Do you think Dash/DataShare meets these AHA Open Data repository requirements?

[MeganLaurance]

I have a researcher at UCSF who is submitting a grant proposal to the American Heart Association in a week and I would like to recommend that he list DataShare (Dash) in his Data Sharing Plan. I'm looking for an honest opinion as to whether or not DataShare meets their requirements for a data repository. Any comments?

AHA Information

Acceptable Repositories

Given the wide range of projects funded by the AHA, no single deposit location is universally applicable. Instead, the AHA provides its grant recipients with a degree of latitude in selecting the most appropriate repository to house their research data. In order for a repository to be approved by the AHA, it must be deemed appropriate across each of the following dimensions:

• Re-Use. The repository must allow any interested party to freely access the data without restriction on research reuse. This should be codified in the repository’s terms of use.
• Security. The repository must describe how datasets are stored, as well as how any confidential information is protected.
• Stability. The repository must have a clearly articulated funding mechanism or business plan to provide reasonable assurances that the data will be available for the indefinite future. It should also have a continuity plan addressing what will happen to the data in the event the repository is discontinued.
• Fee Structure. What is the cost, if any, to deposit data in the repository? Is the payment one-time or recurring? Does the size of the dataset impact the cost? The repository must define its rates and explain how these fees ensure financial stability.
• Subject Focus. There are hundreds of topic-specific repositories in operation at this writing. The grant recipient should endeavor to deposit his/her data in a repository that is appropriate for the subject matter in question. Further, if a repository has emerged within a specific research community as the default resource in that field (e.g., GenBank for DNA sequences), grant recipients should, as a general rule, utilize that repository. This optimizes the ability of others to discover and build upon the data.
• Metadata. The repository must require a depositor to provide sufficient metadata to enable the dataset to be used by others. These metadata should be searchable so that repository visitors can easily discovery appropriate datasets.
• File Formats. The repository should be able to accommodate all aspects of the grant recipients’ dataset, regardless of file type.
• Machine Extraction. The data stored in the repository should ideally be available in a machine-readable and machine-interpretable format.
• Willingness to Accept AHA Data. Finally, the repository must be willing to accept data submitted by AHA-funded researchers.

[cpwillett]

How would you like us to respond to the questions?

[MeganLaurance]

I have a preference for comments to be included here in GitHib so that the thread is viewable by all. But if you want to address it as a group and then respond via e-mail that's OK too. Is that what you meant Perry?

[pcruse]

Agree!

From: MeganLaurance [mailto:notifications@github.com] Sent: Wednesday, July 16, 2014 11:45 AM To: CDLUC3/dash Subject: Re: [dash] Do you think Dash/DataShare meets these AHA Open Data repository requirements? (#26)

I have a preference for comments to be included here in GitHib so that the thread is viewable by all. But if you want to address it as a group and then respond via e-mail that's OK too. Is that what you meant Perry?

— Reply to this email directly or view it on GitHubhttps://github.com/CDLUC3/dash/issues/26#issuecomment-49209011.

[MeganLaurance]

Update: I'll be drafting some text for this researcher about the option of using DataShare to comply with AHA's open data policies, and will get that text to him tomorrow (because -of course- his grant is due next Wednesday.....). Any comments you guys might have about where we are not compliant with their data repository requirements and when we might be compliant would be great. For example - even if we have a ballpark time frame for when the Fee Structure model will be official, or if there is existing wording available on the Stability of Merritt, etc I'll include that.

We are focusing on this portion of their FAQ that does give some leeway with respect to listing "catch-all" repositories and institutional repositories in the data plan portion of the grant, even if they are not on the AHA's list of approved repositories.

AHA FAQ: What if no appropriate repository exists to house my research data?

No one will be expected to develop a de novo database. The list of AHA-approved data repositories is growing rapidly and we anticipate that it will grow to cover a wide spectrum of research areas. If you do not see a repository on this list that is an appropriate home for your data, we recommend the following next steps:

Go to the Registry of Research Data Repositories and look for a subject-appropriate match. NOTE: Not all of the repositories listed in this database will meet AHA standards. If you identify a subject-appropriate repository, please check its policies against the AHA's criteria for acceptable data repositories. If you feel that the identified repository fits the AHA's guidelines, you can submit the repository as “other” in the data plan portion of your application and AHA will review it.

If you cannot identify an appropriate subject repository, consider depositing your data in your institutional repository (IR). Many universities have developed IRs to house works created by their affiliated researchers. The OpenDOAR Directory provides a searchable database of institutional repositories – you can query it to see if your university has an IR. As with the Registry of Research Data Repositories, not all of the repositories listed in this database will meet AHA standards. Please check your IR’s policies against the AHA's criteria for acceptable data repositories. If you feel that the IR fits the AHA's guidelines, you can submit the repository as “other” in the data plan portion of your application and AHA will review it.

Our open data policy is new and we are not able to make recommendations to applicants for repositories other than the AHA-approved list above. We will begin vetting additional repositories that are recommended by our applicants when the applications for this current cycle are submitted. We are currently in the process of reviewing some "catch-all" repositories, including Zenodo, figshare, and Dataverse. All of these will accept image sets. Based on the volume of applicants that were not able to locate subject-appropriate appropriate repositories, we might need to include one or more of these catch-all repositories. If you cannot locate a subject-specific repository using the steps above, you can consider entering one of these catch-all repositories as “other” in the data plan portion of your application.

[slabrams]

Hi Megan:

Here are my responses to the specific AHA requirements: Re-use: The UCSF DataShare collection is configured to allow public access to all materials. This is implied by the online terms of use and explicitly stated in the FAQ. Security: The document http://www.cdlib.org/services/uc3/docs/UC3-Merritt_Preservation.pdf provides summary details of the Merritt infrastructure, including its storage and security mechanisms. Stability/Fee Structure: Merritt is a core service function of the CDL. Currently its operation is fully funded as a CDL budget line. CDL is working with the UCOP budget office for final approval of a price structure based on recovering storage costs at a price point anticipated to be $0.65/GB/year. All content managed in DataShare/Merritt at the time this pricing goes into effect would be grandfathered and not subject to the price model. Subject Focus: As stated in the terms of service and FAQ, the DataShare repository can be used by any UCSF-affiliate for any purpose related to education, research, and public service. Given the nature of UCSF's academic programs, the majority use of DataShare is assumed to be in the biomedical area. Metadata: DataShare provides an opportunity for researchers to describe their data in scientifically meaningful terms using a core subset of the DataCite schema. File Formats: DataShare and Merritt can accept data in any format. Machine Extraction: All data and metadata in Merritt are amenable to automated extraction through public APIs. Willingness to Accept AHA Data: DataShare is open to any data contribution that complies with its terms of service.

[MeganLaurance]

Thanks Stephen! This is very very helpful. I appreciate the quick response.