Interfacing with EBS Analytical Services
The EBS is development team is collaborating with an ILCI project being led by Kelly Robbins at Cornell. The ILCI team developed the following model for integrating BMS phenotypic data with GOBii genotypic data for submission to the EBS analytical pipeline services being developed by the EBS team.
Chart: ILCI analytics flowchart v0.1
The ILCI project team held meetings with the CIWA breeding leaders to understand the requirements from a user perspective for the process. An important outcome of those discussions was an agreement to use the means (BLUEs) and error variances from Single-Site Analyses (SSA) as the inputs for the pipeline rather than the raw data. This has three benefits: 1) it simplifies the process from the pipeline’s perspective; 2) it will provide an important data quality control step to the process, ensuring that the inputs have been reviewed by the breeders and deemed to be good quality data; and 3) it will enable the pipeline to run faster, returning results to the user quicker.
To support acquisition of the data from the BMS as input to the pipeline, a set of API calls were developed in the BMS. These calls provide easily accessible and efficient access to all of the data needed to run the pipeline. In designing the calls, the BMS developers conferred with the BrAPI community and solutions adopted by other projects for working with statistical results data. The calls are demonstrated in the SSA results tab added to the BMS in v22:
The API calls are accessible externally following authentication and acquisition of an authorization token following the standard procedure used for BrAPI calls. The calls and documented use cases will be presented to the BrAPI community as proposed standards for inclusion in a future version of the BrAPI standard.
….
The ILCI team is currently working on developing a User Interface (UI) for selecting studies for inclusion in the analysis. The interface will be based on the Graphical Filtering BrAPP that was developed as part of the BreedBase project and since adapted for the BMS:
The tool will be extended to simplify the process of selecting inputs to the pipeline. Initially, a studies based approach will be implemented, allowing the user to select a set of studies for inclusion. A second approach in which a list of germplasm entries is used as the input, pulling in all studies in which the entries have been included.
The final step in process of analyzing data using the Analytical Pipeline service will ultimately be to provide a mechanism for the user to save the results back to the BMS for making breeding decisions such as whether to advance or discard a given germplasm. A model for how this could be implemented is the Decisions Support BrAPP introduced in the BMS in v22.