GigaScience launches: overseeing the transition from papers to “executable research objects”

July 12, 2012

Research papers have been the predominant form of scholarly communication for the past few centuries, and despite moves towards online publication and open access, the process and structure of publication has not fundamentally changed in that time. With biological and biomedical research becoming increasingly data-driven, and the amount of information, computational tools, and code supporting a publication in areas such as genomics and imaging growing at exponential rates, the lack of access to the resources that the paper is built upon is leading to a growing “reproducibility gap”. Recent scandals relating to falsified data that went long undetected (including this particularly egregious recent example of an author fabricating the data supporting at least 172 of their publications) further highlight the need to make data easily accessible for purposes of validation and to maintain public trust in science.

As research has shifted to work within, and handle, this data-rich environment and to utilize advances such as cloud computing and automated workflow systems,  publishing needs to be able to follow in a similar direction. For a number of years much talk has been made about the potential for executable papers: aiding review and re-use of data by having all of the tools and data associated with a publication accessible in a reproducible and standardized environment. An important first step in the path has been reached today with the first articles published in GigaScience.

Aiming to become a home to research from the growing number of biological and biomedical fields handling “big-data“, GigaScience is a new type of open access, open data journal that provides standard scientific publishing linked directly to a database that hosts its relevant data. Our associated GigaDB database, launched last year, provides a home to all of the supporting data and tools associated with research, thus overcoming one of the biggest challenges holding back reproducible research. Through GigaDB, we assign DataCite DOIs to these accompanying datasets to provide additional credit to the authors who make their data publicly available and to boost data discoverability and tracking of data. Data citation is important in incentivizing the effort needed to present data to the public in a usable form, and recent successes by others and us in promoting its use were highlighted in this recent commentary. Our commitment to data DOIs also means that we are well placed to take advantage of the upcoming data citation index announced recently by Thomson Reuters.

Using the data hosting capabilities and expertise in data handling and cloud computing of our partner, BGI, we are able to host a much larger and broader range of datasets than journal supplementary files are usually able to handle, outside the capacities of most other journals and repositories. We have been testing and building up our new informatics platform by releasing a number of datasets from BGI to demonstrate new mechanisms of pre-publication data-release. For example, the deadly 2011 outbreak E. coli dataset was the first we released, and this led to the crowdsourcing of its analysis, which was cited in the recent Royal Society “Science as an Open Enterprise” report as an example of “The power of intelligently open data”.

Now hosting datasets up to 14TB in size (such as this enormous resource of 88 tumor-normal paired genomes) and containing several datasets from articles currently in press (including this cancer single-cell genome data that will be published shortly in GigaScience), we are providing examples of how data and papers can be combined in our launch issue today. Exemplifying GigaScience and GigaDB’s innovative approach is a research article from Stephan Beck’s group at UCL focusing on ways to conduct whole-genome analyses of DNA methylation. In addition to having the raw data available in NCBI, all of the supporting data and software tools needed to recreate the experiments — a total of 84 GB — are freely available for download and reuse under the most open CC0 public domain waiver from GigaDB. We are also maximizing data-interoperability and re-use by encouraging and rewarding authors for providing much richer metadata. For this dataset, metadata are available in ISA-tab format – the first time a journal has taken ISA-compliant submissions (for more, see the recent ISA-commons community paper in Nature Genetics we contributed to).

The GigaDB website is continuing to evolve and the next version will be released in a few months with a more extensive search interface. As the final step in attaining fully executable and reproducible papers, we will be working with authors to make the computational tools and data processing pipelines described in their papers available and, where possible, executable on an informatics platform we are developing with collaborators at the Chinese University of Hong Kong. We hope that by making both the data and processes involved in their analysis freely accessible, this novel form of publication will help articles published in our journal to have a much higher impact in the scientific literature and to maximize their reuse within the community. Check out our two editorials for more on our goals with the database and journal.

As well as this innovative, big-data-driven publication format, the journal also provides reviews and commentaries that address the many hurdles that still need to be surmounted to improve future big-data handling. Many of these are part of our first thematic series (GSC and beyond) covering the best practices in genomics research, published in concert with the Genomic Standards Consortium. In addition to commentaries discussing data-sharing issues in neuroimagingand genomics, our launch issue has a more detailed review tackling data-compression from Guy Cochrane and Ewan Birney at the EBI, a white-paper on DNA collection for large vertebrate genome projects, and two papers on using genomics data to characterize ecosystem and disease outbreaks. We also have a software/technical note paper detailing a novel data format that facilitates the interoperability of bioinformatics tools and an associated commentary from Jonathan Eisen on ‘badomics’ terminology and the explosion of “omes” (good and bad), a topic that readers of his blog will well know.

We hope you enjoy our first set of articles, and please follow our progress here in our blog, our various social media outlets, and on the journal homepage. We would like to thank the BGI and our collaborators at the Chinese University of Hong Kong, British Library, DataCite, and ISA-Tab for their help and support building the data platform. We’d also like to thank our authors, reviewers and fantastic Editorial Board, as well as BioMed Central for their part in getting this issue out. BGI is generously covering the open access article processing charges for the journal’s first year, so please contact us at editorial@gigasciencejournal.com if you have related work you would like to submit to this series or journal; alternatively submit a manuscript here. This week we will be at the ISMB meeting in Long Beach, so come and get hold of us at the BMC booth (#36) if you’d like to meet the editors.