A strategy to establish Food Safety Model Repositories

Abstract

Transferring the knowledge of predictive microbiology into real world food manufacturing applications is still a major challenge for the whole food safety modelling community. To facilitate this process, a strategy for creating open, community driven and web-based predictive microbial model repositories is proposed. These collaborative model resources could significantly improve the transfer of knowledge from research into commercial and governmental applications and also increase efficiency, transparency and usability of predictive models. To demonstrate the feasibility, predictive models of Salmonella in beef previously published in the scientific literature were re-implemented using an open source software tool called PMM-Lab. The models were made publicly available in a Food Safety Model Repository within the OpenML for Predictive Modelling in Food community project. Three different approaches were used to create new models in the model repositories: (1) all information relevant for model re-implementation is available in a scientific publication, (2) model parameters can be imported from tabular parameter collections and (3) models have to be generated from experimental data or primary model parameters. All three approaches were demonstrated in the paper. The sample Food Safety Model Repository is available via: http://sourceforge.net/projects/microbialmodelingexchange/files/models and the PMM-Lab software can be downloaded from http://sourceforge.net/projects/pmmlab/. This work also illustrates that a standardized information exchange format for predictive microbial models, as the key component of this strategy, could be established by adoption of resources from the Systems Biology domain.

Towards transparent and consistent exchange of knowledge for improved microbiological food safety

“…However, some examples can be pointed out. Recent studies on the creation of food safety model repositories for predictive microbial models [45 ] are likely to provide tailor-made solutions also for this research domain. In November 2016, EFSA launched Knowledge Junction, a curated, open repository that makes use of Zenodo.…”

“…The documentation of the fitting procedure is also important. Recent examples for fitting procedures with different tools and fitting approaches can be found in Plaza-Rodriguez et al [45 ]. The option to provide Supplementary materials in almost every journal in the field should encourage the exchange of knowledge.…”

“…In the last years, the application of existing XML-based information exchange formats for the microbiological food safety modelling domain has been proposed. Specifications were provided by Plaza-Rodriguez et al [45 ], which promoted the consistent adoption of existing standards (NuML/SBML/SED-ML/OMEX) in order to create a common description language for predictive microbial models (Predictive Modelling in Food Markup Language-PMF-ML). PMF-ML has already been used to provide parameterized predictive microbial models to journals [50] and has facilitated the direct information exchange between different software tools from the domain of predictive microbiology (ComBase, GroPIN to PMM-Lab; PMM-Lab to R).…”

“…The creation of standard-based publicly available knowledge repositories could promote this knowledge sharing. Specifically, the development of harmonized data formats, controlled vocabularies and open source software code libraries, as suggested by Plaza-Rodriguez et al [45 ], will facilitate knowledge first, transparency and consistency and second, exchange between software tools in predictive microbial modelling and QMRA.…”

Harmonized terms, concepts and metadata for microbiological risk assessment models: The basis for knowledge integration and exchange

“…Etymologically, metadata is "data about data". To facilitate efficient exchange of data, QMRA and PM models, it is crucial to provide a structured set of metadata that can be used by modellers or data providers to annotate their model or data set (Plaza-Rodríguez et al, 2015). The metadata will define the extent to which existing knowledge can be re-used in a transparent way.…”