From the Deputy Director for Intramural Research

Data Sharing: Greater Than the Sum of All Parts

BY MICHAEL GOTTESMAN, DDIR

Aristotle’s saying “The whole is greater than the sum of its parts” is as true today as it was in the mid-300s B.C.E., especially when it comes to sharing genomic and other kinds of biomedical data. When scientists can draw on huge repositories of research data, the discovery process is enhanced in ways not possible if they simply added up all the findings of individual studies.

Scientists can explore a wider range of research questions; improve the reproducibility and validation of research results; dig deeper into research questions than perhaps was done by the originators of the data; and advance the innovation of methods and tools for research. As a matter of principle and policy, NIH has long had an understanding with its investigators that data obtained using public monies would be widely shared with responsible individuals to accelerate the discovery process.

Although NIH and NIH-funded scientists are already required to share many types of biomedical data, until recently there has not been an explicit requirement for the development of data-sharing plans or the tools to implement these plans. The NIH as a whole, and the intramural program specifically, now have policies that cover both the sharing of genomic data (including human genomic data and data from all model systems) and the more general sharing of data derived from human subjects research or human specimens. These policies were recently sent to all principal investigators at the NIH and can be found at the following websites: http://gds.nih.gov and http://oma1.od.nih.gov/manualchapters/intramural/3016, respectively.

You will be hearing directly from your scientific director or the person designated within your institute or center (IC) to be the point of contact on these data-sharing issues about your requirements for creating data-sharing plans under the Genomic Data Sharing (GDS) policy, which went into effect on August 31, 2015. Please also review the memos and NIH Policy Manual chapters you have already received about data sharing. In brief, the GDS policy contains the following basic requirements.

Sharing Human Data. IRP investigators must submit covered human genomic data and associated data (such as phenotype and exposure) to an NIH-designated repository. As was the long-standing practice under the NIH Genome Wide Association Studies (GWAS) Policy, NCBI’s database of Genotypes and Phenotypes (dbGaP) will, in most cases, be the repository, although alternative repositories can be approved by your IC, and rare exceptions to this requirement can be granted by the DDIR.

Registration. All human genomic data studies covered by the GDS policy must be registered in dbGaP by the time that data-cleaning and quality-control measures begin, regardless of which data repository will ultimately house and distribute it. Investigators should work with their designated genomic program administrator (GPA) to initiate the registration process.

De-identification. Before sharing, data should be de-identified according to current requirements.

Informed Consent and IRB Review. An institutional review board (IRB) will review study protocols and informed-consent materials to determine whether it is appropriate for data to be shared for secondary research use.

Institutional Certification. As was the practice under the NIH GWAS Policy, each IC must complete an institutional certification as part of the data-submission process. This document identifies whether human genomic data are being submitted for unrestricted or controlled-access use and ensures that certain provisions to protect the interests of research participants are satisfied. The institutional certification should be completed by the investigator and signed by the scientific director. Whenever possible, the scientific director, or delegate, should send the institutional certification memorandum to the IC genomic point of contact before research is begun. (For more information, go to http://gds.nih.gov/Institutional_Certifications.html)

The human data-sharing (HDS) policy requires the development of data-sharing plans by October 1, 2015 (see chapter 3016 of the NIH Policy Manual at http://oma1.od.nih.gov/manualchapters/intramural/3016). Once again, your scientific director will explain the requirements, which initially will consist of data-sharing plans for all projects beginning preresearch scientific review after October 1, 2015.

The broader sharing of human data derived from our human subjects research at the NIH requires fastidious attention to the requirements of protecting privacy and consistency with all informed consent documents, as described in the NIH Policy Manual chapter 3016 on HDS. At this time, a data-sharing plan for clinical trials might well consist of sharing with external collaborators and submitting data to ClinicalTrials.gov (as is required for all clinical trials conducted at the NIH), but for natural history trials and other clinical investigations, additional repositories may be necessary. We are exploring the possibility of adapting the Biomedical and Translational Information System to enable HDS as appropriate.

Please work closely with your scientific director and your IC contacts for further information and to determine whether your data are covered by the GDS and/or HDS policies. Thank you for ensuring that the very valuable data you obtain will be shared appropriately with the wider scientific community.