A Pipeline Based Approach for Experimental Neuroscience Data Management

The field of neuroscience is witnessing a huge influx of experimental data thanks to the improvements in the data acquisition tools and techniques. Most of this data is being collected by thousands of experimenters located in various institutions around the world. There is also a growing interest in using this experimental data for building biologically realistic computational models of the neuronal systems. One such project i.e. the Blue Brain Project is developing a bottom-up neuronal modeling and simulation framework using experimental data collected at laboratories worldwide. However, to use the experimental data effectively in computational neuroscience research, the data needs to be annotated with metadata such as details of the experimental protocols and conditions, animal subject used etc. Most of the platforms for experimental neuroscience data management operate under the assumption that the primary data has been validated, and properly annotated before it is uploaded to the system. Thus putting the responsibility of validating, and annotating the experimental data with the experimenter conducting the study. Consequently, most experimenters maintain their own data using ad-hoc systems with non-standard metadata schemes; and only upload the final set of data to the data management platform. As a result, the metadata is usually incomplete, and does not always comply with the requirements of the data management platform, negatively impacting the reusability of their data. The current thesis work explored the question of the experimental data management within the context of the Blue Brain Project, and designs a data management pipeline catering for the acquisition, annotation and validation of the experimental data and associated metadata, in order to ensure that the data is usable for the computational modeling purposes. The pipeline enforces a data review process whereby incoming experimental data was made to go through a series of steps, analogous to the scientific review process, systematically improving the quality of the experimental data and associated metadata.