Sharing biological data: why, when, and how

Data sharing is an essential element of the scientific method, imperative to ensure transparency and reproducibility. Researchers often reuse shared data for meta-analyses or to accompany new data. Different areas of research collect fundamentally different types of data, such as tabular data, sequence data, and image data. These types of data differ greatly in size and require different approaches for sharing. Here, we outline good practices to make your biological data publicly accessible and usable, generally and for several specific kinds of data. Sharing data proves more useful when others can easily find and access, interpret, and reuse the data. To maximize the benefit of sharing your data, follow the findable, accessible, interoperable, and reusable (FAIR) guiding principles of data sharing [[1]] (Box 1), which optimize reuse of generated data. The FAIR principles outline clear standards for ensuring that others can find and access your data and that once accessed, users can easily understand and reuse the data. The FAIR principles provide a clear collection of important details to include within your data and metadata (see ‘Data metadata and documentation’). The first step in (re)using data is to find them. Metadata and data should be easy to find for both humans and computers. Machine-readable metadata are essential for automatic discovery of datasets and services, so this is an essential component of the FAIRification process. Once the user finds the required data, she/he needs to know how can they be accessed, possibly including authentication and authorization. The data usually need to be integrated with other data. In addition, the data need to interoperate with applications or workflows for analysis, storage, and processing. The ultimate goal of FAIR is to optimize the reuse of data. To achieve this, metadata and data should be well-described so that they can be replicated and/or combined in different settings. By GO FAIR [[1]] (https://www.go-fair.org/fair-principles/), provided under the Creative Commons Attribution 4.0 International license. The repositories and practices we recommend below fulfill some of these principles and make it easier for you to follow others. This will not only help others using your data, but can also save you time in the future (see ‘The benefits of sharing data to individual researchers’). The National Institutes of Health (NIH), Canadian Institutes of Health Research (CIHR), Monarch Initiative [[2, 3]], and the Research Data Alliance (https://www.rd-alliance.org/) all recommend FAIR principles for data sharing. Amendments to these recommendations that add measures for traceability (such as evidence and provenance), licensing, and connectedness (such as identifiers and versioning) further improve data reusability [[4, 5]]. Sharing data allows for transparency in scientific studies and allows one to fully understand what occurred in an analysis and reproduce the results. Without complete data, metadata (see ‘Data and metadata’), and information about resources used to generate the data, reproducing a study proves impossible [[6, 7]]. Within the biological sciences, we have a problem of data waste—ostensibly shared data that no one ever uses. Many otherwise useful datasets go underused because researchers cannot effectively reuse the data. The inability to reuse arises from lack of discoverability, lack of important information provided, inconsistencies in data and metadata, and licensing issues. When shared effectively, we can multiply the benefits of large datasets that cost large amounts of funds and research time. Combining previously shared biological data accelerates development of analytical methods used to analyze biological data. Reusing rare samples increases the sample impact. Combining data together in meta-analyses increases study power. Data sharing also leads to fewer duplicate studies. Researchers can build on previous studies to corroborate or falsify their findings rather than repeating the same experiment. Many research projects rely on data from resources such as the Encyclopedia of DNA Elements (ENCODE) Project [[8, 9]]. The existence of a large collection of accessible data also aids in the development of cross-cutting analyses such as recount2 [[10]]. Published manuscripts with reusable data will garner more citations and have more long-term impact on scientific knowledge [[11]]. As such, many funders now require that grant proposals include a data management and sharing plan describing biological data and metadata [[12, 13]]. Many journals have also implemented policies making public data sharing a requirement upon publication. Sharing data increases the impact of a researcher's work and reputation for sound science [[14]]. Awards for those with an excellent record of data sharing [[15]] (https://researchsymbionts.org/) or data reuse [[16]] (https://researchparasite.com/) can exemplify this reputation. Demonstrating a track record of excellence in resource sharing benefits you when applying for funding. A commitment to and detailed plan for sharing data publicly increase the perception of a grant proposal's impact [[14]]. A detailed data sharing plan outlines the types of data you will share, available metadata, and in which repositories you will deposit the data. Preparing to share data publicly reduces unintentional errors within your own research group. When preparing the data for sharing, providing detailed metadata and documentation will eliminate guesswork, lost details, and maintain tacit knowledge that might otherwise remain unrecorded. Posting data on public repositories with links to the publication and links to data deposited within your publication ensure findability of your data. Data citation standards now allow directly citing datasets in journal reference list [[17]]. Citable datasets provide an important incentive to data sharing since those using your shared data can now properly attribute citations to your dataset. Despite the clear benefits of sharing data, some researchers still have concerns about doing so. Some worry that sharing data may decrease the novelty of their work and their chance to publish in prominent journals. You can address this concern by sharing your data only after publication. You can also choose to preprint your manuscript when you decide to share your data. Furthermore, you only need to share the data and metadata required to reproduce your published study. Some have concerns about the time it takes to organize and share data publicly. Many add ‘data available upon request’ to manuscripts instead of depositing the data in a public repository in hopes of getting the work out sooner. It does take time to organize data in preparation for sharing, but sharing data publicly may save you time. Sharing data in a public repository that guarantees archival persistence means that you will not have to worry about storing and backing up the data yourself. You can consider putting off data sharing tasks as incurring a form of ‘sharing debt’, by analogy with the concept of technical debt used in software engineering. Delaying these tasks may appear to save you time in the short run, but sharing the data later will take at least as much time as doing it now. You may also incur interest, as it can take longer in the long run to handle individual requests for data availability. Taking a few hours now to organize data and submit it to a repository will save you much of this time. Sharing of data on human subjects requires special ethical, legal, and privacy considerations. Existing recommendations [[18-24]] largely aim to balance the privacy of human participants with the benefits of data sharing by de-identifying human participants and obtaining consent for sharing. Sharing human data poses a variety of challenges for analysis, transparency, reproducibility, interoperability, and access [[18-24]]. Sometimes you cannot publicly post all human data, even after de-identification [[25]]. We suggest three strategies for making these data maximally accessible. First, deposit raw data files in a controlled-access repository, such as the European Genome-phenome Archive (EGA) [[67]. Controlled-access repositories allow only qualified researchers who apply to access the data. Second, even if you cannot make individual-level raw data available, you can make as much processed data available as possible. This may take the form of summary statistics such as means and standard deviations, rather than individual-level data. Third, you may want to generate simulated data distinct from the original data but statistically similar to it. Simulated data would allow others to reproduce your analysis without disclosing the original data or requiring the security controls needed for controlled access [[21]]. Data consist of recorded observations of the biological artifacts or models studied. Metadata describe the primary data and the resources used to generate it. In a biological context, metadata often provide additional information on samples such as sex, disease, and tissue source site. Metadata often include information about resources such as cell lines and antibodies. You should share metadata alongside every dataset. A lack of clear metadata for your specific dataset makes it more difficult to understand [[27]]. This may make it more difficult to reproduce the research or reuse the data. For example, roughly half of >1700 evaluated research studies lacked sufficient specificity in describing resources such as cell lines, organisms, and antibodies to make the study reproducible [[6]]. In addition to information about samples, metadata also describe experimental protocols and bioinformatic processes. These include tools used to generate the data, hardware and software versions, processing batch information, and details necessary for understanding data generation. Most biological disciplines have specific metadata standards that describe the information expected to accompany datasets. For example, genomic researchers have benefited enormously from consistent minimum standard of metadata reporting. The Minimum Information About a Microarray Experiment [[28]] and Minimum Information About a Next-generation Sequencing Experiment [[29]] (http://fged.org/projects/minseqe/) guidelines have enabled large-scale efforts to combine and harmonize data, promoting reuse. These guidelines require descriptive standards, experimental design information, essential sample information such as tissue or sex, and bioinformatic processing protocols. Repositories of gene expression data, such as Gene Expression Omnibus (GEO) [[26]], have mandated of these We metadata standards for individual biological disciplines controlled or can improve the of describing biological in your are controlled that include both and used biological include the Gene used to gene and the Many repositories or projects require the of a controlled in their metadata standard or data For example, the Project using to describe the source of biological The in an further and reusability a controlled For example, a the Gene using a from of controlled and identifiers to to This and makes it easier to of the concept the scientific To to controlled or a and if your data in three with your with in the metadata by and with files provide information about a collection of files with biological data should observations and in details on the experimental and details on bioinformatic processes. We files as essential for making your data easy to we include specific recommendations on files for different types of biological data. should have all needed to reproduce the study in a not data. These include source and analysis for of can help you your data and in a that will make it easier for you and other to access and reuse them. licensing information to your data of others may reuse it. does not to reuse of such as may clear without an license. Many data resources out not to be as reusable as the to lack of in licensing or licensing your data with a that allows reuse and possibly We recommend your data to the public with the the for others to reuse and the data. guidelines recommend [[4, and many journals and repositories require it. For artifacts with your you may to a with more than include the for and Creative Commons for We you to share data a manuscript by the time of publication. Many and such as now make data sharing an In addition to sharing all data by publication some researchers will go further and make it available when a should have access to data and when a manuscript It may to data access so that only can it manuscript but this and the data and public when the manuscript can this with few Posting a preprint of the manuscript at the same time a public record of Researchers data in tabular an to describe similar information in a of and observations and In biological data, observations usually to samples, or consist of or for Researchers often save tabular data as when you have to to a save the data as a with a data at the of the tabular data as allows for of all at Most have that make it easy to and Despite the of with types of data. some gene to This a that the gene to this of gene can When your data have gene and you have about these when When depositing data in public rather than including it in a manuscript or on the data save the data in This with a and observations of with a Many and can easily data. when possible. the of using to when often within This leads to and that to this make data much more and and provide excellent on how to organize tabular data. First, ensure that you the same in all areas of your data. For example, such as and make the data to and to Second, one of data and remain consistent your data and Third, ensure that you consistent such as using in and as this in many analyses these recommendations makes your data easily and by and others. Data have a in your data, the and their When using add a data as a When using add an additional the data data should provide short for a longer for the a for data (such as or and expected minimum and Data can make what future users would otherwise have to about the of data. the tabular data important for your manuscript as a within the manuscript You can more data or data for as data to the Sharing data the manuscript this can have three First, a may the size of data you can Second, may make the data difficult to to many datasets at Third, have data making it difficult to access later or have it a To these share or more tabular data in repositories such as to data in many with many different biological and technical raw genomic data consist of in or When deposit raw data in with files sequence information, similar to but take up much than a or a you should have the to reproduce a from a When deposit your processed data as data or data with genomic as files instead of files genomic of genomic of in the first three The allows additional in making files for with or data, such as experimental from genomic such as and for makes it easy to analyses on your data with software such as or the of when data in track In or for raw data. For storing processed data, information about genomic such as or in large-scale genomic data to the of for files and for of easily Most genomic data have by to a reference for a the reference you your data to in your manuscript and (see of genomic will reference For example, a of of the the and without the reference used for an the genomic some such as not require reference In these make to which reference you used to your repositories make datasets easily by The repository public gene expression and gene data This data on DNA and the and such as The form requires you to both data files and metadata, such as experimental you with an for your You can an on your data to public access publication of your will allow an of up to but you can the at time. that not in the Archive will submit raw data files to on your so you need not submit to data that DNA or rather than data, in These data include sequence of genomic and and the make up of the International which also DNA Data of and European Archive The take data to of these repositories and make it available in the others. For and information from human a controlled-access only qualified researchers you to access the data. dataset have an data access that access requests and of the data [[25]]. genomic data, generate both raw data and processed data. You should share data in a such as or the raw data in their original also share files in the standard data include of and and of amounts for the such as the require raw data and data for data and data in the standard data in this a allows for of without the of share other essential files the data used the These include files with or used for a with metadata about the including sample metadata (such as and technical metadata (such as and experimental design (such as of technical and biological the and Data for to this information in a metadata to describe the the sample processing and the data processing describe all sample processing including analytical information on the bioinformatic tools used to the data, including the and information of the files used for provide the details of and For a describe the tools used and how the differ from the deposited data. this both in a metadata and in a The which the data repositories such as and a for sharing data To submit data to you the data of all files and raw files to their files and results. By makes datasets and you can publication time to make the data You can include a and in scientific manuscripts so that manuscript can still access the data. Most repositories take of data, some on a specific of data. For example, and only of data. Some including the of of publicly available datasets makes of data as it does not require original raw data files available in public will the new to the original data. image data large amounts of also have metadata with datasets. The from many both biological and in or three and time cell also at different and The biological in size and from to preparation image also Different using different with different often using software and to save image data with different and These data sharing challenges We provide guidelines on how to share data and metadata These guidelines have three distinct these guidelines will the of your data in which will increase the impact of your data. share with at least but no information as one can the to the by For large using may provide and access benefits without much biological information often at image or information, but if the only analysis access and only sharing the may the data repositories for dataset size this may as image datasets in size and To maximize the and impact of your also share processed data such as You for the of of or using tools additional analyses will these instead of the raw the repository if it requires image on experimental and you can also apply an image analysis analysis summary data describing the such as These summary data take much than the original You can this summary data or with provide a standard for many Researchers generally such only for image and these include few if you such you should make When doing include important metadata such as the used to the the time and batch (see to tabular To more and a in and instead a image analysis Many software image analysis and including and These tools can many including and a specific of with and cell in an image analysis, in that require additional data processing. In these one lack specificity for the have no biological and the This a as in which biological the and of from image analysis tools To share data, first experimental metadata in a standard metadata provide a others to find and your data Metadata provide guidelines on what metadata to For example, the data standards and such as for metadata and The these guidelines to data sharing standards research have these For example, have guidelines for cell data data of and standards for describing cell and cell To increase the and of your data, your using consistent your data in an repository data repository a and data that and biological sample For example, Data datasets with future data analyses and in additional data Archive from and biological Archive a for all other image often those of a a variety of biological and for research and The of also a variety of and even a for simulated To the repository, your image dataset to the repository with the biological and When in the repository to the of your data. the repositories provide a for all image datasets. your only in a journal or in a or does not the FAIR principles (Box This will in and artifacts and will make future analysis not share data by to or by using links not share data using a (see not to not For sharing image data, we usually not recommend using repositories such as and (see to These repositories data that not have lack the special necessary to the of which consist in data provide the only to For repositories as the to deposit a of different including and kinds of such as and information from distinct raw data using processing The in raw data files and results. As such, scientific repositories for storing the and of these In addition to sharing the an individual to sharing raw and processed data. Data as a repository for of and Most journals require manuscripts to include of the files and metadata, an of a and a This the files and metadata only to the and the You can an on your to public access publication of your or We recommend access at the time of publication. some journals also require files at the time of or by We you to make all your data accessible upon of the directly with your raw data in the for in in the The usually under a For many you can deposit raw data, in the form of in the The important element of sharing data in the of the data can generate of image can or of on the raw data from a can from to the as as and files in This greatly data and reproducibility. It also software access to raw and processed data and approaches to build and the on the of the on your individual and on the approaches providing and to the in a the and in the Data obtaining and When a dataset provide a and deposit all the models and needed to the of the manuscript in In addition, if in a large of add to an with the raw data. As new more data in the on describing will analysis approaches will more For some types of data not no repositories these kinds of data in repositories that can many different types of data. your data with raw data from results. to collect your data so that can individual files in repositories such as To make your data clear and include a with a detailed of the and an of what of the files First, an data repository in the For no such repository we recommend a repository that allows for of data, analysis, and manuscripts and as as a persistence a for scientific including data, and protocols. It also as a repository, you to share data and other by making you publicly your data using a you may For example, some share their data using public links or even a new repository for their a an of complete In efforts usually in with and in for and not the repositories have more the of the original data and provide and cost repositories also metadata standards that FAIR sharing principles (Box These repositories have access to and that individual with repositories and many have by which generate more than We suggest a for biological researchers to complete when a We you to a publication that the necessary for a to a manuscript and artifacts for publication. the as a and add details specific to you and the kinds of data you work We understand that some will find the recommendations difficult or at We you to what you your data management and sharing practices will still provide for you and other the to share We and for on how to share and image data. This work by the and Research of to the to the Research to and the to and the and in the for the and and and

Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.