Integration of Data Management Approaches in Pre-Clinical Laboratory Workflows

Most efforts in the neuroscience community to organise data and perform standardised analyses have concentrated on complex multi-dimensional data such as in in-vivo recordings and human MRI data sets. Only recently have first attempts started to standardise "simpler" data sets such as electrophysiological recordings in-vitro, as they are often found in pre-clinical studies. Unfortunately, these tools are often not easily approachable for scientists without a coding background and can appear intimidating or require proprietary software, potentially causing deviations from recommended standards. Variations between laboratories and their protocols used introduced further challenges to find common ground for data management and analyses. This is particularly difficult in fields such as cellular electrophysiology, where over the last decades vast amounts of acquisition hardware and software have been developed, resulting in highly diverse data strategies. Following FAIR guidelines is a crucial part of improving reproducibility and transparency in the scientific field. Successful implementation of the guidelines requires research data management (RDM) strategies and an understanding of how RDM can facilitate personal analysis and data sharing capabilities, improving collaboration within and beyond one's own laboratory. Members of the laboratory participated in surveys and 1-on-1 interviews to assess the requirements of scientists and their project's needs and workflows to find common ground and identify existing common standards. To reduce the barrier of following standards in practice, a tool with a low entry barrier and high personal benefit was developed with the help of users. First, existing standards, challenges and analytical needs were discussed. In addition, the workload associated with analysis steps and data organisation was assessed through a questionnaire. Interview feedback was used to inform further development steps of workflows. Agreed standards included methods of documenting experiments and analysis, metadata, and tools to work on the data. Improved accessibility and reusability were achieved by implementing workflows that utilised established open-source data types such as Neurodata without borders (NWB). Convergence on a single data format simplifies the exchange of data within the laboratory and meets requirements of data repositories. Based of previously agreed standards for metadata collection and structures, guidance was provided on how to convert measurement data to NWB. This guidance was partially provided by a Python based laboratory community open-source tool with conversion, metadata input, interface to an electronic lab notebook (ELN), and analysis functionality which integrates already existing data pipelines and makes them available to others. Scientists actively contributed to the development of functionality by being either directly involved in the coding or by providing crucial feedback. As further support, standards and workflows are communicated to new members of the laboratory as part of "data onboarding", facilitating access to analysis pipelines and documentation practices and strengthen established standards. In conclusion, the integration of different RDM concepts into standardised workflows for specialised electrophysiological data structures lets scientists actively participate in the development of procedures with the aim to reduce workload, enable more transparency and lower the burden of data sharing, crucial for improving reproducibility.