General Overview

The STELAR project is developing the Asthma-e-Lab, an application of the HeRC e-Lab platform, bringing together Clinicians, Health Informaticians, Statisticians and Data Managers to establish an improved understanding of early life asthma. There is growing evidence to suggest that asthma is an umbrella diagnosis that includes multiple diseases with different underlying mechanisms. It is thought to be unlikely that these different diseases will respond in the same way to therapeutic treatments. The aim of the STELAR project is to perform a statistical analysis across data gathered as part of multiple UK based birth cohort studies to identify clusters of characteristics (or phenotypes) of asthma that may in turn relate to specific endotypes. Identification of these endotypes may then support an improved stratified approach to the treatment of individuals with the disease.

The e-Lab platform can be used to create collaboration spaces that can be accessed in a straightforward manner using a standard web browser. These spaces can be tailored for specific teams, making available tools that include those to manage documents, wikis and blogs and provide social networking capability. e-Labs offer a standard set of tools that can be incorporated into these spaces, but also implement an architecture that allows new software components to be easily incorporated. Users are able to create sharable snapshots of the collaboration space that include detailed metadata about individual content and the relationships between content. These snapshots are managed as Research Object Bundles and provide an important mechanism for aggregating, identifying and annotating sharable content.

The Variable Manager module has been developed to allow the integration of data sets gathered across the five birth cohort studies. These data sets can be imported into the e-Lab and associated with a data description. The descriptions are developed by the data managers of the cohorts and provide detailed information about the imported data. This includes information about how data are represented, the semantics and additional context required for correct interpretation. These data sets can then be integrated in a manner that allows research questions to be asked across multiple data sets, created by different communities. Data are stored in a domain-independent way such that the software can be easily extended and applied in different contexts.

The Job Manager module supports reproducibility of the research findings of STELAR, allowing users to capture research processes in a manner that enables their repetition. Analysis code can be uploaded to the e-Lab and captured in a Snapshot Research Object. The user can then create Job Research Objects that aggregate all of the content required to execute the analysis code, for example the Snapshot Research Object and input data files. Annotations included in the Job Research Object provide the metadata required for execution, including command line parameters, environment variables and the necessary computational resources (e.g. CPU and memory).

The Job Manager can be used to manage the execution of Job Research Objects on compute resources, including moving code and input files to a remote resource, providing the status of processes and moving output files back into the e-Lab. Each completed execution is captured as an Execution Research Object that aggregates the Job Research Object along with the outputs of the process, including output files, standard input, standard output and standard error. The e-Lab can be used to publish Execution Research Objects, associating them with a DOI that can be used to reference the Research Object in publications.