Abstract
A key aspect of neuroscience research is the development of powerful, general-purpose data analyses that process large datasets. Unfortunately, modern data analyses have a hidden dependence upon complex computing infrastructure (e.g., software and hardware), which acts as an unaddressed deterrent to analysis users. Although existing analyses are increasingly shared as open-source software, the infrastructure and knowledge needed to deploy these analyses efficiently still pose significant barriers to use. In this work, we develop Neuroscience Cloud Analysis As a Service (NeuroCAAS): a fully automated open-source analysis platform offering automatic infrastructure reproducibility for any data analysis. We show how NeuroCAAS supports the design of simpler, more powerful data analyses and that many popular data analysis tools offered through NeuroCAAS outperform counterparts on typical infrastructure. Pairing rigorous infrastructure management with cloud resources, NeuroCAAS dramatically accelerates the dissemination and use of new data analyses for neuroscientific discovery.