PlatformTM, a standards-based data custodianship platform for translational medicine research

Abstract

Biomedical informatics has traditionally adopted a linear view of the informatics process (collect, store and analyse) in translational medicine (TM) studies; focusing primarily on the challenges in data integration and analysis. However, a data management challenge presents itself with the new lifecycle view of data emphasized by the recent calls for data re-use, long term data preservation, and data sharing. There is currently a lack of dedicated infrastructure focused on the 'manageability' of the data lifecycle in TM research between data collection and analysis. Current community efforts towards establishing a culture for open science prompt the creation of a data custodianship environment for management of TM data assets to support data reuse and reproducibility of research results. Here we present the development of a lifecycle-based methodology to create a metadata management framework based on community driven standards for standardisation, consolidation and integration of TM research data. Based on this framework, we also present the development of a new platform (PlatformTM) focused on managing the lifecycle for translational research data assets.

Fig. 1

Research data life cycle highlighting in detail the different stages of the data pipeline between data collection and data analysis, the scope for the proposed data custodianship environment.

Fig. 2

Our proposed data lineage management workflow. Each stage has its own data form, data service (top blue boxes) and data storage resource (bottom grey boxes). Decisions about what data to collect start with the formulation of research questions (modelled as data elements) and culminate in data collection to collect values for specific data elements to produce raw data files. Files are then semantically and structurally annotated by dataset descriptors and consolidated into primary datasets. Content from all datasets is then integrated according to a common observation model each observation semantically defined by an observation descriptor. Finally, user queried data is extracted and saved to analysis-ready annotated datasets.

Fig. 3

Translational Research Metadata Framework (TREMF). The domain model (L4) defines the common elements of a translational research project and the relations between them establishing context for exploring data and cross-study comparisons. Different activities (clinical or assays) within a project generate datasets that are modelled according to generic interoperable meta-model (L3). Dataset content (observational data) is modelled against the common observation model (L2): a vector of related data elements each defined according to the ISO/IEC 11179 data definition model part of the standard model for metadata registries (L1).

Fig. 4

Dataset four-layer metamodel hierarchical architecture for dataset interoperability. Data at M0 level, such as subject demographics, are described by models at the M1 level, such as CDISC SDTM dataset format, which in turn are described by metamodels at the M2 interoperability level, such as subject dataset descriptor, which in turn conforms to a generic dataset descriptor model described in M3 layer.

Fig. 5

Metadata Governance module. For each project activity, a standard-based predefined dataset descriptor is created to define metadata for a primary dataset. First users can browse and search through all preloaded templates. Once a template is selected, user can then customize the structure of the dataset to fit their data.

Fig. 6

Project Drive. This module organizes all uploaded project files and manages the loading process into the platform’s databases.

Fig. 7

Datasets module. Browse and download project’s consolidated datasets.

Fig. 8

Data explorer and query interface. (a) Subject Panel, (b) Clinical data panel, (c) Molecular observations panel. Each panel lists observation features (metadata) on the left and data plots for each clicked observation on the right. Filtering data through the plots, subjects and samples satisfying the filters are automatically updated on top. Clicking on the cart icon in the top right corner lists all observations selected with an option to checkout and generate analysis-ready dataset for the queried data.