus, Hinxton CB10 1SA, United Kingdom; b Department of Genetics, University of Cambridge, Cambridge CB3 0DH, United Kingdom; c European Molecular Biology Laboratory, European Bioinformatics Institute, Wellcome Genome Campus, Cambridge CB10 1SD, United Kingdom; d Broad Institute of MIT and Harvard, Cambridge, MA 02142; e Science for Life Laboratory, Department of Medical Biochemistry and Microbiology, Uppsala University 751 23 Uppsala, Sweden; f China National GeneBank, Shenzhen 518120, China; g Department of Biochemistry and Molecular Biology, Dalhousie University, Halifax, NS B3H 4R2, Canada; h Department of Accelerated Taxonomy, Royal Botanic Gardens, Kew, Surrey TW9 3AE, United Kingdom; i School of Life and Environmental Sciences, Faculty of Science, University of Sydney, Sydney, NSW 2006, Australia; j Institute of Evolutionary Biology, Consejo Superior de Investigaciones Cientıficas-Universitat Pompeau Fabra, Parc de Rechercha Biom edica Barce- lona 08003 Barcelona, Spain; k Catalan Institution of Research and Advanced Studies 08010 Barcelona, Spain; l Centre Nacional d’An alisi Geon omica - Centre for Genomic m Regulation, Barcelona Institute of Science and Technology 08028 Barcelona, Spain; Institut Catal a de Paleontologia Miquel Crusafont, Universitat Aut onoma de Barcelona 08193 Barcelona, Spain; n Bee Research Laboratory, Beltsville Agricultural Research Center, US Department of Agriculture, Agricultural Research Service, Beltsville, MD 20705; o Smithsonian Institution, National Museum of Natural History, Washington, DC 20560-0105; p Computational Biology Institute and Department of Biostatistics & Bioinformatics, Milken Institute School of Public Health, The George Washington University, Washington, DC 20052; q Department of Biological Sciences, Universidad de los Andes 111711 Bogot a, Colombia; r Engineering Biology, Earlham Institute, Norwich Research Park, Norwich NR4 7UZ, United Kingdom; s Genome British Columbia, Vancouver, BC V5Z 0C4, Canada; t BGI-Shenzhen, Beishan Industrial Zone, Shenzhen 518083, China; u Earlham Institute, Norwich Research Park, Norwich NR4 7UZ, United Kingdom; v Institute of Mathematics and Computer Science, Center for Functional Genomics of Microbes, University of Greifswald 17489 Greifswald, Germany; w Vertebrate Genomes Lab, The Rockefeller University, New York, NY 10065; x HHMI, Chevy Chase, MD 20815; y Center for Species Survival, Smithsonian Conservation Biology Institute, National Zoological Park, Front Royal, VA 22630; z The Walter Reed Biosystematics Unit, Museum Support Center MRC-534, Smithsonian Institution, Suitland, MD 20746-2863; aa European Molecular Biology Laboratory, European Bioinformatics Institute, Cambridge CB10 1SD, United Kingdom; bb Halmos College of Arts and Sciences, Guy Harvey Oceanographic Center, Nova Southeastern University, Dania Beach, FL 33004; cc Department of Systems Biology, Max Planck Institute of Molecular Cell Biology and Genetics, Dresden 01307, Germany; dd Science for Life Laboratory in Uppsala, Uppsala University 75123 Uppsala, Sweden; ee Genome Informatics Section, Computational and Statistical Genomics Branch, National Human Genome Research Institute, NIH, Bethesda, MD 20894; ff National Agricultural Library, USDA Agricultural Research Service, Beltsville, MD 20705; gg National Center for Biotechnology Information, National Library of Medicine, NIH, Bethesda, MD 20894; hh Spacetime Ventures, S~ ao Paulo 05449-050, Brazil; ii Florida Museum of Natural History, University of Florida, Gainesville, FL 32611; jj Department of Ecology and Evolutionary Biology, University of Connecticut, Storrs, CT 06269; kk Institute for Systems Genomics, Computational Biology Core, University of Connecticut, Storrs, CT 06269; ll Villum Center for Biodiversity Genomics, Section for Ecology and Evolution, Department of Biology, University of Copenhagen 1165 Copenhagen, Denmark; mm China National Genebank, BGI-Shenzhen 518083 Shenzhen, China; nn State Key Laboratory of Genetic Resources and Evolution, Kunming Institute of Zoology, Chinese Academy of Sciences 650223 Kunming, China; oo Center for Excellence in Animal Evolution and Genetics, Chinese Academy of Sciences 650223 Kunming, China; pp BGI-Qingdao, BGI-Shenzhen 266555 Qingdao, China; qq University of California Davis Genome Center, University of California, Davis, CA 95616; and rr Department of Evolution and Ecology, University of California, Davis, CA 95616 1 H. A. Lewin et al., The Earth BioGenome Project 2020: Starting the clock. Proc. Natl. Acad. Sci. U.S.A., 119, e2115635118 (2022). 2 H. A. Lewin et al., Earth BioGenome Project: Sequencing life for the future of life. Proc. Natl. Acad. Sci. U.S.A. 115, 4325–4333 (2018). 3 E. Pennisi, Sequencing all life captivates biologists. Science 355, 894–895 (2017). 4 E. S. Lander et al.; International Human Genome Sequencing Consortium, Initial sequencing and analysis of the human genome. Nature 409, 860–921 (2001). 5 R. H. Waterston et al.; Mouse Genome Sequencing Consortium, Initial sequencing and comparative analysis of the mouse genome. Nature 420, 520–562 (2002). 6 Arabidopsis Genome Initiative, Analysis of the genome sequence of the flowering plant Arabidopsis thaliana. Nature 408, 796–815 (2000). 7 M. D. Adams et al., The genome sequence of Drosophila melanogaster. Science 287, 2185–2195 (2000). 8 C. elegans Sequencing Consortium, Genome sequence of the nematode C. elegans: A platform for investigating biology. Science 282, 2012–2018 (1998). 9 A. Goffeau et al., The yeast genome directory. Nature 387, 5–6 (1997). 10 A. Rhie et al., Towards complete and error-free genome assemblies of all vertebrate species. Nature 592, 737–746 (2021). 11 M. R. Vollger et al., Improved assembly and variant detection of a haploid human genome using single-molecule, high-fidelity long reads. Ann. Hum. Genet. 84, 125–140 (2020). 12 S. B. Kingan et al., A high-quality de novo genome assembly from a single mosquito using PacBio sequencing. Genes (Basel) 10, 62 (2019). 13 The Earth BioGenome Project Working Group, The Earth BioGenome Project Website, https://www.earthbiogenome.org (2021). 14 Secretariat of the Convention on Biological Diversity, Text and Annex of the Nagoya Protocol on Access to Genetic Resources and the Fair and Equitable Shar- ing of Benefits Arising from their Utilization to the Convention on Biological Diversity. 1st ed. [ebook] (Secretariat of the Convention on Biological Diversity, Montreal, Montreal: United Nations, 2011). https://www.cbd.int/abs/text/ (Accessed 21 February 2015). 15 J. S. Sherkow et al., Ethical, legal, and social issues in the Earth BioGenome Project. Proc. Natl. Acad. Sci. U.S.A. 119, e2115859119 (2022). 16 M. Arita, I. Karsch-Mizrachi, G. Cochrane, The international nucleotide sequence database collaboration. Nucleic Acids Res. 49, D121–D124 (2021). 17 F. Shaw et al., COPO-linked open infrastructure for plant data, in Semantic Web Applications and Tools for Life Science (SWAT4LS, 2015), pp 181–182. 18 R. M. Waterhouse et al., BUSCO applications from quality assessments to gene prediction and phylogenomics. Mol. Biol. Evol. 35, 543–548 (2018). 19 K. Howe et al., Significantly improving the quality of genome assemblies through curation. Gigascience 10, giaa153 (2021). 20 B. L. Aken et al., The Ensembl gene annotation system. Database (Oxford) 2016, baw093 (2016). 21 K. D. Pruitt et al., RefSeq: An update on mammalian reference sequences. Nucleic Acids Res. 42, D756–D763 (2014). 22 S. Ou, J. Chen, N. Jiang, Assessing genome assembly quality using the LTR Assembly Index (LAI). Nucleic Acids Res. 46, e126 (2018). 23 M. D. Wilkinson et al., The FAIR Guiding Principles for scientific data management and stewardship. Sci. Data 3, 160018 (2016). Downloaded at Universitat Autonoma de Barcelona (UAB) on January 24, 2022 8 of 8 j PNAS Lawniczak et al. https://doi.org/10.1073/pnas.2115639118 Standards recommendations for the Earth BioGenome Project