Data management and digitalization
Livestock research is accumulating large amounts of heterogenic data from multiple sources including laboratory data, data from the experimental station and data from an increasing number of sensors that monitor animal parameters. Often such data sources are linked to special formats and proprietary data processing software which makes integration of data from multiple sources particularly difficult.
Accordingly, manual integration of data in research can traditionally take 50 to 80% of the time that is actually spent on the data itself and often leads to data sets that are adjusted to only test hypotheses in the scope of a single trial within a single research institution.
This traditional use of data disagrees with the 3R principle for ethical treatment of animals in experiments, as gathered information could also be used to test hypotheses outside the mentioned scope of a single trial and accordingly reduce the number animals in experiments.
To address this problem, the working group Data management and digitalization works on applying (local) FAIR standards (Fig. 2) to storage and processing of data at the Institute of animal nutrition in a community driven manner.
Local FAIR data:
- Findable: Data and relevant metadata are stored in a local database system using persistent identifiers relevant to the institute.
- Accessible: (Meta)data can be queried using persistent search terms relevant to the institute.
- Interoperable: Data can be combined by metadata based on persistent terms, through links between data and by common persistent identifiers.
- Reusable: Rich description of (meta)data enables combination of independent data or reusing data for hypothesis testing independent of the original context.
This means that an expert group consisting of members from all facilities associated with the Institute of animal nutrition regularly meets to define standards on what data are relevant and what information needs to be conserved. The results of these discussions are then implemented by a core working group consisting of specialized staff.
Harmonizing heterogenous data, making data reusable and accessible outside of trial scopes and reducing time spent on data integration does not only help reducing labor but addresses also ethical treatment of animals in experiments.
Consequent application of FAIR data standards is also useful when depositing data in public repositories and serves as a keystone of modern cooperative research approaches with other research institutes focused on animal nutrition or with other disciplines of (livestock) research.