Metacuration Standards and Minimum Information about a Bioinformatics Investigation

Abstract

Many bioinformatics databases published in journals are here this year and gone the next. There is generally (i) no requirement, mandatory or otherwise, by reviewers, editors or publishers for full disclosure of how databases are built and how they are maintained; (ii) no standardized requirement for data in public access databases to be kept as backup for release and access when a project ends, when funds expire and website terminates; (iii) the case of proprietary resources, there is no requirement for data to be kept in escrow for release under stated conditions such as when a published database disappears due to company closure. Consequently, much of the biological databases published in the past twenty years are easily lost, even though the publications describing or referencing these databases and webservices remain. Given the volume of publications today, even though it is practically possible for reviewers to re-create databases as described in a manuscript, there is usually insufficient disclosure and raw data for this to be done, even if there is sufficient time and resources available to perform this. Consequently, verification and validation is assumed, and claims of the paper accepted as true and correct at face value. A solution to this growing problem is to experiment with some kind of minimum standards of reporting such as the Minimum Information About a Bioinformatics Investigation (MIABi) and standardized requirements of data deposition and escrow for enabling persistence and reproducibility. With easy availability of cloud computing, such a level of reproducibility can become a reality in the near term. Through standards in meta-curation and minimum standards of reporting that uphold the tenets of scientific reproducibility, verifiability, sustainability and continuity of data resources, the knowledge preserved will underpin tomorrow's scientific research. Other issues include disambiguation of authors or database names, and unique identifiers to support non-repudiability, possibly in multiple languages. The International Conference on Bioinformatics and its publications are now in the process of making attempts at addressing these issues and this presentation will highlight some of the current efforts.