All about data sharing | Nature Cancer

"Making data available is essential for validating and furthering scientific discoveries. Helping authors navigate whether, how and in what form to share the data is also essential....

Over the past several years, many major funding agencies, including the National Institutes of Health and National Science Foundation in the USA, the European Research Council, and the Wellcome Trust, Cancer Research UK and the Research Councils in the UK, have adopted more stringent data-sharing requirements for their grantees. Many publishers also have relevant mandates, including the Nature Portfolio journals. However, the days when a dataset could be provided within the few pages of a scientific publication are long gone. Especially in multidisciplinary fields such as cancer research, the results reported in one paper are, more often than not, supported by several distinct and complex datasets, from classic cell and molecular biology experiments and animal work, to large-scale ‘-omics’ data and analyses of clinical samples obtained from human participants. Navigating whether, how and in what form to share the data associated with a paper can be complicated....

Making data available to readers without undue qualifications is a condition for publication in Nature Cancer. Sharing the data should be the norm unless there are justifiable restrictions on availability, which we ask authors to disclose at the time of submission. If these restrictions are found to be unduly prohibitive, we may decline further consideration of the manuscript. As a general rule, we advise authors to ensure that all relevant data are available either within the manuscript files or through deposition in appropriate public repositories. For certain data types, we mandate deposition and ask authors to provide access to editors and referees before we send a manuscript to peer review. These include gene expression, DNA and RNA sequencing and proteomic data; nucleic acid and protein sequences; and macromolecular structures, genetic polymorphism and linked genotype and phenotype data. We also strongly encourage public provision of other data types for which community-endorsed repositories exist, such as metabolomics and imaging data. In cases for which discipline-specific repositories do not exist, we recommend the use of unstructured, generalist repositories such as Figshare, Dryad and Zenodo. In general, we may require deposition in a public repository of any dataset and data type that is deemed to be central to the main message of the study or essential for reproducibility of the reported findings. To help authors identify the appropriate public repository for their data, our sister journal Scientific Data maintains a dedicated list of approved and recommended data repositories. Apart from the deposition of datasets before peer review, we also advise inclusion with the submitted manuscript files of unprocessed images of gels and blots, and of all raw numerical data behind graphs and statistical analyses. We ask authors to include these source data files when we invite a revision and ultimately require them for publication of the study...."