I realize I haven’t written a post in over a month, and I feel horribly guilty about it. The one good thing about not having the time to write blog posts frequently is that I now have a stockpile of ideas, and plenty of material to write more frequent posts.
What I would like to address in today’s post is some of the ongoing efforts from journals, government agencies, and open source communities have taken to address the need to publish data, in all of its messy and intricate formats. Similar to my previous posts, I will describe each of the efforts that I find to be promising in terms of their ability to tackle this massive, and complicated task. In case readers are unfamiliar with the concept of a data publication, I define the concept based on a hybrid of different viewpoints from papers by Borgman, Lynch, Reilly et al., Smith, and White:
A data publication takes data that has been used for research and expands on the ‘why, when and how’ of its collection and processing, leaving an account of the analysis and conclusions to a conventional article. A data publication should include metadata describing the data in detail such as who created the data, the description of the type of data, the versioning of the data, and most importantly where the data can be accessed (if it can be accessed at all). The main purpose of a data publication is to provide adequate information about the data so that it can be reused by another researcher in the future, as well as provide a way to attribute data to its respective creator. Knowing who creates data provides an added layer of transparency, as researchers will have to be held accountable for how they collect and present their data. Ideally, a data publication would be linked with its associated journal article to provide more information about the research.
Nature Publishing Group – Scientific Data
Scientific Data is the first of its kind in that it is an open access, online-only publication that is specifically designed to describe scientific data sets. Because the description of scientific data can be a complicated and exhaustive, this publication does an excellent job of addressing all of the questions that need to be asked of researchers before they even think of submitting their data. Scientific Data just came out with their criteria for publication today, and the questions they ask are exactly what is needed to ensure that the data publication will be able to be reused through appropriate description.
Then comes the next great component – the metadata. Scientific Data uses a ‘Data Descriptor’ model that requires narrative content about a data set such as the more traditional descriptors librarians are familiar with such as Title, Abstract and Methodology. What is excellent about the Data Descriptor model is that it also requires structured content about the data. This structured content uses the an ‘Investigation’, ‘Study’ and ‘Assay’ (ISA) open source metadata format to describe aspects of the data in detail. These major categories are apparently designed to be ‘generic and extensible’, and serve to address all scientific data types and technologies. You can check ISA out HERE.
Overall I think that Scientific Data is the beginning of a new trend in publishing where major journals will begin to publish data publications more frequently on top of traditional research articles. This publication is the first step towards making research data available, reusable and transparent within the scientific research community.
F1000Research – Making Data Inclusion a Requirement
F1000Research is an excellent new open science journal that has caught my attention for its foray into systematic reviews and meta analyses and for its recent ‘grace period’ to encourage researchers to submit their negative results for publication. I think that this publication that medical librarians should be aware of, and potentially encourage researchers to submit to should they be looking for a more frugal option. What really impresses me with F1000Research though, is their commitment to ensuring that data associated with research articles is made readily available.
Currently, F1000Research reviews data that is submitted in conjunction with an article, and then offers to deposit the data on the authors behalf in an appropriate data repository. The journal is open to placing in data in any repository, but they work mainly with figshare - a popular platform for sharing data. Together figshare and F1000Research have created a ‘data widget’ that allows figshare to link data files with its associated article in F1000Research – which is excellent! There was a recent blog post written about this widget here that can give it the attention it deserves: http://blog.f1000research.com/2013/05/23/new-f1000research-figshare-portal-and-widget-design/). F1000Research is also apparently working on a similar project with Dryad. I think that moving forward we will see more efforts from journals like F1000Research to seamlessly connect their publications with associated data. This is a crucial component to publishing data as the journal article provides the context in terms of how the data was used.
Dryad – Integrated Journals
Dryad is a data repository and service that offers journals the option of submission integration with their system. The service is completely free and is designed to simplify the process of submitting data, and ensure biodirectional links between the article and the data. Currently Dryad provides an option for data to be opened up to peer review, but I would like to see that become more of a requirement going forward. Here is a link to Dryad’s journal integration page: http://datadryad.org/pages/journalIntegration
Currently there are a number of journals currently participating in this effort, and a complete list of them can be seen HERE. Carly Strasser also did a great job of outlining other journals that require data sharing in her post about data sharing on the excellent blog Data Pub. I think Dryad is a perfect example of the other side of traditional publishing. We need data repositories like Dryad and figshare to continue supporting data publication and storage, as they represent half of the picture that will allow articles and data to be connected.
The Dataverse Network
The Dataverse Network is a data repository designed for sharing, citing and archiving research data. Developed by Harvard and the Data Science team at the Institute for Quantitative Social Science, Dataverse is open to researchers in all scientific fields. As a service, Dataverse organizes its data sets into studies; each study contains cataloguing information along with the data, and provides a persistent way to cite the data that has been deposited.
Dataverse also uses Zelig (an R statistical package) software that provide statistical modeling of the data that is submitted. Finally, Dataverse can also be installed as a software program into their own institutional data repositories. I see the ability to download Dataverse for institutional purposes to be an excellent prospective strategy; as more academic institutions begin to develop data storage capabilities to their institutional repositories, Dataverse will provide some much needed assistance in this arena.
GitHub: Git for Data Publishing
Although I would not call myself an expert of the GitHub world, I will say that I recognize a fruitful initiative to publish data when I see one. In a recent blog post by James Smith talking about how the tools of open source could potentially revolutionize open data publishing. The post is great and you can read it here: http://theodi.org/blog/gitdatapublishingutm_source=buffer&utm_medium=twitter&utm_campaign=Buffer&utm_content=buffer6c57f James’ idea is to upload data to GitHub repositories and use a DataPackage to attach metadata that will sufficiently describe the data. Ultimately the goal of using GitHub for data publication would enable sharing and reuse of data within a supporting and collaborative community. While some of this can get complicated, working through the links from his post really provides you with a sense of how an open source community is coming together to address the need to publish data.
Biositemaps is a working group within the NIH that is designed to:
(i) locating, (ii) querying, (iii) composing or combining, and (iv) mining biomedical resources
‘Biomedical resources’, in this case can be defined as anything from data sets to software packages to computer models. What is most interesting about Biositemaps is that they provide an Information Model that outlines a set of metadata that can be used to describe data. Using the Information Model as a base for data description, it then uses a Biomedical Resource Ontology (BRO); BRO is a controlled terminology for the ‘resource_type’, ‘area of research’, and ‘activity’ to help provide more information about how data is used, and how it can be described in detail using biomedical terminology. I will admit this resource is still pretty raw, but I think it has a lot of potential for being an excellent resource moving forward. The basic idea behind Biositemaps is that a researcher fills in a lengthy auto-complete form describing themselves, their data, and the methodology used to create the data. Once the form is complete, it produces an RDF file that is uploaded to a registry where it can be linked to, and from anywhere. If you are a medical librarian and you have researchers interested in publishing data, I encourage you to take a look at this resource.
SHARE Program – Association of Research Libraries (ARL), Association of American Universities (AAU), the Association of Public and Land-grant Universities (APLU)
This effort just came out last week, but the ARL, AAU and APLU are joining together to create a shared vision of universities collaborating with the Federal government and others to host institutional repositories across the the memberships to provide access to public access research – including data. While it is not entirely clear how this will be achieved – especially in the realm of data – I think that this is the type of collaboration that will provide a well researched, evidence based solution moving forward. I hope that SHARE continues to expand beyond the response to the OSTP memo, as I think Canadian academic institutions could benefit greatly from this effort. Here is a link to the development draft for SHARE: http://www.arl.org/storage/documents/publications/share-proposal-07june13.pdf
For Medical Librarians
My goal in presenting these data publication efforts is an attempt to get medical librarians to think more about the options that are available for data publication. Journals, government agencies and open source communities are all trying to address the issues surrounding data publication, and I think it is our duty as medical librarians to familiarize ourselves with journal policies around data sharing; data publication initiatives like DataCite, Dryad, and figshare; and new government efforts like Biositemaps that are becoming more heavily used every day, and will be relevant for our liaison and research areas of practice moving forward. I have tried to provide a lot of links within this post, but I’ve included some more reading below that may be useful. I’d also like to mention that this is by no means an exhaustive list, but rather some of the interesting efforts i’ve seen throughout my work with data. Please feel free to add as you wish in the comments section.
1. Borgman CL, Wallis JC, Enyedy N. Little science confronts the data deluge: habitat ecology, embedded sensor networks, and digital libraries. International Journal of Digital Libraries [Internet]. 2007;7:17–30. Available from: http://escholarship.org/uc/item/6fs4559s#
2. Lynch C. The shape of the scientific article in the developing cyberinfrastructure. CT Watch Quarterly [Internet]. 2007;3(3):5–10. Available from: http://www.ctwatch.org/quarterly/articles/2007/08/the-shape-of-the-scientific-article-in-the-developing-cyberinfrastructure/
3. Piowowar H, Chapman W. A review of journal policies for sharing research data. Nature Precedings [Internet]. 2008. Available from: http://www.academia.edu/904922/A_review_of_journal_policies_for_sharing_research_data
4. Reilly S, Schallier W, Schrimpf S, Smit E, Wilkinson M. Report on Integration of Data and Publications [Internet]. 2011: p. 1–7. Available from: http://www.alliancepermanentaccess.org/wp-content/uploads/downloads/2011/10/ODE-ReportOnIntegrationOfDataAndPublications-exesummary.pdf
5. Smith VS. Data publication: towards a database of everything. BMC research notes [Internet]. 2009 Jan [cited 2013 Mar 3];2:113. Available from: http://www.pubmedcentral.nih.gov/articlerender.fcgi?artid=2702265&tool=pmcentrez&rendertype=abstract
6. Whyte A. IDCC13 Data Publication: generating trust around data sharing. Digital Curation Centre [Internet]. 2013 Jan 23; Available from: http://www.dcc.ac.uk/blog/idcc13-data-publication-generating-trust-around-data-sharing